Marketing leaders at Snap, Overwolf, Eos, and Doist are trading the CMO title for CRO, President, or Chief Growth Officer as the job outgrows the old name.

Overwolf CMO Shahar Sorek splits this into three tiers:

1. The channel CMO who runs campaigns

2. The integrator CMO who owns the full customer path

3. The structural CMO who answers bigger questions, like how the company creates value in the first place

His own role now includes investor relations, which he calls marketing aimed at a different audience.

At Doist, Brenna Loury moved from CMO to CRO, merging marketing and customer experience to shift the company's revenue from mostly consumer to mostly B2B.

This isn't a rebrand for its own sake.

Snap's Grace Kao points to the CMO-CFO relationship as the real catalyst.

Eos CMO Soyoung Kang calls her CFO her most valuable internal relationship, and Chief's Sabrina Caluori says the CFO should rank right behind the customer.

The shift carries real cost, too.

Loury says she struggles to relate to peer CROs who came up through enterprise sales, and Kang had to pull back from marketing work she loved once she took a bigger seat.

Pinterest's AI shopping pivot just posted real numbers 🛒

Pinterest reported its second quarter revenue of $1.18 billion, up 18% year over year, its fourth straight quarter above $1 billion.

About 30% of lower-funnel ad revenue is from Pinterest's automated ad tool, and bidding within it lifted ROAS 28% in testing this quarter.

U.S. and Canada revenue growth accelerated to 18%, a five-point jump from last quarter, driven by AI bidding pulling in more retailer spend.

If you buy Pinterest ads and haven't tested Performance+ bidding, this quarter's numbers make the case for turning it on. See the breakdown →

Teads sues Google over 6.88 trillion lost ad impressions ⚖️

Teads filed an 85-page antitrust lawsuit against Google on August 3, becoming the fifth ad exchange to sue the company in the past year.

The complaint claims Google's decision to tie its Ads buying platform to its AdX exchange cost rival exchanges 6.88 trillion ad impressions between 2017 and 2023.

Google rejects the claims, saying advertisers and publishers choose its tools because they work, cost less, and are easy to use.

If your team routes spend through Google's ad stack, this fight over auction rules is worth tracking before it reshapes pricing. Read more →

Klaviyo grew 26%, Wall Street punished it anyway 📊

Klaviyo's second quarter revenue hit $370.6 million, up 26% year over year and ahead of the $362.1 million Wall Street expected.

The company raised its full-year guidance to $1.526 billion to $1.534 billion. Investors sold the stock down 10% in after-hours trading anyway.

Net revenue retention held at 109%, and buyers spending over $50,000 a year grew 36% to 4,477, so the growth itself wasn't the problem.

If you run lifecycle or CRM, this is the bet Klaviyo is making: fewer point tools, one platform doing more work. Read the breakdown →

📌 An AI system that emails you content ideas daily (6-minute read)

Buffer's Aiza Domingcil built a system where ChatGPT emails her three buckets of content ideas every morning: Substack topics, LinkedIn posts, and email experiments.

She set it up with ChatGPT's Schedule feature and one saved prompt.

The system helped her launch two newsletters, one of which was named a top B2B marketing newsletter in Southeast Asia.

📌 Reverse-engineer your company's data into lifecycle moments (prompt)

This prompt turns Claude or ChatGPT into a lifecycle strategist that maps every proprietary data signal your company could use to personalize customer relationships.

It then builds two structured tables: one for smarter lifecycle decisions, one for experiences competitors can't copy, and forces specificity.

Its Costco example pairs fuel-purchase telemetry with membership data to catch members who are gassing up but skipping the warehouse.

Worth running against your own company before your next lifecycle planning meeting.

📌 A step-by-step guide to making infographics with AI (9-minute read)

Ruben Hassid breaks down two ways to make infographics with AI.

Claude turns a prompt and reference images into precise, editable HTML infographics, while ChatGPT's diffusion model produces more visual ones.

He also shows how to save your style as a reusable Claude skill, so future infographic prompts take one line. Worth trying before your next LinkedIn carousel.

💡 Prompt of the day

Today's lead story is about CMOs whose job outgrew their title before their business card caught up. Most marketers never check whether their title matches what they own.

This prompt forces that comparison before your next org chart conversation.

THE MARKETING MANDATE AUDIT

You are a skeptical org-design consultant auditing whether a marketing leader's title matches their real scope of ownership, not a career coach telling them what they want to hear and not an HR generalist reciting generic title conventions. Your only job is to compare what this person actually owns against what their title implies, using only what I tell you. Do not recommend a title change without evidence, and do not assume ownership of something I haven't described.

Ask me these questions one at a time, and wait for each answer before asking the next:
1. What is your current title, and what does your organization officially say you're responsible for?
2. List everything you actually decide or drive day-to-day: budget lines, hires, campaigns, revenue targets, board narrative, investor relations, product input, anything. Be specific, not just "marketing."
3. Who do you report to, and what's the one metric that person cares about most when they evaluate you?
4. Name one decision in the last quarter that you made, but that technically wasn't "yours" to make, and one decision that was technically yours but someone else made instead.

Once I've answered all four, produce a report with these exact sections:
- Title-Scope Gap: Compare my answer to question 1 against question 2, quoting my own words as evidence, and state plainly whether my title undersells, oversells, or matches my real scope.
- The Real Job: Based on questions 2 and 3, name the actual function I'm performing right now (for example, integrator, structural leader, or channel operator), not the title on my badge.
- Accountability Check: Using question 3, state whether my success metric matches what I actually control, and flag it if I'm being measured on something I don't own.
- Boundary Friction: Based on question 4, name the specific tension between my formal authority and my actual decisions, and what that friction is likely costing me: credibility, speed, or leverage.
- What This Audit Can't Tell You: State plainly what's missing, such as whether a title change would actually get approved, or whether the org even wants this function formalized.

If my answer to question 2 is vague or mostly restates my job description, don't produce the rest of the audit. Ask me to get specific first, since a vague scope makes the whole comparison meaningless. Do not invent responsibilities I haven't described.

Save the output as a Markdown file I can bring into my next 1:1 with my manager or into a comp and title review.

🔗 How Amazon built the fastest-growing ad business (14-minute read)

Mike traces how Amazon built the fastest-growing ad business in internet history, from reluctant 2006 banner ads to $56 billion in revenue last year.

The turning point came through unglamorous choices.

They built their own ad platform in 2012, beat Google to acquire Twitch in 2014, and landed NFL Thursday Night Football rights in 2021. Worth reading if you want the real mechanics behind how the "Duopoly" of Google and Meta became a "Triopoly."

🔗 Why your product release emails get ignored (4-minute read)

Richard lays out why most product release emails get a 4% open rate. His fix is a six-step checklist, from writing a subject line to giving each email exactly one call to action.

He also pushes segmentation: power users, new users, and enterprise admins each need a different angle on the same feature, not the same blast.

Worth reading before your next launch email goes out.

🔗 13 of the best SEO blogs, beyond the usual suspects (13-minute read)

Ahrefs' Louise Linehan rounds up 13 SEO and AI-search blogs worth following beyond the usual names, each with a specific example of what earns the bookmark.

Kevin Indig's Growth Memo found the word "best" triggers brand mentions in nearly 70% of AI prompts, and Chris Long found ChatGPT runs a live search one in three times to ground its answers.

The list favors writers who publish original tests and data over recycled advice, including one blog Ahrefs liked enough to acquire outright. Worth a scroll if your RSS feed still points at the big-name SEO sites everyone already reads.

Today's stories all came down to the same question: does the label still match the job? What's one thing you are fully responsible for at work that has nothing to do with your actual job title?

Your Paid Media Has Blind Spots. We'll Find Them for Free.

Your ad platforms tell you what happened. They rarely tell you what you are missing. HK Digital's senior paid media experts will review your Google, Meta, and other active campaigns at no cost. We assess performance, account structure, creative, targeting, media mix, and reporting to uncover wasted spend and overlooked growth opportunities.

You will receive a clear, prioritized set of recommendations you can put to work, whether or not you hire us.

Get My Free Audit

Partner with us

Snap reported second-quarter earnings. Revenue hit $1.6 billion, up 19% from a year ago, and ad revenue grew 9% to $1.28 billion.

The company's AI-powered Smart Campaign tools drove a 56% jump in platform conversions. Cost per app install fell 8%, and cost per purchase fell 18%.

App purchase volume jumped 128%. Snap's stock jumped double digits.

Ad platforms have promised AI-driven efficiency for years, mostly through demos and roadmap slides. And now, the actual performance numbers are showing up for advertisers who use the tools.

Snap's NA daily users, its highest-value market, still slipped 7% even as global users grew 5%. It means the ad gains came from the AI layer.

If Smart Campaign-style tools keep compounding, budget conversations shift from reach to what the AI-run campaign actually converted.

Every platform selling AI ad tools now has a real number to beat.

Pie just made a TikTok creator its CMO 📱

Social platform Pie named creator Nadya Okamoto as co-founder and CMO. Okamoto built her following on TikTok with over 4 million followers.

Pie founder Andy Dunn argues most marketing chiefs are older and learn about Gen Z secondhand, through research decks rather than daily life.

He wanted someone who already lives inside that audience. Okamoto also holds an advisory "chief creator officer" title at a startup, Cherub.

If your brand still treats creator deals as one-off posts, this signals some are becoming full executive hires. Get the complete details →

Publicis's $2.2B data bet raises an awkward question 💰

Publicis paid $2.2 billion for LiveRamp, adding a data-identity platform to its stack. LiveRamp's client list includes several of Publicis's biggest rivals.

But the deal raises a question: can LiveRamp stay neutral while its owner competes with its clients? Other holdcos are chasing different bets.

The era of holdcos buying more agencies is fading. If your agency of record just got acquired, ask what business its new owner is building toward. Get the analysis →

Japan Airlines built a tool to track its AI visibility 🔍

Japan Airlines' website traffic fell about 10% year over year, as AI tools increasingly answer travel questions before people click through.

They partnered with Jellyfish, an agency, to build a tool that tracks how AI models mention and compare its brand. They used those insights to rework PMax campaigns.

Conversion rate rose 56%, cost per acquisition fell 21%, and media spend dropped 12.5%. Japan Airlines performs well on Gemini but barely shows up in ChatGPT.

So it's launching paid ChatGPT ads this month. If you have never checked how AI models describe your brand, this is a playbook worth copying. Read more→

📌 Gamma turns a prompt into a finished deck (tool)

Gamma is an AI design tool that turns a prompt, outline, or file into a full presentation, website, or document. It exports to PowerPoint, PDF, or Google Slides.

Over 100 million people use it. It works best for slide decks, one-pagers, and social graphics you don't want to build from an empty page. Worth trying if your team still burns hours formatting decks instead of writing them.

📌 10 AI-powered link-building tactics, one thread (7-minute read)

Growth marketer Borja breaks down 10 AI-powered link-building tactics, from broken-link outreach to building a statistics page journalists cite.

Each tactic pairs a classic PR or SEO play with an AI shortcut. The tactics themselves, especially the statistics-page and unlinked-mention ideas, are solid and reusable.

📌 A prompt framework built for GTM teams (10-minute read)

Tofu's guide breaks prompting into one formula: Signal Context + AI Role + Job + Output Format + Tone. It includes six full example prompts, covering everything from competitor takeout pages to win-loss pattern analysis.

It's built for B2B and ABM teams, but the framework fits any prompt that needs business context. Worth saving for the next time a prompt keeps missing the mark.

💡 Prompt of the day

Today's lead story is about Snap's AI ad tools posting efficiency numbers.

Most marketers get an AI ad-tool claim from a vendor and never test it against their own account. This prompt forces that test before you shift real budget.

THE AI AD TOOL CLAIMS AUDIT

You are a skeptical performance marketer auditing whether an AI ad tool's claim holds up on my own account, not a vendor evangelist repeating the pitch deck and not someone who assumes every AI claim is inflated. Your only job is to size up whether this specific claim is trustworthy enough to shift budget toward, using only what I tell you. Do not validate the claim without evidence, and do not assume my results will match the vendor's case study.
Ask me these questions one at a time, and wait for each answer before asking the next:
1. Which AI ad tool or feature are you evaluating, and what exact claim is it making? Quote the number if you have it.
2. What does your account's performance look like right now, before turning this on: cost per result, conversion rate, and spend level?
3. Have you already tested it on a live campaign? If yes, what were the before and after numbers? If no, what has stopped you so far?
4. What result would you need to see to trust this claim enough to shift real budget toward it?

Once I've answered all four, produce a report with these exact sections:
- Claim Breakdown: Restate the claim in plain terms, quoting my answer to question 1 as the exact wording under review.
- Baseline Gap: Compare the claim against my answer to question 2, and name what's missing to make a fair before-and-after comparison.
- Evidence Status: Mark this Tested or Not Tested based on question 3. If tested, state the real delta in my own numbers, not the vendor's.
- Trust Threshold: Propose the smallest test that would satisfy my answer to question 4, with one clear pass or fail metric.
- What This Audit Can't Tell You: State plainly what's missing, such as whether results hold once spend scales up.

If my answer to question 3 shows no real test has run yet, do not rate the claim as trustworthy. Build the audit around designing that test first instead. Do not invent performance numbers I haven't given you.
Save the output as a Markdown file for a platform-mix review.

🔗 7 ecommerce blogs AI actually cites (9-minute read)

Backlinko studied seven ecommerce blogs, from Garmin to a small pet brand, to find what makes AI cite a blog. He found that blogs that answer product questions get into AI Overviews, while generic posts get skipped.

Garmin's blog ranks for 46,000-plus keywords and gets cited by AI 52,000-plus times, mostly because its docs are specific. Worth reading before your next content calendar meeting, especially if your blog still chases traffic instead of citations.

🔗 A 30-minute trick moved 20 articles to page one (2-minute read)

SEO advisor Irina Maltseva tried something simple on older articles that weren't ranking: update the publish date and republish. No rewrites or new links, just a freshness signal.

Over 20 articles climbed from page 10-plus to page one within two weeks. She says the trick works on 10% to 40% of older posts, since Google and AI weigh freshness.

🔗 The best time to pitch a journalist is 8 AM (14-minute read)

BuzzStream studied 4.5 million PR pitch emails and found peak engagement between 8 and 9 AM local time. Monday is the best send day. UK journalists peak around 7 to 9 AM, and US journalists peak around 9 to 11 AM.

The bigger finding is that none of this matters much for a journalist, since habits vary by beat and publication. Worth reading before your next mass send, especially the tips on using a journalist's bio and past publish times.

Today's stories all came down to claims getting tested against real numbers.

What's one AI marketing tool your team adopted mostly on a vendor's claim, without testing it against your own numbers first?

Buying Flat Fee Media Has Never Been Easier

Somewhere out there is an audience that already trusts someone. A newsletter they open every morning. A site they read. A creator they actually listen to.

Getting your brand in front of that audience usually means weeks of email tag and a rate card that isn’t standardized.

MediaPact is a flat-fee marketplace for paid media. Newsletters, publishers, podcasts, and creators list real inventory at real prices. Filter by vertical, send an offer, e-sign the IO, and pay. All in one place.

See the price and the deliverable before you commit. Guaranteed placement, 90% faster. Free for buyers.

Browse live inventory

Partner with us

LinkedIn spent years pushing AI tools that write your headline, your About section, and even full posts, and now, it's reversing course. The company is cutting the reach of generic AI content and adding a button that lets members flag posts as "AI slop."

It's also killing "Enhance Your Post," the feature that let AI draft your writing, and replacing it with a lighter, proofreading-only tool. Buffer analyzed 52 million social posts and found LinkedIn's median engagement rate fell 56% year over year in 2025.

Chief Product Officer Hari Srinivasan called AI slop "a top priority for all of us."

This flips the last two years of advice upside down. By Q1 2026, roughly half of new online articles were AI-generated, up from 36% within a year of ChatGPT's launch.

Engagement fell due to that flood of generic content, and that threatens the ad and Premium subscription revenue LinkedIn depends on.

LinkedIn is now treating authentic human perspective as a ranking signal.

For two years, using AI was your competitive edge. Now everyone has it, so your edge is whatever AI can't fake: your judgment, failures, opinion.

A judge just let a privacy case against LinkedIn move forward 🔒

A federal judge ruled Thursday that a Nevada woman can move forward with her privacy lawsuit against LinkedIn. She says LinkedIn's Insight Tag collected medication and dosage details she entered on Nevada's state health marketplace, Health Link.

LinkedIn said the health questions were marked "optional," so she had no real privacy claim, but the judge disagreed. The case traces back to reports that state health exchanges sent visitor data to LinkedIn's tracking pixel.

If you run tracking pixels near health or financial forms, this case shows exactly how that risk plays out in court. Read the ruling breakdown →

Heineken turns Serena Williams' old ticket stubs into new fans 🎾

Heineken 0.0 is turning old ticket stubs into a shot at the 2026 US Open.

Fans submit a photo of an old stub and a memory for a shot at the Heineken Suite at Arthur Ashe Stadium. The push is part of Heineken's "Fans Have More Friends" platform, built on research showing fans have richer social lives.

Williams, a six-time US Open champion and Heineken 0.0's ambassador, calls it a way to reconnect with old fans. This is Heineken's second big fan stunt this year.

If your brand has old customer moments archived somewhere, this shows how to turn one into a reason to come back. See the Tickets campaign →

Deliveroo opens its most-viewed screen to programmatic ad buyers 🛵

Deliveroo Media opened its highest-traffic ad screen to programmatic buyers for the first time. The Order Tracker page, checked up to six times per order, now routes ads through Google Ad Manager to three major buying platforms.

The rollout covers the UK, France, Italy, and the UAE, with a 20% discount for agencies booking through September. Two features are coming this year: audience targeting on the placement, and a tool that matches your ad to the food someone ordered.

If you buy retail media, this new post-checkout spot is worth testing before competitors crowd it. Get the full rollout details →

📌 How an AI startup made LinkedIn its top lead source (9-minute read)

Clarify, an AI CRM startup, turned LinkedIn into its top lead source in six months. Their marketing lead broke down the exact playbook: treat LinkedIn like its own channel.

The piece breaks the strategy into six moves, from founder interviews to tracking which posts to repeat. It's a clear guide for any B2B team that wants LinkedIn to pull its weight.

📌 A step-by-step way to see if AI bots are reading your site (tutorial)

Screaming Frog published a tutorial on tracking AI bots such as GPTBot, ClaudeBot, and PerplexityBot. It shows exactly which pages these bots hit, ignore, or fail to load.

This is for anyone doing SEO or AEO who wants proof about AI visibility. You will see if your content ever reaches the bots that decide what gets cited.

📌 Why your old followers don't see your posts anymore (9-minute read)

Rachel Karten breaks down why big-follower brand accounts now get barely any views. The follower-based feed is gone, replaced by one that shows your post to everyone or no one.

She lays out three ways brands are adapting: flooding feeds, building serialized content, or riding campaign waves. Any SMM planning next quarter's content should read this.

💡 Prompt of the day

LinkedIn just said it wants real voices. This prompt reverse-engineers your writing voice from real posts, so every future draft, human-written or AI-assisted, sounds like you.

LinkedIn Voice Spec Builder

You are a professional ghostwriter and linguistic analyst. Do not act as a generic copywriter or offer generic "LinkedIn tips." Your only job is to reverse-engineer a specific person's writing voice from real samples of their work, then produce a structured, reusable voice spec.

Before you write anything, ask me these questions one at a time. Wait for my answer before asking the next one.
1. Paste 15-50 of my past LinkedIn posts (or links/screenshots). The more, the better. If I give you fewer than 15, tell me and ask if I want to proceed with a lighter analysis instead.
2. Who is my target reader on LinkedIn, and what do I want them to feel after reading a typical post?
3. Name one writer, creator, or public figure whose tone I admire, not to copy, just as a reference point.
4. Are there any words, phrases, or formatting habits I specifically want to avoid (em dashes, hashtags, emojis, etc.)?

Once I've answered, analyze the samples and produce a Voice Spec as a Markdown file with these exact sections:
- Sentence Patterns: average sentence length, how often I use fragments, my typical opener words, how I end posts. Quote three real sentences as evidence for each pattern.
- Vocabulary Fingerprint: 10-15 words or phrases I use unusually often, with one quoted example of each. Flag any words most LinkedIn writers overuse that I never touch.
- Structural Habits: how I open posts, how I use line breaks and white space, whether I use lists, how I close.
- Tone Boundaries: a 2x2 map of what's "on-voice" vs. "off-voice" for me: how casual vs. formal, how vulnerable vs. guarded, with one real example on each end.
- Do-Not-Do List: five specific things this voice would never say or do, based on the gap between my real samples and generic AI-written LinkedIn content.
- Test Draft: write one new 150-word post in this voice, on a topic I haven't covered yet, and mark in brackets exactly which voice-spec rule justifies each unusual choice.

If I give you fewer than 10 posts, or posts that contradict each other in tone, stop and tell me specifically what's inconsistent before producing the spec. Do not guess or average it away.
Save the final Voice Spec as a Markdown file I can drop into a Claude Project or hand to a ghostwriter, so every future draft starts from my real voice, not a generic one.

🔗 This messaging app runs India's small businesses (15-minute read)

Dharmesh Ba maps out how WhatsApp became invisible infrastructure for India's small shop owners. He tracks the real cost: ₹0.78 per marketing message, and delivery rates that fell from 95% to under 75% overnight.

He compares it to WeChat Work in China, which separates business and personal chats, and pitches an open protocol as the fix. It's a sharp read for anyone thinking about platform dependency in emerging markets.

🔗 AI-made ads are beating human-made ones in testing (5-minute read)

Marketing Week's Tom tested 18 AI-generated ads through System1, finding they scored on average 3.4 stars vs. 2.3 for typical ads. UK-made AI ads scored 4.6.

He pushes back on the "AI slop" story and argues the story is about effectiveness. Worth reading before you write off AI creative as a shortcut that hurts results.

🔗 Instagram is the app most Americans want to delete (8-minute read)

GetResponse pulled Semrush search data and found nearly 2 million U.S. users search each month to delete or deactivate a social account. Instagram tops that list, and TikTok users are five times more likely to reactivate than Instagram users once they leave.

The full breakdown covers ten platforms, including which ones people regret leaving and which ones almost nobody tries to quit. Useful context for any brand betting its full content budget on one platform's permanence.

One of today's biggest platforms changed its mind about how much AI should write for you. That's a decision worth making on purpose. Do you write your posts first and let AI edit, or start with AI and edit it yourself?

Saylor, a social-first agency founded in 2021 by a former Netflix and Disney executive, launched Fleet. It bundles creator casting, content development, production, and partnership execution into one unit.

They hired Nneka, previously of Amazon Advertising, Hello Sunshine, Mass Appeal, React Media, and Authentic Entertainment, to lead it.

Fleet starts with five staffers pulled from Saylor's existing creator work.

Founder Will says it's less a new offering than a formal stake in the ground, since Saylor built creators into its writing room from day one.

It's the industry catching up to how brands already spend.

Streaming services now sign deals to bring creator content to their platforms, and digital publishers are adopting influencer formats to attract attention.

Agencies used to treat social strategy and creator casting as two separate jobs. Saylor is betting that the split is over and that creators belong in the writing room.

This is a signal the client-facing structure is about to change as well, if your agency still buys influencer posts as a separate line item.

An agency just showed why your AI agents might be burning budget 🤖

Draft Digital published the architecture behind its AI buying system.

Instead of buyers making 12 different AI-agent calls to ad platforms, the agency now runs everything through "Buyer Agent,” which sits between the strategist and platforms.

Independent testing this year found conventional media buyers still beat AI agents by 13% on cost, so the token-savings claim hasn't been proven on a full campaign yet.

Postmus says they move up into strategy roles instead of disappearing, but no agency has published numbers to back that up yet. See the breakdown →

Microsoft built the most complete stack in B2B marketing 🔒

Microsoft now touches every stage of B2B: LinkedIn for identity and discovery, Bing and Copilot for search, Dynamics 365 for the customer record, and Teams for the workplace.

If Microsoft becomes the landlord for identity, discovery, and the customer record, B2B marketers lose the room to negotiate that comes from spreading risk across platforms.

The tradeoff is the same walled-garden problem that shadowed Google and Meta since it now stretches across the complete buying process. Read the full analysis →

Brands are putting 70% of festive budgets into digital and OOH now 🎪

As India's Onam festival approaches, brands are breaking down their seasonal budgets very differently than they did five years ago.

One D2C jewelry brand now puts 70% into digital performance marketing, while a consumer electronics brand skipped TV and put half its budget into OOH instead.

Creator partnerships are the other constant across nearly every brand quoted, running alongside digital and OOH rather than replacing either.

If you still plan "digital" and "influencer" as separate items for a seasonal push, this is a sign to test folding them into one plan instead. Read more →

📌 Pew just mapped exactly who uses which platform (and why) (9-minute read)

Pew surveyed over 5,000 U.S. adults and found YouTube (84%) and Facebook (71%) still lead, while Instagram, TikTok, WhatsApp, and Reddit continue to grow.

The splits are the real find: women lead on Instagram and TikTok, men lead on X and Reddit, and Reddit skews hard toward college graduates.

Worth pulling up before your next persona review, primarily if your targeting still treats "adults" as a segment rather than one platform at a time.

📌 The gap between marketers' beliefs and what customers feel (7-minute read)

Invoca surveyed 600 B2C marketing leaders and found 86% believe AI is improving the buying experience, while only 35% of their own customers agree.

90% trust AI-automated ad tools, but most use them on a quarter or less of total ad spend. The report also flags a lag teams haven't priced in: only 2% can turn a phone call insight into a live campaign change the same day.

Worth reading before your next AI budget review, especially the section on why confidence in your own rollout might be outrunning the evidence.

📌 4 pre-header tests most email marketers still skip (4-minute read)

Email marketer DJ Schwedelson breaks down four specific pre-header line tests, covering angles most teams never touch beyond the "here's what's inside" line.

Pre-header copy is one of the few free upgrades left in email, since it costs nothing to test and it shows up before a single word of body copy does.

If your pre-header text has read the same way for the past six months, this is a steal.

💡 Prompt of the day

Today's lead story is about an agency betting that creators belong in the room from day one, not bolted on after the concept is locked. Most brands still treat every creator partnership as a single transaction and then wonder why nothing compounds.

This prompt forces the question before your next renewal conversation.

THE CREATOR PARTNERSHIP RENEWAL AUDIT

You are a skeptical partnerships lead deciding whether a one-off creator collaboration deserves to become an ongoing retainer, not a talent agent pitching me on renewal and not someone who reflexively cuts every deal after one post. Your only job is to size up whether this specific creator relationship earned a second round, using only what I tell you. Do not recommend renewal without evidence, and do not assume good vibes on set mean the content performed.

Ask me these questions one at a time, and wait for each answer before asking the next:
1. Which creator and campaign are we evaluating, and what was the original brief or goal?
2. What were the actual results: reach, engagement rate, click-through, or conversions, compared to your other creator posts in the same period?
3. How did the creator handle the process: on-time delivery, usable content, willingness to revise, and any friction worth noting?
4. What would a second collaboration need to prove that this one didn't?

Once I've answered all four, produce a report with these exact sections:
- Performance Verdict: Compare the results from question 2 against your other creator work, and state plainly whether this creator over- or under-performed.
- Process Score (1-10): Rate the working relationship from question 3, quoting my own words as evidence.
- Compounding Potential: Based on my answers, name what a second collaboration could build on that a first one-off can't, or say plainly if nothing compounds here.
- Renewal Verdict: Renew, Test Once More, or Pass, tied directly to the evidence above, not to how much you liked working together.
- What This Audit Can't Tell You: State plainly what's missing, such as whether the creator's audience actually matches your buyer.

If my answer to question 2 shows no real performance data, don't recommend a renewal. Build the audit around getting that data first instead. Do not invent engagement numbers I haven't given you.

Save the output as a Markdown file for a partnerships review.

🔗 Why "consumers are influencers too" might be the lesson (8-minute read)

Rachel Karten interviews Mojo Supermarket's Mo Said about how his agency built Burger King's "There's A New King And It's You" campaign around customer stories.

Said's core argument is that marketing used to mean telling a story to an audience, and now it means building a story with them.

Worth reading before your next campaign brief, especially the part about a customer complaint: that the Whopper gets smushed, turning into a product fix.

🔗 What ML in email moves the needle (and what's hype) (12-minute read)

HubSpot breaks down which ML applications in email marketing have a track record.

It lays out rollout plans by team size, from a solo marketer's first four weeks to an enterprise governance framework. The useful part is the limits: ML fails on messy data, can't replace brand judgment on tone, and needs real volume to learn from.

Worth reading before your next martech pitch meeting, especially if "AI-powered" has become a reason to buy rather than a feature to test.

🔗 The case for un-gating your best content (7-minute read)

Storyarb interviews Vector's head of marketing Jess Cook on contact-based marketing.

Cook says the approach has doubled to quadrupled qualified pipeline for companies that tested it against a traditional gated setup. The tactic works by mapping content to buying-intent stages, then triggering behavior-based follow-ups.

Worth reading before your next content audit, mainly if your best-performing gated asset still gets fewer reads than it deserves because of the form in front of it.

Today's lead story is about an agency betting that creators belong in the room from day one, not bolted on after the concept is locked. Does your team cast creators before the creative concept is set, or after?

Upside's "Escaping the Loyalty Plateau" report found the average buyer now belongs to 17.4 loyalty programs, up 60% over the past decade.

Active participation grew just 8% over that same stretch, to 8.4 programs.

Grocery shoppers average 2.4 grocery programs, fuel shoppers 2.3, and restaurant customers 3.2. Only 59% of grocery members use their program regularly.

That drops to 49% for fuel and convenience, and 38% for restaurants. Joining takes a number or an email, and staying active takes something brands haven't figured out yet.

Signing up costs a brand almost nothing, so enrollment numbers look great in a board deck even when they mean very little. Upside's loyalty influence score measures how much a program actually changes spending and purchase frequency.

Restaurants rank highest at 45 out of 100, fuel and convenience at 42.8, and grocery at 39.2. The programs that do move behavior share three traits. They personalize offers, build benefits beyond points, and add perks that create real status.

Many still measure loyalty by signups, not by any of those three things.

Journeys turns back-to-school into a music video 🎬

Journeys launched "Don't Just Show Up. Turn Up," a back-to-school campaign.

The 30-second film shows students arriving at school by speaker-loaded bikes and horseback, trailed by personal DJs and five-necked guitars.

If your back-to-school creative looks like every other retailer's, this is a reminder that a strong visual travels further than another discount code.

The campaign runs across TV and social with six-second cutdowns built specifically for short-form platforms. See the film →

A fitness creator skipped brand deals and built her own platform 💪

Fitness creator Gabby George turned down brand deals to build her own subscription platform instead. GGStudio, which she launched in August 2025, now counts more than 30,000 members across 93 countries.

A single brand deal barely covers half a month's rent, George says, so she skips most of them and sticks to medically sound advice.

If your influencer budget is built entirely around one-off brand deals, this is a live example of a creator who bet the other way and won. Read the playbook →

Amazon's ad business hits $19.8B, barely gets a mention 💰

Amazon's ads made $19.8 billion in Q2, up 26% YOY from $15.7 billion.

That kind of growth would be headline news for any other ad platform, but it barely registered on Amazon's own earnings call.

Advertisers using Amazon's AI-powered Ads Agent saw 8% lower CPI and 6% lower CPA than those who skipped it. If you haven't tested Ads Agent yet, the CPA numbers alone are worth a look before your next budget cycle. Get the breakdown →

📌 A/B tests show page speed's effect on conversion (12-minute read)

Ronny Kohavi, Jakub Linowski, and Lukas Vermeer ran eight large-scale A/B tests on page load time, each with a median of 2.4 million users.

Only two of the eight showed a statistically significant lift in the expected direction, once the tests were properly powered.

The summary breaks down exactly how sample size and minimum detectable effect distorted earlier claims about page speed's impact on conversion.

Worth reading before your next quarter of "let's just make the site faster" gets pitched as a growth lever without any real data to back it up.

📌 A marketer walked away from performance-only marketing (8-minute read)

Ward Rushton spent years inside Omnicom's precision-marketing arm before deciding that clean dashboards were hiding a real growth problem.

He points to Airbnb, which cut performance marketing and grew revenue 40% by investing in brand instead. His core argument is that hyper-optimized targeting shrinks your addressable market even as the metrics keep improving.

Worth reading if your own reporting has started to feel like it's optimizing a smaller and smaller circle of the same buyers.

💡 Prompt of the day

Today's lead story is about loyalty programs that grow enrollment but not engagement. Most retention audits stop at signup numbers and never check what happens after.

This prompt forces the harder look.

THE LOYALTY GHOST-MEMBER AUDIT

You are a skeptical loyalty and retention consultant auditing whether my program builds real engagement, not a vendor selling loyalty software and not a cheerleader assuming every enrolled member is a win. Your only job is to size up the real health of my program using only what I tell you. Do not treat enrollment growth as success, and do not assume any of my segments are engaged unless I show you activity data.

Ask me these questions one at a time, and wait for each answer before asking the next:
1. How many people are enrolled in your loyalty, rewards, or email program, and how many took an action (purchase, redemption, open, click) in the last 90 days?
2. What does someone get for joining, and does that reward change based on anything about them, or is it the same offer for everyone?
3. What is the single most common reason a member becomes inactive, if you know it? If you don't know, say so.
4. What would "beyond points" look like for your program: a service, an experience, a status tier? Have you built any of it yet?

Once I've answered all four, produce a report with these exact sections:
- Ghost-Member Rate: Calculate the gap between my answer to question 1's enrollment and activity numbers, and state plainly how it compares to industry benchmarks (roughly 50% active is average, above 65% is strong).
- Personalization Gap: Based on question 2, state whether my program is running blanket offers or real personalization, and name one specific way to close that gap using data I already have.
- The Unknown Problem: If my answer to question 3 shows I don't actually know why members go inactive, build the audit around finding that out first, not fixing anything yet.
- Beyond-Points Roadmap: Using my answer to question 4, propose one concrete beyond-points benefit I could pilot with my most active segment this quarter.
- What This Audit Can't Tell You: State plainly what's missing from my answers, such as whether inactive members are gone for good or just dormant.

If my answer to question 1 shows I don't track 90-day activity at all, don't produce the rest of the audit. Build the report around setting up that tracking first, since nothing else here works without it. Do not invent activity numbers I haven't given you.

Save the output as a Markdown file I can bring into a retention planning meeting.

🔗 Why every marketer needs a GitHub in 2026 (5-minute read)

A growth marketer, Louise, argues that marketers now need a public GitHub the way they once needed a portfolio site. She walks through how to set one up, pin your best projects, and use Claude to write clear documentation for non-technical visitors.

Her own LinkedIn post making that case pulled in 1,200 new followers in a single day. Worth reading if you have built any skill, prompt, or tool with AI.

🔗 What the first half of 2026 revealed about AI's real impact (14-minute read)

Growth advisor Kevin Indig pulls together six months of data on AI search, and the picture is messier than the hype suggests. ChatGPT's share of AI search traffic fell from 78% to 56%, while Gemini climbed from 15% to 30%.

Referral traffic to publishers fell a third YOY, and major outlets are now suing or blocking AI crawlers outright. Worth reading before you lock in next year's AI-search strategy.

Today's stories all came down to the same gap: people show up once, and brands assume that means they will stay. What's the real active rate on your biggest list, program, or community, and have you ever actually pulled that number?

This week's earnings calls turned marketing effectiveness into a boast.

Starbucks CEO praised CMO by name for how she's spending the chain's $190 million ad budget. P&G's CFO told analysts the company still has room to make its media spend work harder.

Kraft Heinz CEO said fewer, better media partners lifted the company's global ROAS 8%. Reckitt Benckiser and Canada Goose made similar points on their own calls.

Most marketers aren't there yet: a Gain Theory survey found only 49% feel confident defending their data to a CFO.

Starbucks and Kraft Heinz aren't flexing confidence for fun.

Starbucks closed stores and fired its South Korea CEO after a promotion blunder, while P&G issued a profit warning in January.

When companies under pressure start bragging about measurement instead of hiding from it, that tells you the data got good enough to trust.

Holding companies are chasing the same: Publicis pitches sports spend as "measurable at scale," and Omnicom frames its AI spend around better outcomes.

Agencies are quietly redirecting pitches from CMOs to CFOs.

Apple's ad business just hit a record $30.7B 💰

Apple's services division, which includes its ad business, pulled in a record $30.7 billion in the June quarter, up 12% year over year.

That capped a quarter where iPhone revenue jumped 22%, and total company revenue hit $109.4 billion, beating Wall Street's forecast.

Apple CEO Tim Cook is set to step down on September 1. If you buy Apple Search Ads or App Store placements, a leadership change at the top is worth watching. Get the breakdown →

Time now sells ads to AI bots, not just humans 🤖

Time magazine has started serving ads to AI bots, not just human readers, in an attempt to monetize its growing bot traffic.

The ads are formatted as FAQ-style brand content and labeled as sponsored, built to be picked up inside an AI agent's response. Mobian CEO Jonah Goodhart says influencing an AI agent may matter more than influencing one human.

Worth testing on a small budget before every publisher starts charging for the same audience your brand is already chasing for free. Read the roundup →

A site asked users to trust 1,741 ad partners at once 🔒

Privacy group noyb filed a GDPR complaint against dictionary site dict.cc after a single "accept all" click asked a visitor to consent to data sharing with 1,741 named partner companies at once.

noyb calculates that actually reading every partner's privacy policy would take more than 172 hours, about seven full working days.

Worth auditing your consent setup this week, especially if you have never counted how many partners are hiding in that submenu. See the writeup →

📌 How a founder built $1M in pipeline from LinkedIn alone (3-minute read)

Jonathan breaks down the LinkedIn playbook that drove over $1 million in pipeline for his bootstrapped startup, on a $300-a-month ad budget.

The playbook also covers partnering with others to post about you, and spending a few minutes daily commenting inside your ICP's feed.

Worth stealing if your company page gets more silence than engagement, and nobody's posting under their own name yet.

📌 A framework for naming your company (or product) (18-minute read)

First Round's Head of Brand lays out the process she's used to name dozens of startups, from a naming brief to a namestorm to a final shortlist test.

She breaks down when AI helps, mostly brainstorming raw material, and when it doesn't, since its own name suggestions tend to be generic.

Worth bookmarking before your next launch or rebrand, especially the "two-layer name" test for judging your shortlist.

📌 5.7 million videos later: What actually works in short-form (8-minute read)

Metricool analyzed 5.7 million short-form videos from 582,000 accounts to map out where each platform actually stands right now.

• TikTok posts grew 156% year over year, while Instagram Reels views fell 59% overall even as small accounts kept gaining.

• Facebook became a reach machine for big accounts through Reels.

• YouTube Shorts posted strong volume growth but the weakest engagement of the four.

Worth pulling up before your next content calendar meeting if your team still treats every short-form platform the same way.

💡 Prompt of the day

Today's lead story is about CEOs bragging on earnings calls about how well they measure ad spend. Most marketing reports still rely on activity metrics that would fall apart in front of a real CFO.

This prompt runs your own reporting through that same pressure test before you are the one getting asked hard questions.

THE CFO-READY MARKETING REPORT AUDIT

You are a skeptical CFO reviewing a marketing report before approving next quarter's budget, not a marketing cheerleader looking for reasons to say yes and not an auditor trying to kill every line item on principle. Your only job is to test whether this report proves marketing drove real business results, using only what I give you. Do not praise a metric just because it went up, and do not assume a metric connects to revenue unless I show you the connection.

Ask me these questions one at a time, and wait for each answer before asking the next:
1. Paste the top 3-5 metrics from your last marketing report, word for word, including the actual numbers.
2. For each metric, what business outcome is it supposed to predict or drive: revenue, pipeline, retention, or something else?
3. What was your total spend for the period this report covers, and what result did that spend produce in dollar terms, if you know it?
4. What question did your CFO or leadership actually ask about this report last time, if any?

Once I've answered all four, produce a report with these exact sections:
- Metric-by-Metric Verdict: For each metric from question 1, quote it back and label it Proof, Signal, or Vanity, based on how directly it ties to the business outcome from question 2.
- The Missing Dollar: Based on question 3, state plainly whether this report connects spend to a dollar result, or just activity. Flag it if I can't answer this.
- What the CFO Actually Wants: Using my answer to question 4, or general CFO priorities if I have none, name the one number missing from my report that would change the conversation.
- Rewrite Direction: For each Vanity metric, give a specific direction for what to replace it with, grounded in data I actually have access to.
- What This Audit Can't Tell You: State plainly what's missing from my answers, such as whether my attribution model is trustworthy.

If my answer to question 3 shows I don't know the dollar result of my spend, don't rate any metric as Proof. Build the audit around closing that gap first instead. Do not invent revenue numbers I haven't given you.

Save the final audit as a Markdown file I can bring into my next budget meeting.

🔗 Why people pay $30 a month for "Free" email (6-minute read)

Superhuman charges $30 a month for something Gmail gives away free, and built a half-million-person waitlist doing it.

Tom Snyder breaks down how the company turned "commodity" email into a status symbol through scarcity, onboarding calls, and read receipts.

Worth reading before your next pricing conversation, especially if your product competes in a category everyone assumes should be free.

🔗 What Reddit moderators actually see about your account (11-minute read)

Foundation interviewed two Reddit marketing specialists on what moderators can see, including post history across subreddits and metadata to catch you switching VPNs.

The piece lays out exactly what gets a brand account banned versus welcomed, down to specific disclosure language that works.

Worth reading before your next Reddit push, since the piece argues the biggest brand mistake is disclosing affiliation at the bottom of a post.

🔗 How Crocs turned ugly shoes into a culture machine (14-minute read)

Sprout Social breaks down how Crocs grew from $1.2 billion to $4.1 billion in revenue partly through collaborations nobody expected.

The brand also runs region-specific social accounts rather than a global feed, posting in local languages and pairing with regionally relevant celebrities.

Worth reading if your brand strategy still treats "going global" as one account and one voice instead of a set of local ones.

Today's stories all came down to who can prove their spend worked, and who's just hoping nobody asks. If your CFO grabbed your last marketing report right now, would it survive five minutes of questions?

Lululemon got sued over "phantom discounts." A shopper says the brand listed leggings at $59, next to a "regular" price of $98.

Lululemon hadn't actually sold them at $98 since October 2025. Nike faces a nearly identical suit. Both cases cite the same California rule.

It says a reference price must reflect a real sale in the past 90 days, or the ad must say otherwise. Filings like these roughly doubled from 2024 to 2025.

This isn't a fringe risk anymore. Two of the biggest names in retail got named in the same month, and the Nike suit reaches back to every purchase since 2022.

Professor Anita Rao says the surge tracks with how price-sensitive shoppers have become, since people now scan every number on a page.

Her own research found something uncomfortable: a lower "starting from" price boosts engagement even when nothing about it is misleading.

That's the incentive pulling brands toward strikethrough pricing. Rao doesn't expect the lawsuits to change that math much, since converting sales may outweigh the legal risk.

Snapchat opens its ad platform to AI agents 🤖

On July 25, it completed its MCP server rollout.

Three days later, Snap added new agency tools to Business Manager. It also expanded its Indy Agency Partner Program into France and the UK.

Nine US agencies, including Whalar and Wpromote, lined up to pilot a Creator Agency Partner Program later this year. If your team already runs AI agents on Meta, check whether that workflow now reaches Snapchat as well. Get the breakdown →

Bad AI ads are quietly wrecking brand trust 📊

A new global study found most consumers can't spot AI-generated ads. But quality still matters: 42% say low-quality or "uncanny" AI ads hurt their opinion of a brand.

Polished AI ads work the opposite way. 40% of consumers see them positively. Marketers feel the same split, and 48% say they are wary of using AI to build ad creative at all, rising to 63% in North America.

Just 34% view ads in AI chat platforms positively, vs. 46% globally and closer to 68% across LATAM and APAC. See the findings →

B2B's "Last Mile" problem just got two fixes 🎯

Demandbase and DemandWorks announced an integration that links account intelligence to campaign activation. Once an account meets the qualification threshold, it automatically moves into coordinated outreach.

6sense is fixing a different piece of the same problem.

Its AI-Recommended Leads feature, shipping in August, turns a qualified account into a short list of named contacts sales can call the same day. If your ABM program still relies on manual exports, this is the moment to close that gap. Read the breakdown →

📌 80% of this B2B ad budget goes to thought leadership (4-minute read)

Ashley Lewin runs paid LinkedIn ads for B2B brands, and 80% of her budget goes to thought leadership. She breaks her approach into seven lessons.

She also tests Engagement against Reach instead of assuming the winning objective. Worth reading if your B2B paid social budget goes mostly to polished, traditional ads.

📌 Turn your server logs into a ChatGPT demand report (7-minute read)

Siteline's David Kaufman shows how to mine your server logs for ChatGPT's "ChatGPT-User" bot. That bot only fires when a real person asks ChatGPT something needing your site's live data.

He covers pulling hits, clustering pages, then comparing volume against content. Worth running this quarter if your AI-visibility reporting leans on guessed prompts.

📌 Train Claude on your own best LinkedIn posts (6-minute read)

Ruben Hassid lays out a full workflow for training Claude on your own best LinkedIn posts. Scrape your public posts for a couple of dollars, then upload the file to Claude.

Claude then builds a report on what patterns drove your best-performing posts. The skill creator then turns that into a reusable skill you can invoke for every new post.

💡 Prompt of the day

Today's lead story is about brands getting sued over prices that didn't hold up once someone checked them. Most marketing claims never get tested this hard before they go live. This prompt forces that test early.

THE PRICING CLAIM DEFENSIBILITY AUDIT

You are a skeptical compliance-minded marketing reviewer, not a lawyer giving legal advice and not a marketer looking for reasons the claim is fine. Your only job is to size up whether a specific price, discount, or marketing claim would survive outside scrutiny, using only what I tell you. Do not clear a claim without evidence, and do not assume a claim is safe just because it is common practice.

Ask me these questions one at a time, and wait for each answer before asking the next:
1. What is the exact claim, price, or discount you want to check, word for word as it appears?
2. What evidence do you have that the claim is true right now: a sale history, a data source, an internal record? Paste it if you have it.
3. How long has this claim, price, or offer been running, and has anything like it run before?
4. If someone outside your company checked this claim today, using public tools, what would they most likely find?

Once I've answered all four, produce a report with these exact sections:
- Claim Breakdown: Restate the claim in plain terms, quoting my answer to question 1 as the exact wording under review.
- Evidence Gap: Compare what the claim states against my answer to question 2. Name specifically what is missing, not just whether evidence exists.
- Outside Risk: Based on my answer to question 4, state plainly what an outside reviewer would likely find, and whether that matches the claim.
- Fix Direction: One specific change to the claim or the evidence behind it that would close the gap, not a full rewrite.
- What This Audit Can't Tell You: State plainly that this is not legal advice, and name what a lawyer or compliance reviewer should check that this audit cannot.

If my answer to question 2 has no real evidence, do not clear the claim. Build the audit around the missing evidence instead. Do not invent sales history or data I have not given you. Save the output as a Markdown file I can bring into a legal or compliance review.

🔗 Empathy beats reviews as a reason people buy (7-minute read)

A Stanford psychologist and Zurich's CCO surveyed 12,000 buyers across 11 countries. They found empathy beats reviews as a driver of purchase decisions.

The piece treats empathy as a system something you can build and measure. Worth reading before your next CX planning cycle, especially if "be more empathetic" has never been a real metric.

🔗 Inside the VC team built to win your launch day (6-minute read)

Andreessen Horowitz built an in-house media team that acts like an agency for its portfolio companies. The team ships a podcast and newsletter five times a week, betting daily reps beat weekly polish.

It's a look at brand-building infrastructure from the venture side. Worth reading if you are rethinking what "owned distribution" means for a growing brand.

Today's stories all came down to the same test: does the shortcut hold up once someone checks it? Which price, claim, or ad on your site right now could you not fully back up if someone asked?

Similarweb's report tracked a year of AI platform traffic worldwide. ChatGPT's share of that traffic fell from about 76% in June 2025 to roughly 53% by May 2026.

Gemini picked up most of the difference. Claude grew the fastest in percentage terms.

The overall AI category grew fast enough, up 70% year over year, that ChatGPT's actual visit count barely moved even as its share fell.

A single-platform AI search strategy now misses half of the AI search audience.

Gemini's growth rides distribution it owns through Search, Android, and Workspace. Claude's growth comes from developers and power users.

ChatGPT also changed how it displays brand links back in May.

Homepage clicks have held 62% of all referrals since then. Roughly a quarter of ChatGPT responses now carry an ad, so early, clean citations are worth it now.

X and advertisers officially end their GARM war ⚖️

X and the World Federation of Advertisers settled their two-year legal fight over the vanished GARM brand-safety initiative. The joint statement called it an effort to reset the relationship between the two organizations.

X sued WFA in 2024. They said members organized an illegal ad boycott after Elon Musk's takeover. A federal judge dismissed that claim in March.

Both sides say they are now aligned on brand-safety innovation. If you buy media on X, watch for what replaces GARM's role. Read the recap →

WPP Media bets on AI agents for India commerce 🤖

WPP Media partnered with Kily, an agentic AI startup. The deal plugs Kily's autonomous commerce agents into WPP Open, WPP's own agentic marketing platform.

Kily's agents handle listing, pricing, and ad decisions across marketplaces and quick-commerce apps. This is a global holding company handing real pricing calls to an agent.

If you run ecommerce media in a fragmented market, this is worth watching before agencies pitch it as standard. See the coverage →

This AI agent rules on whether a backlash is real 🕵️

Cyabra launched an AI agent that investigates suspicious online activity and returns a clear verdict. It checks account authenticity, timing, and content patterns.

Then, it labels a scan Confirmed Coordination, Likely Coordination, Insufficient Evidence, or No Coordination Detected. The whole process runs automatically on every scan and finishes in 30 seconds, with a downloadable evidentiary report attached.

Worth knowing about before your next real crisis. Read the announcement →

📌 Why growing through partners beats chasing another channel (9-minute read)

Emily Kramer, former head of marketing at Asana, argues that inbound, outbound, and lifecycle email are all getting noisier and less effective at once.

Her fix is growing through partners, creators, and communities who already have your audience's trust, rather than chasing that audience directly.

Worth reading if your team still treats influencer and partner relationships as a side channel instead of the main growth lever.

📌 Why chatbots will monetize with ads, not affiliate links (12-minute read)

Ad-tech analyst argues that personalized advertising will end up being how chatbots make money. He points to Amazon actively blocking AI shopping agents as proof the biggest retailer has zero incentive to let a chatbot broker the sale.

He also flags that product reviews, which chatbots lean on for recommendations, are themselves funded by affiliate links that a chatbot checkout would undercut.

Worth reading if you are budgeting for AI-driven commerce and assuming today's referral model survives another year unchanged.

📌 A tactical playbook for Pinterest ads most agencies skip (10-minute read)

Paid media expert Karen Nelson breaks down why Pinterest ads have less competition and a longer content lifespan compared to Meta or TikTok.

She also gives setup details: 2:3 vertical creative, 9-to-15-second video hooks, and starting budgets around $500 to $1,000 per month to get a real read.

Worth trying if your paid budget is entirely Meta and TikTok.

💡 Prompt of the day

Today's Breaking News has two brands bolting a new AI agent onto their stack in the same week. Most teams say yes to a new AI tool based on a demo.

This prompt helps you with a real fit check.

THE AI VENDOR FIT AUDIT

You are a skeptical martech buyer deciding whether a new AI tool or agent deserves a real budget line, not a vendor sales engineer running the demo and not someone who reflexively distrusts anything with "AI agent" in the name. Your only job is to size up whether this specific tool solves a problem I actually have, using only what I tell you. Do not recommend adoption without evidence, and do not assume a vendor's case study will repeat inside my team.

Ask me these questions one at a time, and wait for each answer before asking the next:
1. What is the tool or agent, and what specific task does it claim to automate or speed up?
2. Who on your team does this task today, how long does it take them, and how often does it happen?
3. What happens right now when that task is done wrong or missed: is the cost small and recoverable, or big and visible?
4. What would you need to see in a 30-day pilot to justify paying for this after the trial ends?

Once I've answered all four, produce a report with these exact sections:
- Problem-Fit Score (1-10): Rate how directly the tool's claimed function matches the real task from questions 1 and 2, quoting my own answer as evidence.
- Stakes Check: Based on question 3, state plainly whether this is a low-stakes efficiency tool or a high-stakes judgment tool, and flag if I'm about to hand a high-stakes call to something unproven.
- Pilot Design: Propose a specific 30-day test structure using my answer to question 4, with one clear pass/fail metric, not a vague "see how it goes."
- Renewal Verdict: Adopt, Pilot Further, or Pass, tied directly to the evidence above.
- What This Audit Can't Tell You: State plainly what's missing from my answers, such as real output quality once the tool runs on my own data.

If my answer to question 3 shows this task is high-stakes (legal, financial, reputational), do not recommend full adoption without a human review step in the loop. Do not invent vendor performance data I haven't given you. Save the output as a Markdown file for a vendor review meeting.

🔗 Why your least AI-literate customers adopt it fastest (4-minute read)

A study found that people with lower AI literacy adopt AI tools faster, because they experience AI as magical rather than scrutinizing its limits.

Consumers who understand AI better are more critical of it and slower to adopt, even though they rate it as more capable and ethical.

The researchers warn this creates a tension. Worth reading before your next AI-feature launch, especially if your messaging assumes technical users are your early adopters.

🔗 Why your PR budget is aimed at the wrong sources (7-minute read)

Kevin Indig's analysis of 1,000 domains found that Authority Score is the strongest predictor of whether a site's backlinks become AI citations.

He argues AI doesn't form a fresh opinion about your brand on every query. It reuses trust already attached to the sources it already associates with your topic.

Worth reading before your next earned-media push, especially if your PR plan was built for a different topic and just relabeled for this one.

🔗 Why one founder spent $20K on branding to survive AI (4-minute read)

Jonathan Martinez, founder of GrowthPair, explains why he spent nearly $20,000 on a full rebrand just fourteen months into his startup.

His logic is that AI has dropped the cost of building a competing product to near zero, so brand feel and voice are the only lever left to stand out.

He rewrote his own copy to sound conversational, and leans on his personal LinkedIn over his company account to keep the brand human.

Worth reading if your own product has gotten easy to copy.

Today's stories all came down to who, or what, gets to make the call without you in the room. Which decision in your stack have you already handed to an AI agent, and which one would you never hand over?

Ladder, a strength-training app, launched a new campaign, "Stronger, not smaller."

The campaign stars actor and singer Hilary Duff in a 60-second hero ad about the value of strength training. A second spot, styled as a fake corporate compliance video, runs about four minutes and busts myths about lifting weights.

Ladder built the campaign to push back on a resurgence of "thin is in" messaging tied to Gen Z's early-2000s nostalgia. Philip Edsel, Ladder's VP of brand and creative, said: "We're trying to get ahead of that misinformation."

This is a brand choosing a fight, and that's a bigger bet than picking a celebrity face.

Ladder is betting that naming the "thin is in" story builds more trust than another generic strength-training ad directly. The choice of Duff matters as well.

She reads as familiar to millennials and newly visible to Gen Z through SkinnyTok itself. That cross-generation appeal is rare, and it lets one campaign do double duty.

The corporate-compliance-video format also tells a shift: fitness brands are borrowing comedy and internet-native formats. Any brand selling something tied to body image now has to decide whether to name the pressure or stay quiet and hope it passes.

Claude's shared chats and tools turned up on Google 🔒

Shareable Claude apps and conversations are turning up in Google search results.

Users on Reddit and X found they could search for public links and pull up other people's documents and chats. Anthropic said links only become searchable once someone shares them somewhere a search engine can already see, like a forum.

ChatGPT and Meta AI hit similar exposure problems last year. If your team uses AI tools, check if "share" on your account means one person or the entire internet. Read more →

Microsoft Ads adds one-click creative previews to PMax 🎯

Microsoft Advertising added its Ad Preview Hub to PMax campaigns.

Advertisers can open an asset group, select Preview ads, and see how creatives look across devices and placements before a campaign goes live.

PMax has always made creative approval hard, since a campaign can produce dozens of asset combinations across Bing, MSN, and Outlook placements. If clients still ask for screenshots before sign-off, this cuts that back-and-forth to one link. See the rollout →

Vusion bets in-store screens are the next retail media platform 🛒

Vusion, a French company known for digital shelf labels, agreed to buy In-Store Media, a Barcelona-based in-store retail media company.

In-Store Media works with 90 retail banners and more than 1,600 brands across nine countries and made about €120 million in revenue last year.

Vusion's CEO said, "the next big digital media is the physical store."

If your brand already buys retail media online, watch whether your retail partners start pitching in-store screens the same way. Read the announcement →

📌 growthroles.com tracks live marketing jobs and pay (tool)

growthroles.com is a job board built only for marketing roles.

It shows pay ranges for most listings, the skills that appear most often in job descriptions, such as go-to-market and demand generation.

It's worth a look if you are weighing a move, since OpenAI, Stripe, and Snowflake post the most roles with real salary ranges attached. It also works as a market-read tool: watching which titles and skills spike shows where marketing budgets are growing.

📌 How do you measure the ROI of a timeless brand? (3-minute read)

This piece breaks brand ROI into four parts: trust, design, consistency, and resilience, instead of treating "brand building" as one vague bucket.

It cites WARC research linking a 4% rise in brand equity to about 1% more annual revenue growth. Each section comes with its own measurement checklist, so you can pick one and start tracking it this quarter.

Worth reading if your CMO gets asked to defend brand spend in a budget meeting and needs a framework beyond "it builds awareness."

📌 How the best products remove guilt (6-minute read)

VC Rex Woodbury breaks down why the gummy-vitamin brand Gruns hit a $300 million run-rate: it turned a guilty pleasure into a responsible habit.

He applies the same formula to AI products, arguing enterprise tools win by turning boring tasks into something people enjoy doing.

Worth reading if you are positioning any AI or wellness product, since it gives you one concrete question: does this flip guilt into permission, or pain into pleasure?

💡 Prompt of the day

Today's lead story is about a brand that decided to name an uncomfortable topic rather than dancing around it. Most brands never run that decision through a real test.

This prompt forces the audit before you write the next line of copy.

THE SENSITIVE TOPIC POSITIONING AUDIT

You are a skeptical brand strategist deciding whether I should name a sensitive tension directly in my marketing. You are not a hype writer trying to make every topic feel bold, and not a risk-averse compliance reviewer trying to talk me out of everything. Your only job is to size up whether naming this tension helps or hurts, using only what I tell you. Do not recommend a direct approach without evidence, and do not assume boldness always wins.

Ask me these questions one at a time, and wait for each answer before asking the next:
1. What is the sensitive topic or tension you're considering naming directly in your marketing?
2. Where have you already seen customers bring this topic up unprompted: reviews, comments, sales calls, support tickets? Paste a real example if you have one.
3. What is the worst realistic reaction if you name this directly, and what is the cost of staying vague instead?
4. Who on your team or in your category has already tried naming a similar tension, and what happened?

Once I've answered all four, produce a report with these exact sections:
- Signal Strength: Rate how strong the evidence is that customers already care about this topic, quoting my answer to question 2 as the main evidence.
- Risk vs. Silence Cost: Compare the downside from question 3 against the cost of staying vague. State plainly which risk is actually bigger.
- Precedent Check: Interpret my answer to question 4. If no one has tried this in my category yet, say so and flag it as higher risk, not lower.
- Test-First Recommendation: Propose one small, testable piece of content that names the tension, not a full campaign, with one metric that decides if it's worth scaling.
- What This Audit Can't Tell You: State plainly what's missing from my answers, such as real audience research beyond anecdotes.

If my answer to question 2 has no real examples, don't recommend naming the topic yet. Build the audit around finding that evidence first instead. Do not invent customer sentiment I haven't given you.

Save the output as a Markdown file I can bring into a creative review.

🔗 Why advertisers can't overlook veteran marketing (4-minute read)

Nielsen looked at the 16 million veterans in the US and found a specific buying power advertisers ignore. Veteran households average $115,139 in income, above the national median, and veterans are 19% more likely to subscribe to a live TV streaming service.

They also spent $6.8 billion online and more than $10.7 billion on home improvement in the past year, and they're 23% more likely to own a dog.

Worth reading before your next audience-segmentation exercise, especially if your persona work has never named this group at all.

🔗 Middle-aged viewers now power YouTube's long-form growth (2-minute read)

New Ampere Analysis data proves the biggest consumers of full-length movies and TV shows on YouTube are viewers aged 35 to 64, not teens.

Nearly one in five users in that age group watch full movies or shows on the platform, and the share is highest in India at 32%. Ampere calls this group "content super-consumers" because they also watch more genres than the average YouTube user.

Worth reading if your media plan treats YouTube as a young audience platform only.

🔗 Marketers must wrestle with a vanishing middle class in 2026 (4-minute read)

WARC's annual outlook found 73% of marketers now think "middle class" is a meaningless label, as income growth slows and costs keep rising.

Only 54% of the industry believes 2026 will be better than this year.

WARC's advice is to help people through what it calls "affordability tension," the gap between what people want and what they can spend.

Worth reading before your next planning cycle, especially if your audience segments still assume a stable middle-income buyer sits in the middle.

Today's stories all came down to a choice between naming something uncomfortable and staying quiet about it. Is there a topic your customers keep bringing up that your own marketing has been dancing around?

Josh Grant Webinar - 30-day AI search pipeline recovery sprint

Why your top pages still miss AI search

Pricing, feature, and comparison pages are what sales leans on, and what AI search summarizes wrong or skips entirely.

Josh Grant, founder of StackedGTM and former VP of Growth at Webflow, joins AirOps on turning AEO into pipeline this quarter. Not a future-proofing project. A short-horizon conversion play.

He covers why comparison pages capture mid-funnel AI demand fastest, what makes pricing and feature pages extractable and citable, why third-party signal from G2, Capterra, and Reddit outweighs on-site credibility, and what a measurement stack looks like once last-touch attribution stops capturing AI's real influence.

Live August 5, 2 pm ET on Zoom. Recording goes to everyone who registers.

REGISTER NOW

Partner with us

OpenAI sent ChatGPT Ads Manager beta users a product email packed with eight new features. The headline addition is oCPC, or conversion-optimized cost-per-click bidding.

It nudges delivery toward clicks likely to convert while still billing per click.

Daily budgets have changed as well, now based on a seven-day average instead of a fixed daily cap, matching how Google and Meta already work.

Advertisers also got geo-exclusion targeting, a Bulk API for bulk campaign edits, and mobile measurement links to AppsFlyer and Adjust.

This turns an ad platform into something a media buyer can run a budget through. Google and Meta have had outcome-based bidding and bulk campaign tools for years.

ChatGPT Ads caught up now. Three holding companies, WPP, Omnicom, and Dentsu, already joined the ad pilot, and better tooling makes their spend easier to defend.

Trust is the bigger shift. Automatic advanced matching and mobile attribution mean advertisers can measure what a ChatGPT ad drove. Once you can measure it, you can defend the spend in a budget meeting, and that's what turns a pilot into a line item.

Nielsen rebuilds Ad Intel around AI and live agents 📊

Nielsen renamed its ad-spend tool "Ad Intel AI," rebuilding it around AI.

Ad Intel AI now connects through the MCP, so a client's own AI agents can query Nielsen's data directly inside their workflow. CPO Akhil Parekh called it the first phase of "an end-to-end ad ecosystem engineered for the AI era."

If your team still exports Nielsen reports into a slide deck once a quarter, this is the moment to ask what a live feed changes. Know more →

Born Social builds a creator army to fight AI slop 🎨

Croud-owned agency Born Social launched the Born Creator Community, a global network built to help brands scale real human content.

• Stanley pulled 869,000 organic impressions and 23,000 new followers across five European markets in three months.

• Premier Inn saw over 500,000 organic reach and a 2.5% engagement rate from the same setup.

Chief strategy officer Callum McCahon said the bet is that brands that win will "scale human creativity rather than replace it." If your feed strategy relies on AI posts, this is a live test of whether real creators still outperform them at scale. Read the details →

Brands now pay creators with just 500 followers 💰

The WSJ reports that a new wave of brand advocacy programs now only needs 500 followers to start paying creators for branded content.

Momfluencer told her four-figure followers on Instagram and TikTok were too small to interest big brands for most of the last five years, until now.

If your influencer program still sets a follower-count floor in the tens of thousands, this is a signal to test if that line is filtering out creators worth paying. Read the report →

📌 How Morning Brew turned a newsletter into a $75M company (7-minute read)

This piece breaks down the three pillars behind Morning Brew's growth: a referral program, a distinct conversational voice, and native ads.

At its peak, advertisers paid roughly $50,000 for a single primary ad slot.

It also covers how Morning Brew built a network of vertical newsletters and podcasts. Worth reading if your own newsletter growth has stalled.

📌 How a non-tech CEO built an AI workflow from sales calls (40-minute watch)

Suzy's CEO built a full AI workflow on Zapier that turns raw sales call transcripts into summaries, sentiment scores, and even SEO blog posts.

This piece also drafts follow-up emails and pulls keywords straight into Google Ads campaigns, all without an engineer touching the pipeline.

Worth watching if your team still treats "the CEO learns to prompt" as optional rather than the fastest way to find what's worth building.

📌 Every Meta ad concept ranked into a 2026 tier list (9-minute read)

A creative strategist ranks 20 Meta ad concepts, from AI UGC to offer statics, scored on scalability, production speed, and funnel fit.

AI UGC and Pixar-style animation top his S-tier list. He also maps every concept to a funnel stage, so TOF education formats don't get mistaken for BOF urgency plays.

Worth bookmarking before your next creative brief, especially if your account still leans on the same two or three ad formats. 

💡 Prompt of the day

ChatGPT Ads just got real bidding and measurement tools, and every emerging ad channel makes the same pitch: getting in early. Most teams either chase every new platform or ignore all, and neither is a strategy.

THE NEW AD CHANNEL TEST BUDGET AUDIT

You are a skeptical media buyer deciding whether a new ad channel deserves a real test budget, not a platform rep pitching you on early-mover advantage and not someone trying to talk you out of every new channel out of habit. Your only job is to size up whether this specific channel fits my account right now, using only what I tell you. Do not recommend a test budget without evidence, and do not assume a new channel will perform like an established one.

Ask me these questions one at a time, and wait for each answer before asking the next:
1. Which new ad channel are we evaluating, and what specific new features or capabilities made you consider it now?
2. Where does your current media budget concentrate, and which channel would this new one most likely pull spend from?
3. What does your ideal customer profile look like, and is there any evidence this channel's audience overlaps with it?
4. What is the smallest test budget and timeframe you could run without it distorting your other channel results?

Once I've answered all four, produce a report with these exact sections:
- Readiness Score (1-10): Rate how ready this channel is for real spend, quoting my answer to question 1 as the main evidence.
- Cannibalization Risk: Based on question 2, name what you would need to watch closely if this test underperforms and pulls budget back.
- Audience Fit Check: Interpret my answer to question 3. Flag plainly if I have no real evidence of audience overlap, rather than assuming it exists.
- The Test Plan: Propose a specific structure for the test budget and timeframe from question 4, with one clear metric that decides whether it graduates to a real line item.
- What This Audit Can't Tell You: State plainly what's missing from my answers, such as real performance benchmarks for a channel this new.

If my answer to question 3 shows no real evidence of audience overlap, don't recommend a test budget. Build the report around getting that evidence first instead. Do not invent channel performance data I haven't given you. Save the output as a Markdown file I can bring into a budget planning meeting.

🔗 Why younger shoppers trust brands more than older ones (5-minute read)

India Stronach digs into new research showing under-35s trust brands 12% more than older shoppers, but argues it hides a different story.

Older consumers name charities and pharmacists as their most trusted brands, while younger ones name supermarkets and online retailers. That split suggests two different definitions of trust: ethical trust and reliable-delivery trust.

Worth reading before your next brand-values campaign.

🔗 What 180 top-ranked Google Ads have in common (12-minute read)

WordStream analyzed 1,750 headlines and 550 descriptions from its highest Ad Rank campaigns and found "today" beats every other word for urgency.

Numbers, trust words like "certified," and the humble ampersand all showed up far more than low-price language. The bigger surprise: exclamation points and question marks barely appear, while formal title-case headlines dominate the top spots.

Worth a skim before your next ad copy sprint.

🔗 What actually triggers a Google AI Overview (8-minute read)

Ahrefs analyzed 146 million search results and found AI Overviews appear on just 21% of keywords overall, but that number swings by category.

Long, question-based searches are far more likely to trigger one than short navigational queries, and news searches rarely do. Worth reading before your next AEO push, especially if your content is in a category the data says AI Overviews mostly skip.

Today's stories all came down to the same question: who gets access, and at what threshold. What follower-count floor does your influencer program actually use, and have you ever tested going lower?

Joe Spector co-founded Hims & Hers before building Dutch, a 24/7 veterinary telehealth startup with more than 100,000 members.

Late last year, Dutch was paying an outside agency $50,000 a month for five pieces of creative, about $10,000 each. Spector called that expensive and limited.

So he fired the agency and built an in-house team that leans almost fully on AI tools, from static ads to full animated and live-action commercials.

The team now ships about 50 pieces of content a month, 10X the old output.

Dutch's customer acquisition cost fell 20% comparing January to April, the first four months after the new setup took hold.

Its Meta ad spend grew from about 5% of the budget to a quarter, managed through AI for A/B testing, brand-lift studies, and search strategy. Spector is trying to move Dutch upstream, building awareness so pet owners think of Dutch before an emergency hits.

That's why Dutch brought on Olympic figure skater Tara Lipinski as its first brand ambassador, a bet that a recognizable face builds trust an ad alone can't.

Reddit quietly became women's go-to for hormone health 📊

Reddit and Omnicom released "The New Health Ecosystem," a report on how women research hormone and reproductive health. It found 81% of women feel more comfortable discussing on Reddit than with friends or family, at 72%.

AI-driven searches on Reddit grew 15X over the past year. 7% said the process left them feeling more prepared for their next doctor's visit.

If your brand touches health, wellness, or anything people feel awkward asking about out loud, this is worth stealing. Read the report →

SaaS could learn trust from industries that never had it easy 🏢

At B2B Marketing UnBoxed 2026, a panel of marketers from insurance, logistics, fintech, and media argued SaaS has had it easy for years.

Free trials, PLG, and viral loops did the selling, while their industries had to earn trust and create demand with none of those shortcuts.

Panelists said that as SaaS products commoditize and PLG stalls, the "earn it the hard way" instincts from regulated, low-trust categories are becoming the edge.

Read it if your plan still assumes the product will do all the talking. See the recap →

📌 44% of social marketers feel their boss doesn't get social (9-minute read)

Link in Bio surveyed 834 social media experts and compared results YOY.

Instagram is the platform marketers would keep if forced to pick just one, but Reddit's share as a "priority platform" doubled since last year.

44% of respondents said their boss doesn't understand social media, and 56% are actively moving away from a platform, mostly X. Worth reading for the vibe check alone if you are building a case for more headcount or a bigger budget this planning cycle.

🛠️ Podium Phones: an AI receptionist for missed calls (tool)

Podium Phones is an AI-powered VoIP system built for local and service businesses that automatically texts back missed or after-hours calls.

It also transcribes and summarizes every call, so your team has context. Worth a look if your business still loses leads to an unanswered phone during peak hours.

🛠️ Ahrefs adds a free social scheduler to its SEO stack (tool)

Ahrefs expanded its free, beta Social Media Manager, letting you plan, schedule, and publish to LinkedIn, X, Instagram, and Facebook straight from the same dashboard.

It includes AI writing suggestions, a shared content calendar, and per-post analytics. It's free on every plan, including the free tier, while still in beta.

Worth testing if your team juggles a separate scheduling tool.

💡 Prompt of the day

Today's lead story is one founder's bet that an in-house AI team beats an outside agency. Most teams have never actually run the math on whether that bet would work for them. This prompt forces the audit before you fire anyone.

THE AGENCY-VS-IN-HOUSE AI PRODUCTION AUDIT

You are a skeptical marketing operations lead reviewing whether to move creative production in-house with AI tools, not an agency salesperson defending the status quo and not an AI vendor overselling automation. Your only job is to size up whether this move makes sense for me, using only what I tell you. Do not recommend a switch without evidence, and do not assume AI tools will work as well as they did for someone else's case study.

Ask me these questions one at a time, and wait for each answer before asking the next:
1. What do you currently pay an outside agency or freelancer per month, and how many finished pieces of creative does that produce?
2. What is your current customer acquisition cost, and has it moved in the last two quarters?
3. Who on your team, if anyone, already has hands-on experience with AI creative tools like image, video, or ad-copy generators?
4. What's the one thing your current agency does that you'd genuinely be nervous to lose: quality, speed, strategic thinking, or something else?

Once I've answered all four, produce a report with these exact sections:
- Cost Per Piece Today: Calculate my current cost per piece from question 1, and state it plainly.
- Readiness Score (1-10): Rate how ready my team is to run this in-house, quoting my answer to question 3 as the main evidence.
- What I'd Actually Be Risking: Based on my answer to question 4, name the specific failure mode I should watch for if I make this move, not a generic warning.
- A 30-Day Test, Not a Full Switch: Propose one narrow experiment I could run this month to compare in-house AI output against my current agency, using my own numbers from question 1 as the baseline.
- What This Audit Can't Tell You: State plainly what's missing from my answers, such as real output quality, and that this isn't a guarantee of the same results Dutch saw.

If my answer to question 3 shows nobody on my team has AI tool experience, don't recommend a full switch. Build the audit around a hiring or training gap instead. Do not invent capabilities my team hasn't described.

Save the output as a Markdown file I can bring into a budget planning meeting.

🔗 AI visibility is a topic game, not a prompt game (9-minute read)

Semrush and Kevin Indig studied 1,094 topic categories across 50,000 brands in ChatGPT and found only 15% have a brand that owns the topic.

Winning one prompt doesn't count. Ownership means showing up in at least four of five related prompts with a real lead over the runner-up.

Worth reading before your next AI-visibility report, mainly the finding that once a brand builds a wide enough lead, it tends to keep the topic for months.

🔗 Stop selling AI as a replacement for people (7-minute read)

A growth advisor argues the biggest positioning mistake in AI marketing is selling a product as a replacement for people, a pattern he calls "substitution positioning."

He backs it with data from Yale researchers who found no clear evidence of AI-driven job displacement across 33 months of employment surveys.

Worth reading before your next AI product launch, especially if your messaging leans on "does the work your team used to do."

🔗 Why marketing keeps missing the revenue mark (8-minute read)

A marketing consultant breaks down why teams that look busy, like full dashboards and buzzing Slack channels, often aren't moving the business at all.

Her core argument is that marketing gets rewarded for visibility metrics like traffic and MQL volume rather than pipeline and closed-won revenue.

Worth reading before your next KPI review, especially if "vanity metrics look good in a deck" describes more than one of your current reports.

Joe Spector didn't switch overnight. He ran the math first.

If you swapped one piece of this month's creative for an AI-made version, which single metric would decide if it's a keeper?

Behavioral health SEO is entering a new phase. Traditional strategies such as keyword research, content creation, backlinks, service-page optimization, and Google rankings still matter, but the search experience around them is changing.

Google AI Overviews can now answer informational questions directly in search results, while tools such as ChatGPT, Gemini, Claude, and Perplexity are becoming part of how patients and families research symptoms, compare treatment options, and understand levels of care.

As a result, discovery is no longer driven only by ranked links. AI systems can summarize information from multiple sources and influence a patient's understanding before they ever visit a provider's website. This also means fewer clicks are available for traditional organic results, especially for informational searches where users can get the answer directly from an AI-generated response.

A 2025 Pew Research Center analysis found that users clicked a traditional search result in 8% of visits when an AI summary appeared, compared with 15% of visits without one.

Behavioral health providers still need strong traditional SEO, but they also need well-structured content that AI systems can understand, verify, and accurately represent.

How AI Search Differs From Traditional Behavioral Health SEO

Traditional behavioral health SEO focuses on helping individual pages get crawled, indexed, and ranked for relevant searches.

A treatment center may create a service page for inpatient drug rehab, publish supporting articles about addiction and recovery, optimize its Google Business Profile, and earn links from trusted websites. Search engines evaluate these signals and decide where the pages should appear in the results.

AI search works differently.

Instead of simply presenting a ranked list of webpages, generative search tools can extract information from several sources, evaluate the context, and combine those details into a single answer. The user may receive a summary of treatment options, symptoms, costs, or levels of care without opening every source used to create that response.

Figure 1: A Google AI Overview answering whether outpatient therapy is considered rehab. Screenshot by the author.

This changes what content must accomplish.

A traditional SEO page must be relevant enough to rank. Content designed for both traditional and AI search must also make its key information easy to identify, interpret, and verify.

For behavioral health providers, that means important facts should not be hidden inside long, vague paragraphs. Treatment descriptions, eligibility requirements, levels of care, clinical approaches, locations, credentials, and insurance information should be presented clearly and consistently.

AI systems also need enough context to understand what makes one provider different from another. Generic claims such as "we provide compassionate care" offer very little useful information. A more specific explanation of who the program serves, which services are available, who provides the care, and how treatment is delivered gives search systems more substance to work with.

Why AI Visibility Matters in The Behavioral Health Patient Journey

The behavioral health patient journey is rarely a single search followed by an immediate admission.

Patients and families often move through several stages of research. They may begin by asking about symptoms, withdrawal, treatment effectiveness, insurance coverage, or the difference between inpatient and outpatient care. Later, they may compare facilities, read reviews, evaluate staff credentials, examine locations, and contact admissions teams.

AI search is becoming part of those early and middle stages.

A family member might ask ChatGPT what type of treatment is appropriate for someone experiencing opioid dependence. A prospective patient may use Google's AI Overview to understand alcohol withdrawal. Someone else may ask Perplexity to compare residential and outpatient programs before searching Google Maps for nearby providers.

In each case, AI can influence which treatment options the user considers and what criteria they use to evaluate them.

This makes AI visibility important even when it does not generate an immediate website visit. If a provider is consistently cited, accurately described, or recommended during the research process, it may build recognition before the patient reaches the point of contacting a facility.

The opposite is also true. If AI systems cannot understand a provider's services, location, credentials, or areas of specialization, the organization may be excluded from answers or represented inaccurately.

Behavioral health marketers should therefore think beyond traffic alone. Organic sessions and rankings remain important, but they do not capture every way search influences patient decisions. Visibility may also include citations in AI answers or branded searches after an AI interaction.

The Foundations of Behavioral Health AI Search Optimization

Behavioral health AI search optimization depends on whether information can be easily extracted, verified, and accurately included in an AI-generated response. For a behavioral health organization to become a trusted source that AI tools can reference or recommend, its content must be clear and supported by enough context.

Clarity

AI systems should be able to quickly understand what services a provider offers, who those services are designed for, where treatment is available, and what makes the organization different from other providers. Clear headings, concise explanations, and specific service information make it easier for both patients and AI systems to understand the page.

Context

Content must provide enough context for AI systems to represent the provider accurately. A page should clearly explain the level of care, conditions treated, treatment methods, location, admissions process, and any relevant limitations. Without that context, an AI tool may misunderstand the organization or exclude it from an answer entirely.

How Behavioral Health Content Should be Structured for AI Search

Content written for AI search should be organized into clear, self-contained sections that can be easily understood without relying on the rest of the page for context.

Each heading should clearly describe the topic or question being addressed, and the paragraphs beneath it should provide a direct answer before expanding into additional detail. AI systems often retrieve and interpret individual sections of a page rather than reading the entire article from beginning to end. When a paragraph only makes sense after reviewing several earlier sections, it may be more difficult for an AI tool to interpret accurately.

For behavioral health providers, service pages should clearly explain who the program serves, which conditions are treated, what level of care is offered, where services are available and whether insurance is accepted. These details should be stated directly and placed where both patients and search systems can find them quickly.

Content should also use descriptive headings, concise definitions, relevant FAQs, and strong internal links between related articles and service pages. Structured data can provide additional context by helping search systems identify information about the organization, its locations, services, clinicians, and frequently asked questions.

The goal is not to write mechanically for an algorithm. It is to make the content easier for both people and AI systems to understand. Marketing Monk’s guide to getting ChatGPT and Perplexity to recommend a brand reinforces this idea: AI visibility depends on creating authoritative, clearly structured, citation-worthy content.

Why Credibility and Clinical Expertise Matter More in Behavioral Health

Credibility has always mattered in healthcare marketing, but it becomes even more important when AI systems are summarizing information and recommending sources.

Generic behavioral health articles are now easy to produce. A treatment center can use AI to publish large numbers of pages about addiction, mental health disorders, withdrawal symptoms, and treatment options. However, publishing more content does not automatically make the information trustworthy or useful.

Clinical input can help separate credible resources from generic content. Whenever possible, behavioral health organizations should involve licensed clinicians, physicians, therapists, or other qualified professionals in the writing and review process.

Google Search Central's guidance on helpful, reliable content emphasizes clear authorship, first-hand expertise, and additional scrutiny for health-related topics. That reinforces the importance of identifying who created or reviewed behavioral health content and why that person is qualified to do so.

A medical or clinical reviewer can:

• Correct inaccurate or oversimplified claims.

• Add practical insight based on real experience.

• Ensure treatment descriptions reflect the services actually provided.

• Identify statements that need additional context.

• Strengthen the accuracy of health and safety information.

The author and reviewer should be clearly identified, with profile pages explaining their qualifications, professional experience, and connection to the organization.

Trust also extends beyond informational articles. Service pages should include transparent details about staff, credentials, accreditations, treatment approaches, locations, insurance, and admissions. Reviews and consistent business information across the website, Google Business Profile, and reputable directories can further reinforce that the organization is legitimate.

In behavioral health, the strongest content gives patients, families, search engines, and AI systems clear reasons to trust the information being provided.

How AI Search is Changing Behavioral Health SEO Measurement

Traditional SEO can be measured through rankings, impressions, clicks, organic sessions, and conversions. AI search is more difficult to attribute because much of its influence may occur before a patient ever visits the provider's website.

Figure 2: AI search visibility can lead to measurable referral traffic, directional brand search, or direct visits that are difficult to attribute. Graphic created by the author.

When a user clicks a citation from ChatGPT, Perplexity, or another AI platform, that visit may appear as referral traffic in analytics. Behavioral health marketers can track those identifiable AI referrals and evaluate whether they lead to meaningful actions.

However, many AI-assisted journeys leave no clear attribution trail. A patient may discover a treatment center in an AI-generated answer, close the tool, and later search for the provider by name or visit the website directly. In that case, analytics may record the visit as branded organic search or direct traffic without showing that AI influenced the decision.

Citation-monitoring tools can show whether a provider appears in AI-generated answers, while branded-search growth may offer a directional signal that awareness is increasing. Neither metric proves that AI generated a specific inquiry or admission.

Adobe has similarly noted that AI agents can influence decisions before a person reaches a website, while conventional analytics may classify that activity as direct traffic or fail to capture it entirely.

At this stage, there is no reliable way to fully connect every AI interaction to downstream patient actions. Behavioral health marketers should track measurable AI referral traffic and citation visibility, but they should be honest about the limits of attribution. In many cases, AI's role in the patient journey will remain partially unmeasured.

Why Google and Traditional SEO Still Matter

The growth of AI search does not mean behavioral health organizations should stop investing in Google or traditional SEO.

Patients and families may use AI tools to research symptoms, treatment methods, or levels of care, but they still rely on Google when they are ready to evaluate specific providers. They use search results, maps, reviews, websites, and business profiles to compare treatment centers and decide whom to contact.

Traditional SEO also supports AI visibility. Search systems still need accessible websites, clear page structures, strong internal linking, reputable backlinks, consistent business information, and authoritative content. These signals help search engines understand the organization and can make its information easier for AI systems to interpret.

AI search is becoming another part of discovery, not a complete replacement for traditional search. Behavioral health organizations need to optimize for both.

How Behavioral Health Marketers Should Adapt in 2026

Behavioral health marketers should begin evaluating their SEO strategies with AI search in mind, but that does not require abandoning traditional SEO or rebuilding everything around a new set of tactics. The fundamentals still matter. The difference is that writers, strategists, and subject-matter experts must now consider how content may be interpreted, summarized, and cited by AI systems.

This also means being cautious of exaggerated claims about entirely new AI SEO methods. Much of the current conversation centers on shortcuts, mass publishing, and tactics that promise instant visibility in LLMs. In reality, the strongest approach is usually an extension of good SEO: clear structure, reliable sourcing, useful context, and authoritative input.

Although AI search has not replaced Google, it is already shaping how people discover and evaluate behavioral health providers. Organizations that strengthen their existing strategies now will be better positioned as AI plays a larger role in the patient journey.

About the author: Will Warren is the founder of Admita, a marketing agency that helps addiction treatment and behavioral health organizations build focused digital marketing strategies around qualified patient demand.

A new tier of "middle class" creators is proving you don't need a million followers to earn money. Micro-creators average 3.2% engagement versus 1.1% for celebrity accounts, and top performers clear 10%, a gap brands can't keep ignoring.

• Abi Platock had 8,000 followers when she landed her first four-figure brand deal, and expects $50,000 this year from creator work.

• Bonsa Nemera, a dentist earning $76,000 a year at his clinic, pulled in $180,000 from brand deals in two months of posting about oral care and money.

• Anna Fenstermacher made $140,000 last year from content and kept her day job anyway, for the stability it still offers.

None of them are chasing fame, just a second income they can build.

This changes who your influencer budget should actually chase.

A handful of engaged micro creators can outperform one expensive macro-influencer post whose real reach you can't fully verify.

It also changes the deal shape, since these creators lean on affiliate links and ongoing commissions. The instability cuts both ways, though.

A creator's income can swing from zero to $10,000 in a single month.

Watch this segment closely if you sell a niche, considered-purchase product, since niche credibility is exactly what's selling here.

Paramount+ is building a free tier to hook non-subscribers 🎬

Paramount+ is building a free tier, "Free Front Porch," per a Business Insider report. Viewers create a free account and watch select shows.

The free tier follows Pluto TV, and the two will share tech later. Paramount is betting free viewers convert once a paywalled show hooks them.

Hulu and Peacock both tried a similar free layer before, but pulled back. If you buy streaming or CTV media, watch this test closely before fall ad rates lock in. Read more →

PropellerAds lets you run campaigns from inside Claude or ChatGPT 🤖

PropellerAds launched an MCP Connector, letting advertisers run and manage live campaigns through Claude, ChatGPT, or Cursor.

You can create, edit, or pause campaigns across five ad formats without ever opening a dashboard. The connector sets full targeting in chat.

PropellerAds' Julia Larionova said skipping an AI agent will soon feel like refreshing a dashboard by hand. If you run paid social or push campaigns, this is a preview of AI agents touching live budgets. See the announcement →

📌 How Delve planned and ran a $750K OOH campaign (9-minute read)

Delve, the AI compliance startup, ran one of last year's largest out-of-home campaigns across San Francisco, New York, and Austin.

Its guide covers panel counts, lead times, and real numbers, including a roughly $750,000 spend for the San Francisco leg alone.

It also pushes back on testing OOH with a billboard, since the format only proves out at scale. Worth a read before you pitch OOH, mainly on how early you must book locations.

📌 The 3-part formula behind a 45%+ hook rate (6-minute read)

Meta ads creator Lorenzo Pravata breaks down what drives hook rate across the 500-plus ads his team ships each month. His framework is that a hook needs relevance, curiosity, or emotion, and the visual has to match that tone or the script falls flat.

A high hook rate is worthless if it's stopping the wrong people. Worth reading before your next brief, especially if your team reports hook rate with no buyer context.

📌 Build one portable AI marketing workspace (17-minute watch)

This video shows how to build one AI marketing "workspace" that runs across Claude Code and Codex. The creator splits the setup into five parts.

Context, instructions, skills, sub-agents, and tools. It also covers a migration skill that syncs your setup between tools in under two minutes.

Worth watching if your team runs AI agents and wants one system.

💡 Prompt of the day

Most teams report hook rate as one flat number and call it a win the moment it climbs.

That misses the only question that matters: are the people stopping your buyers, or just anyone who will watch anything for three seconds?

THE HOOK QUALIFICATION AUDIT

You are a skeptical direct-response creative strategist, not a brainstorming partner. Your only job is to determine whether a hook attracts qualified buyers or just broad attention, using only the input I give you. Do not praise a hook for being catchy if it doesn't match the buyer.

Ask me these questions one at a time, and wait for each answer before asking the next:
1. Paste your top three current ad hooks, word for word.
2. For each hook, what is the actual 3-second hook rate, if you know it? Say "unknown" if you don't have the number.
3. Describe your ideal buyer in one sentence: who they are and the specific problem they have.
4. What is your current cost per purchase or lead from the ads using these hooks?

Once I've answered all four, produce a report with these exact sections:
- Hook-by-Hook Diagnosis: For each hook, quote it back and score it against relevance, curiosity, and emotion. State which one it's actually using, or say if it's using none.
- The Qualified vs. Broad Read: For each hook, judge whether it would stop your ideal buyer specifically, or just stop a broad audience of scrollers, based on my answer to question 3. Flag any hook that reads generic enough to work for almost any product.
- Visual Pairing Check: Note what kind of visual and pacing each hook needs to land, and flag any hook that would fail if it opened on a static product shot.
- Rewrite Direction: For each weak hook, give a specific direction to fix it, not a full rewritten hook. Ground the direction in the actual problem from question 3.
- What This Audit Can't Tell You: State plainly what's missing, such as real hook-rate data if I answered "unknown," and what to test to fill that gap.

If my answer to question 2 is "unknown" for a hook, do not estimate a number. Score it on the framework alone and say so. Do not invent details about my buyer that I haven't given you.

Save the final audit as a Markdown file for a creative brief.

🔗 How Reddit engineered its comment section into the product (6-minute read)

This piece shows how Reddit turned appreciation into a status game, using karma and awards. The loop: a good comment earns upvotes, upvotes earn karma, and karma pulls people back to comment again.

The piece also covers Reddit's branching comment trees, which turned every post into parallel mini-conversations. Worth reading if your team is thinking past likes and shares toward real engagement loops.

🔗 YouTube shows up in 80% of B2B "demo" searches (8-minute read)

Foundation Inc studied 8,566 keywords across 130-plus SaaS categories and found something stark. YouTube lands in Google's top 10 for "demo" searches 80% of the time, at a steep $49 average CPC per click.

On 166 of those keywords, YouTube outranks five or more SaaS brands at once, covering nearly 54,000 searches a month. Worth reading if your B2B plan treats YouTube as awareness only, not the BOF channel the data says it is.

🔗 Why On overtook Hoka, and Hoka didn't see it coming (7-minute read)

Brand strategist Michelle Wiles explains why On grew to 150% of Hoka's size in four years. Her framework splits category growth into three parts: product innovation, credibility, and code expansion.

She argues Hoka nailed only the first part, the product, the same gap that once slowed Nike's early push into basketball. Worth reading if your brand wants a new category but has only funded the product side.

Today's numbers all came from creators nobody outside their own niche would call famous. That's worth sitting with for a second before your next budget meeting.

If you had $1,000 to spend on one creator partnership this month, would you rather spend it on someone with 5 million followers or someone with 5,000 who actually talks back to their audience?

Unilever CEO Fernando Fernandez told investors that big brand advertising is dead.

He's backing 300,000 creators instead, and social now eats half the company's digital marketing budget. Cannes turned that claim into a full debate last month.

Digiday's Friday briefing laid out both sides. SharkNinja's CMO backed Fernandez, saying social is now the brand's main way to tell its story.

Other marketers pushed back hard, pointing to CTV attention numbers.

This fight is not really about killing TV ads. It's about whether trust must be borrowed creator by creator now, or whether one campaign can still reach everyone at once.

Millions of creators now make content at a professional level, which makes borrowing trust a real option, not a niche one. The next place people search is not a search bar.

It's a chatbot, and chatbots lean on creator content and reviews more than a paid TV buy. CTV still pulls stronger attention than social video, though.

Faire turns retail media into a wholesaler play 🛒

Faire expanded its Promoted Listings ad platform into Europe.

The wholesale marketplace lets brands target the retail buyers who decide what gets stocked, not the shoppers who buy it in stores.

Faire's CRO says targeting a buyer who orders on repeat feels like "fishing with dynamite" compared to scouring the open web for one-time shoppers.

If your brand sells through independent retailers, this is a new lane worth watching. Video ads and takeover placements are on Faire's roadmap. Read more →

Agency owners are ready to sell. Buyers are circling. 💰

Half of independent agency owners want to sell. It's the first data release from Adweek's new Dealroom vertical, which covers marketing M&A.

Owners, buyers, and investors all answered the same questions. The report found more appetite on all sides, but a standoff still keeps most of that appetite from closing deals.

If you run an independent shop, or are evaluating one as a vendor, this data shows the exact gap holding deals back. Read the report →

An audit of 71 businesses found most are invisible to AI search 🔍

An audit of 71 businesses found the average one resolves only 15.6% of what an AI system needs to cite. The gaps ranged from unreadable sites to leadership information.

22 of the 71 businesses had named leadership somewhere on their own site, but AI systems missed it because it sat on an "Our Team" subpage.

These are cheap, fast fixes. Run the same check on your own site before you assume your AI visibility problem is a content problem. Read the full audit →

📌 How Delve ran a $750K billboard campaign (7-minute read)

Delve's growth team shares exactly how it planned and ran one of 2025's biggest OOH campaigns, across San Francisco, New York, and Austin.

The piece breaks down budget math, timing, which media operators own what, and how to write copy that survives a five-second glance.

It's built for any team weighing a first OOH campaign and wondering what "go big or don't go OOH" costs. Read it before asking the CFO for the budget.

📌 Turn Screaming Frog into a visual AI-search auditor (tool)

SEO practitioner Metehan built a free custom script that runs Screaming Frog through OpenAI's Vision API, showing the hidden search queries your product images could match.

It simulates Google's "visual search fan-out" feature, so you can see the gaps before Google does. The code is free on GitHub. A full-site crawl can get pricey.

Worth running on your top product pages if your images say "blue" and the shopper's query says "slate blue button-down."

📌 4 Claude prompts that force your writing to lead with the point (6-minute read)

Marketer Nav Toor breaks down Barbara Minto's Pyramid Principle, the McKinsey writing framework, into four reusable Claude prompts covering structure, opening, logic-checking, and a final executive read test.

Each prompt rewrites a messy draft into an answer-first message.

State your recommendation in line one, then back it up. Worth bookmarking for your next client email, or any message where the reader might stop at sentence two.

💡 Prompt of the day

Today's lead story is a fight over where trust comes from now: one campaign at a time, or one creator at a time. Most media plans still get built on gut feel about that split.

This prompt forces the math.

THE REACH VS. TRUST MEDIA MIX AUDIT

You are a skeptical media strategist, not someone building a case for whatever channel I already prefer. Your only job is to size my current media mix by how it splits between reach (mass, one-to-many channels like CTV, OOH, paid social) and borrowed trust (creator and influencer spend), using only the numbers I give you. Do not recommend a specific split until you have real data from me.

Ask me these questions one at a time, and wait for each answer before asking the next:
1. List every channel you spent marketing dollars on last quarter, and roughly what percent of budget went to each.
2. Of that spend, which channels are "reach" plays (one message, many people, you control it) and which are "borrowed trust" plays (a creator or influencer's audience, not fully yours)?
3. Pick your best-performing campaign from the last two quarters. What was its channel mix?
4. What's one assumption your team makes about where your audience actually is that you've never tested?

Once I've answered all four, produce a report with these exact sections:
- Current Split: State my actual reach-versus-trust percentage split, quoting my own numbers from question 1 as evidence.
- What Worked, and Why: Compare the channel mix from question 3 against my overall split from question 1. Flag if my best campaign looked nothing like my average spend.
- The Untested Assumption: Interpret my answer to question 4. Name one cheap way to test it next quarter.
- Rebalance Recommendation: One specific shift to test next quarter, tied directly to the evidence above, not a generic "do more creators" line.
- What I Can't Tell You: State plainly what's missing from my answers, and what would sharpen this audit.

If my answer to question 1 is vague, like "mostly social," push back and ask for real numbers before scoring the split. Do not invent a percentage I have not given you. Save the final audit as a Markdown file for a budget planning meeting.

🔗 How Buffer turned its whole team into creators (6-minute read)

Buffer built a company-wide hub, complete with workshops and platform-specific playbooks, to help every employee post under their own name.

The bet is that living the same creator journey as Buffer's customers makes for better product decisions and a genuine connection with them.

The program stays intentionally light. Worth reading if your brand is leaning into employee voices but hasn't built any real structure around it yet.

🔗 The audience you are picturing is probably wrong (5-minute read)

SparkToro ran an audience research brief on electrical contractors who do commercial backup power work, a niche worth $4.2 to $5.8 billion annually.

The buyer was a seasoned project superintendent, not the electrician on a ladder most picture. That wrong mental picture would send a media plan toward the wrong channels.

Worth reading as a reminder to check your own buyer persona against real data, not gut instinct, before your next campaign.

🔗 What A24 does differently to build a cult following (7-minute read)

This piece breaks down how the indie film studio built a devoted audience through a distinct brand identity and unconventional, low-cost marketing.

It traces A24's target audience decisions and how they differ from a major studio like Disney. 21 Oscars in 13 years is the proof the approach works.

Worth reading if your brand competes on distinctiveness rather than budget size.

Today's stories all came back to the same tension: reach that everyone sees, or trust borrowed from a smaller group of people already follow.

Where does your current media mix lean, and is that on purpose or by default?

Adidas built a World Cup ad around Lamine Yamal and Lionel Messi two days before the final, betting its two biggest stars would carry the brand.

That kind of deal is a 1980s-era strategy, and athletes are done with it.

The NBA players' union, the NBPA, launched a campaign in June starring Steph Curry and Jalen Brunson, shot by Kendrick Lamar's own agency.

NBPA executive director David Kelly told the union now pushes brands to deal directly with players instead of going through teams or leagues.

"We're trying to get rid of certain gatekeepers," Kelly said, so players and brands can build a deeper relationship from the ground up.

This is the creator playbook overtaking the ambassador playbook, years after a Supreme Court decision opened up NIL deals for college athletes.

In practice, that means longer contracts, creative say, and sometimes equity. Roger Federer took a 3% stake in On Running back in 2019.

Unilever and L'Oréal already run creator armies in the thousands for a sporting event, but those deals are transactional and short-lived.

Dundas argues the smarter move is a timeline strategy: fold the athlete into an ongoing story rather than dropping a product on them once.

Treat the relationship like a one-off campaign, and you will keep losing out to brands offering real say in exchange for real loyalty.

World Cup ad spend shows a sports market that's splitting up 🎯

Sports brands spent $3.3 billion on digital ads across over 50 markets between Q3 2025 and Q2 2026, up 26% year over year.

The NBA drew the heaviest sportsbook advertising of any property measured, at 26% of impressions, nearly double the Super Bowl's 13%.

If your media plan still treats "sports sponsorship" as one line item, this data says the channel mix and the advertiser mix both shift property by property. See more →

Creator video now eats up 26% of daily screen time 📊

Media Dynamics Inc. sized something marketers already felt: creator video now takes 26% of daily TV and video time, or 1.7 hours a day.

That's out of 6.7 total daily hours spent with TV and video across all devices. Linear TV still leads with a 35% share and streaming holds 33%.

The age split matters more: whoever you are chasing under 30 spends more time with creators than anything else on a screen. Read more →

Google's ad business is booming. Investors didn't ask about it. 💰

Alphabet's Q2 call barely mentioned advertising: "advertiser" and "advertising" came up 21 times combined, while "TPUs" came up 28 times.

Investors didn't raise Google's ad business once in the Q&A. Q2 revenue hit $119.8 billion, cloud revenue grew 82% to $24.8 billion, and YT ad revenue grew to $11.1 billion.

If your roadmap assumes Google Ads is the company's top strategic focus, this quarter's plan says to plan for less attention. Get the story →

🛠️ MCP HQ: Find the right MCP server (Directory)

MCP HQ is a directory for browsing maintained MCP servers by what you are building.

Servers are ranked by install count, each with an install config you can copy straight into your setup. Worth a bookmark if your team is wiring Claude or another AI agent into your stack and keeps guessing which server to trust.

📌 How to get people to tell you what they think (7-minute read)

Email Love rounds up survey emails from Sony, Tecovas, and Huggies, then pulls out the pattern behind the ones that get replies. The best ones state how long it takes, explain what's in it for you, and let you answer without leaving your inbox.

Athletic Brewing's version lets you rate a product built into the email. Worth stealing if your lifecycle program still sends a feedback link to a form nobody opens.

📌 How to answer annual planning questions (in days) (6-minute read)

This piece breaks planning season into three questions every marketer gets asked: growth, budget, and whether you will hit the number.

The answer to the first is to skip the strategy doc and list 3 to 5 big bets. The other two are backed with free tools: a budget calculator and a bottom-up revenue forecaster. Worth reading if "we'll figure out the number later" is still your default.

💡 Prompt of the day

Today's lead story is about brands moving from one-off athlete campaigns to ongoing partnerships. Most talent contracts never get audited for which one they actually are.

This prompt forces that audit before your next renewal.

THE PARTNERSHIP STRUCTURE AUDIT

You are a skeptical talent partnerships strategist, not an agent trying to justify a deal already in motion. Your only job is to determine whether a creator, athlete, or influencer partnership is built for longevity or for a single transaction, using only the details I give you. Do not assume a deal has more structure than what I actually describe.
Ask me these questions one at a time, and wait for each answer before asking the next:
1. Name the person and the deal length: one campaign, one quarter, one year, or ongoing with no set end.
2. What does this person actually get beyond a posting fee: creative input, product say, equity, a multi-year term, or nothing beyond the payment?
3. Paste the original brief or one-line goal this partnership was built around.
4. Has this person's contract ever been renewed, or given them a say in creative direction? Describe it, or say no.

Once I've answered, produce a report with these sections:
- Structure Verdict: Campaign or Timeline, quoting my own answers as evidence.
- Longevity Risk: What's missing that would make this person walk at the next renewal, based on question 2.
- One Upgrade to Offer: A single, specific change to the deal structure, not a full renegotiation, grounded in what I've described.
- Talent-Facing Pitch: A short paragraph I could send this person, using only the details I've given you.
- What This Audit Can't Tell You: State plainly what's missing from my answers and what would sharpen the verdict.

If my answer to question 2 is just a posting fee with nothing else, say so plainly and build the audit around that gap instead of inventing structure I haven't described. Save the output as a Markdown file for a partnerships review doc.

🔗 AI brand visibility: You are tracking it wrong (9-minute read)

This piece says most AI visibility tracking follows old keyword-rank logic. It doesn't work since a prompt can return a different answer each run.

Prompts like "best CRM in 2026" describe a buyer who doesn't exist, and a brand they audited had strong visibility there but zero on the high-intent prompts real buyers use.

The fix is better ones: build them around real personas and intent stages. Worth reading before your team reports AI visibility like a search ranking.

🔗 The entire marketing strategy of the 2026 World Cup (15-minute read)

This piece ranks and roasts over 100 World Cup campaigns, splitting them into celebrity-driven ads, reactive stunts, and brands that built a story.

The sharper data point is that non-sponsor brands like Nike and Levi's out-engaged official World Cup sponsors online by nearly two to one.

Worth reading before your next big-event sponsorship pitch, especially the case that a bigger logo budget doesn't guarantee bigger attention.

🔗 The best AI-native GTM plays you are not running (9-minute read)

Kyle Poyar and Brendan Short update their 2025 list of automated GTM plays for a world where an AI agent runs the workflow.

Closed-lost re-engagement no longer fires on a 9-month timer. An agent now scans for signals and drafts an email saying the reason the deal died. Read if your GTM team still treats automation as a trigger and a template, not a system that decides when to act.

Some of today's stories are about relationships built to last one campaign.

Which relationship in your marketing stack do you wish you had structured for the long haul instead of one deal at a time?

Omnicom Media Group kept Uber's global media account after a review that ran through the second quarter of 2026. Uber added Brazil to Omnicom's scope.

The account is worth close to $880 million a year, split across NA ($490 million), EMEA ($210 million), APAC ($140 million), and Latin America ($40 million).

WPP kept its slice of the business in Asia-Pacific, worth about $136 million a year.

Omnicom first won the global account away from WPP back in 2023. Uber had already handed Omnicom sports marketing duties back in May.

This is what account loyalty looks like when a client keeps growing.

Uber didn't only keep Omnicom around. It handed over more work: a new region, plus sports marketing added months before this review.

That kind of expansion usually means the agency is proving real return.

For agency teams, an $880 million account with growing scope is about the strongest job security right now, since holding companies face real financial pressure elsewhere.

It also says something about how global reviews play out: incumbents tend to win when they can show results across many markets at once.

Intersnack buys Utz Brands for $2.9 billion, no longer a US outsider 🥔

Germany's Intersnack Group agreed to take Utz Brands private for $14.25 a share, a 91% premium over Utz's stock price. The deal values Utz at $2.9 billion and marks Intersnack's first foothold in the US snack market.

Utz CEO Howard said Intersnack's "marketing, manufacturing, and technology capabilities" would help the brand invest in its portfolio.

Expect Utz's ad and shopper-marketing budgets to eventually plug into a much bigger European parent with category muscle. Read more →

Apple quietly fixed a year-old bug that unmasked "hidden" emails 🔒

Apple patched a flaw in its Hide My Email tool, after sitting on it for a year.

The bug let almost anyone see the real address behind a Hide My Email alias. Apple says any alias made before July 7 could still be exposed.

If your signup forms see a lot of Hide My Email addresses, don't assume those users are now fully protected. Treat any list built before July 7 as if some real addresses leaked, and avoid holding raw email logs longer. Read the report →

Scrutiny grows over the "data broker loophole" ⚖️

Privacy researchers and lawmakers are pushing harder to close a gap that lets government agencies buy Americans' location and behavior data.

Just Security argues AI tools make that purchased data more powerful, since a model can now find patterns across huge datasets in minutes.

If your stack includes any third-party location or behavioral data vendor, this is the same data supply chain now under scrutiny. Read the analysis →

📌 Turn Claude into a LinkedIn carousel designer (6-minute read)

Charlie Hills shows how to reverse-engineer your best LinkedIn carousel into a Claude Skill. Upload one example, and Claude turns it into a file that captures your structure, tone, and branding.

Then, one prompt produces a full carousel brief, and another builds the slides. Worth trying if your team redesigns carousels from scratch instead of working from a system.

📌 Google's GA4 reporting playbook, straight from the source (Playbook)

Google breaks down GA4's four reporting surfaces: Standard Reports, Explore, APIs, and BigQuery. It also maps UA terms to GA4 equivalents. It's a reference guide.

It clearly explains why your GA4 numbers don't match across tools. Bookmark it for the next time someone asks why Explore and Standard Reports disagree.

📌 How Owner.com grew its audience 42x (7-minute read)

Eric Linssen interviews Owner.com's content lead on the three-word strategy behind its growth. Every idea has to pass what they call the "busy restaurant owner test."

The piece also breaks down the roles you need to actually pull this off, from scriptwriter to a dedicated retention editor. Worth reading if your content team is execution-poor.

💡 Prompt of the day

Most marketers only check whether a vendor contract renewed.

But that misses the signal: if the relationship's scope is growing or shrinking. This prompt forces a harder look, using your account history as evidence.

THE AGENCY RELATIONSHIP HEALTH AUDIT

You are a skeptical procurement and marketing operations advisor, not someone trying to justify keeping an existing vendor. Your only job is to assess whether an agency or vendor relationship is actually healthy, using only the evidence I give you. Do not treat "we renewed the contract" as proof of health on its own.
Ask me these questions one at a time, and wait for each answer before asking the next:
1. Which agency or vendor relationship are we auditing, and how long have we worked with them?
2. What was their scope of work a year ago, and what is it today? Be specific about channels, regions, or deliverables added or dropped.
3. Has our spend with them grown, shrunk, or stayed flat over that time, in real numbers if you have them?
4. Name one thing they have done in the last quarter that surprised you, in a good or bad way.

Once I've answered all four, produce a report with these exact sections:
- Scope Trend: State plainly whether their job has grown, shrunk, or stayed flat, quoting my own answer to question 2 as evidence.
- Spend-to-Scope Ratio: Compare the spend trend from question 3 against the scope trend. Flag it if spend grew faster than scope, since that often means quiet price creep.
- The Surprise Signal: Interpret my answer to question 4. A good surprise suggests real initiative; a bad one suggests coasting.
- Renewal Verdict: Renew As-Is, Renegotiate Scope, or Run a Review, tied to the evidence above.
- What I Can't Tell You: State plainly what's missing from my answers, and what data would sharpen this audit.

If my answer to question 2 is vague, like "about the same," push back and ask for specifics before scoring the scope trend. Do not invent scope details I haven't given you. Save the final audit as a Markdown file for a vendor review meeting.

🔗 Inside the agency rebuilding the US Army's recruiting machine (7-minute read)

Monocle goes inside AEMO, the in-house marketing arm the US Army built after firing its old ad agency over a corruption scandal in 2019.

Its $4 billion, 10-year contract with DDB Chicago runs like a Fortune 500 marketing operation. The Army hit its recruiting target for the second year running.

Worth reading for a long view on how one brand's messaging survives, and bends, across very different leadership eras.

🔗 When algorithms curate culture, what do we lose? (5-minute read)

Observer looks at what happens when Spotify, TikTok, and other algorithmic feeds become the main way people discover music, style, and ideas.

The piece argues that real curation used to mean a human with taste narrowing down the noise, and algorithms replace that judgment with pure engagement math.

Worth reading if your team leans on algorithmic personalization and hasn't stopped to ask what a human curator would catch that the algorithm won't.

🔗 A more accurate way to measure consumer engagement (8-minute read)

McKinsey's Kabir Ahuja argues that audience size and time spent are the wrong metrics.

He proposes an "attention equation" that mixes a commercial quotient, who's watching, and how easy they are to sell to, with an attention quotient, how focused they are.

By his numbers, a third of why content gets monetized comes down to attention quality.

Worth reading before your next media plan, especially if your buys still lean on reach and time spent as the main numbers that matter.

Today was all about relationships getting tested, whether that's an agency proving its scope, a brand changing hands, or a data pipeline facing new scrutiny.

Which vendor or agency relationship have you not reviewed in over a year, even though the deliverables have changed?

You're paying Klaviyo prices. You don't have to.

Omnisend's migration team moves everything - automations, templates, full setup - in 5 days. You do nothing. Show up on day six paying up to 35% less.

And once you're in, it doesn't feel like any ESP you've used before. You don't really "use" it. You ask Claude or ChatGPT things like "where am I losing revenue?" and it pulls answers straight from your account. Not generic AI guesses. Your real data.

It audits deliverability, builds segments, writes campaigns, and hands you a monthly summary ready to drop into a client report. All from a chat window.

150,000+ brands run on Omnisend, averaging $79 back for every $1 spent.

4.8 stars on Shopify, 4-minute human support, and 250+ integrations, such as Triple Whale, Smile.io, Aftership, and more. Five days to switch. A fraction of the cost.

Start Saving Up to 35%

Partner with us

Brussels published its final guidelines and Code of Practice for AI content transparency on July 20. The rules come from Article 50 of the EU AI Act.

Full obligations kick in on August 2, but by today, brands have to sign the Code for a compliance safe harbor. Anyone who builds with AI must watermark synthetic content.

Anyone who deploys it, including advertisers, agencies, and publishers, must label deepfakes and AI-generated text on public-interest topics.

Also, in advertising, the lighter rules for art, satire, and fiction exclude persuasive commercial content, so a synthetic spokesperson in an ad needs the full label.

Fines up to 15 million euros or 3% of worldwide revenue, whichever is higher.

It comes right after Google put AI ad-labeling onto advertisers. Together, the two moves put the compliance burden on whoever signs off on the ad, not the platform running it.

Disney and Kraft Heinz ink a landmark alliance 🤝

Disney and Kraft Heinz announced a multi-year alliance spanning Disney's North American parks, resorts, cruise line, and streaming platforms.

Ten Kraft Heinz brands, including Heinz and Kraft Mac & Cheese, will appear across Disney venues and can feature Disney characters in their own marketing.

For any legacy brand, this is a clean case study in using entertainment licensing to reach families where they already spend time.

Watch whether this playbook shows up next in Kraft Heinz's NFL deal or gets copied by a rival chasing the same turnaround. Read the announcement →

A study says identity errors can erase 70% of your measured ROI 📊

LiveRamp and the MMA published a study modeling how measurement errors distort reporting. Poor identity precision destroyed simulated ROI by roughly 70%, turning a true $1.50 return into a measured $0.43.

The research splits the problem into missing data and identity mismatch. Either one can make a working campaign look like a loser worth cutting.

Worth reading, but the numbers come from a synthetic simulation, and LiveRamp sells the identity-resolution tools it says fix the problem. See the findings →

E*Trade goes from goat-herders to aliens to sell crypto trading 👽

E*Trade rolled out spot trading for Bitcoin, Ethereum, and Solana.

The launch campaign, "What's Next," traces trading history from ancient goat-herders through a 1980s day trader to a hipster alien reaching for cold brew.

The ad leans on institutional trust to make crypto feel familiar rather than fringe, landing while crypto markets have a rough stretch going.

If your brand sits in a category customers see as risky, this is an example of using humor and history to lower the guard. Watch the E*Trade spot →

📌 Cuey: compare ChatGPT, Claude, and Gemini side by side (PDFs)

The best AI is right only 61% of the time, per Artificial Analysis.

Cuey sends one prompt to ChatGPT, Claude & Gemini in one tab, so you cross-check the answer before you ship it. No tab-switching. No copy-pasting. Add to Chrome, free.

📌 How to onboard non-technical marketers to automation tools (9-minute read)

HubSpot lays out a 14-day framework for getting non-technical marketers comfortable with automation tools, from a practice sandbox to a live campaign.

It leans on HubSpot's own platform as the running example throughout.

Still worth a skim for the structure: sandbox first, real send by day seven, workflow by day ten, even if you swap in a different tool.

📌 The ever-expanding branding briefcase (tool)

A designer built a free bundle of branding resources: a logo inspiration ebook, a brand strategy workbook, a discovery-call question bank, and a brand personality matrix.

Everything is downloadable. It's built for designers and brand strategists doing client work. Worth bookmarking before your next client kickoff call.

💡 Prompt of the day

Today's EU rules force a question most teams have never checked: would any of your AI-generated marketing assets need a label under Article 50 right now?

This prompt runs your content library through the test.

THE AI CONTENT LABELING COMPLIANCE AUDIT

You are a skeptical compliance-minded marketing operations lead, not a lawyer and not someone trying to reassure me everything is fine. Your only job is to flag where my AI-generated content might need a label under EU rules, using only what I tell you. Do not give legal advice or a final compliance verdict. Flag risk, not certainty. Ask me these questions one at a time, and wait for each answer before asking the next:
1. List every place you currently use AI-generated images, video, voice, or avatars in marketing content. Include ads specifically if any exist.
2. Does any of that content show, or closely resemble, a real person, real event, or real place? Describe each case, or say "none."
3. Do you publish AI-generated or AI-modified text about news, politics, health, or other public-interest topics without a named human editor reviewing it first? Describe your review process, or say "no review."
4. Does any of your audience sit in the EU or EEA, even as a small share of total reach?

Once I've answered all four, produce a report with these sections:
- Likely Labeling Risk: For each item from question 1, state whether it plausibly falls under the deepfake or public-interest-text rules, quoting my own description as the basis for that call.
- The Ad Exception That Isn't: Flag explicitly if any AI content sits inside a persuasive ad, since ads don't get the lighter creative-work exemption.
- Editorial Review Gap: Based on question 3, state plainly whether my current review process would meet a "substantive human review" bar, or whether it looks more like a formality.
- What Needs a Label Today: A short, direct list of content that should carry a visible AI label before it runs again.
- What I Can't Tell You: State plainly what's missing from my answers, such as jurisdiction detail or a lawyer's read, and that this isn't a substitute for one.

If my answer to question 4 is "no EU audience," say clearly that domestic rules may still differ, and this audit only covers the EU angle. Do not invent content categories I haven't described. Save the output as a Markdown file for a compliance review meeting.

🔗 Modern luxury: How creativity translates ideas into value (7-minute read)

Brandingmag's Maurizio Ribotti argues creativity in premium brands is an upstream function that directly shapes decisions, systems, and financial outcomes.

He draws a hard line between creativity, generating ideas, and innovation, executing them at scale, arguing most brands blur the two.

Worth reading if your team still treats creative as the last step before launch.

🔗 When AI becomes the gatekeeper of brand visibility (5-minute read)

Adweek recaps a Cannes panel where marketing leaders discussed earning trust when AI answers a question before a customer reaches an owned channel.

Unilever's Selina Sykes makes the sharpest point. If a product claim reads differently on Walmart than what a creator says on Reddit, she argues, the AI won't recommend it.

Worth reading if you have never checked if your brand says the same thing everywhere an AI might look.

So, which number, claim, or "AI-generated" label in your own marketing could you not confidently defend if someone questioned it today?

AI Search / AEO Playbook

2026 AEO Playbook: The 4 plays behind Ramp, Carta, and Webflow's AI search strategy

Most teams still measure AI search like they measured Google. That's the mistake.

Ranking and being cited aren't the same job. A page at position one can miss every AI answer. Another on page two can get cited in ChatGPT, Gemini, and Perplexity the same week. What changed is the selection logic.

AI engines aren't ranking pages; they're choosing which source is safe to quote. That runs on structure, evidence, freshness, and off-site presence.

AirOps analyzed 15M+ AI queries to find what earns citations. Clear heading hierarchy: 2.8x more likely to be cited. 85% of brand mentions come from off-site sources. 70% of cited pages were updated in the last 12 months.

The playbook breaks down the 4 plays behind how Ramp, Carta, and Webflow operationalize this, plus a 90-day action plan.

Get the playbook

Partner with us

The FDA said that a Taylor Farms lettuce sample that tested positive for cyclospora was a false positive. The recall on Taylor Farms' central Mexico iceberg lettuce still stands.

Traceback data and patient interviews still point to that lettuce as the source. Over 1,600 cases have been confirmed, with 94 hospitalizations and no deaths.

Foot traffic at Taco Bell sank 19% below average on the outbreak's worst day. Yum Brands shares fell 9% in the week before the reversal landed.

This isn't a food safety story for marketers. It's a lesson in what happens when the facts change mid-crisis and your statement doesn't catch up.

Taco Bell said "immediate action" on day one, which was the right instinct. But the false-positive reversal created a second news cycle on top of the first.

Most brands have no plan for round two. Foot traffic and the stock price had already moved before the science was settled, which is the real risk here.

Markets and customers punish you on the first draft of a story, not the corrected one.

Coca-Cola's global rebrand skips the logo 🎨

Coca-Cola rolled out a new global design system with agency JKR, built to make the brand feel "more Coca-Cola" when it shows up everywhere.

The brief skipped a logo and asked for tighter consistency across the red-and-white palette, the Ribbon, the Arden Square, and the Spencerian text.

If your brand has drifted between what people picture and what actually ships across markets, study this fix pattern. See JKR's approach →

Stadiums have become full ad-tech platforms 🏟️

Sports venues have turned into full-blown ad-tech platforms with LED signage, "Instagrammable" fan zones, and Wi-Fi tracking. The shift built up over the past decade and is now standard at nearly every major stadium, rink, and arena.

Venues use "sophisticated market-segmentation strategies" to pull in higher-spending fans. If you plan sports spend, ask your media partner what data layer comes with the signage. Read the breakdown →

7-Eleven bets discovery beats convenience 🏪

7-Eleven Singapore launched "Find the Good Stuff," a brand campaign done with agency TAG that repositions the chain around surprise.

It follows a DFI Retail Group survey, which found "exciting," "community-minded," and "innovative" now beat convenience as trust drivers.

The campaign's hero film has passed 3 million views on Instagram and Facebook. If your category gets picked mostly on default, this is what building a reason to choose it on purpose looks like. Read the breakdown →

📌 6 examples of what real brand differentiation looks like (9-minute read)

HubSpot walks through six brands that made a real difference, from Revolut's frictionless banking app to Liquid Death turning canned water into a punk-rock bit.

It closes with an eight-step framework covering mission, culture, tools, and visual identity, built around a "Difference Engine" model.

Worth a skim before your next positioning meeting, especially the Liquid Death and Fenty sections if your category still reads as interchangeable.

📌 How to market through every phase of the holiday season (6-minute read)

Startups Magazine breaks the holiday shopping season into seven phases, from early "proud preparedness" shoppers in October to post-Christmas "Self-Gifters.”

Each phase comes with a unique psychological driver, and it leans on Intuit Mailchimp's Holiday Shopping Unwrapped report. Worth reading now if your team is starting Q4 planning, since the framework only helps if it's built into the calendar before October.

📌 How one team turns 20 pieces of content into 160 posts a week (7-minute read)

Dickie Bush breaks down his team's content engine across six platforms, and the throughline is that only about 20 pieces get created from scratch each week.

He lists three rules: one CTA per platform, build the cadence first, and don't try to copy this until you have been consistent for 90 days. Worth stealing if your content ops still means one person manually posting the same idea five different ways.

💡 Prompt of the day

Taco Bell had a statement ready on day one. Most brands don't, and even fewer have a plan for when the facts shift after they have already spoken.

This prompt forces you to find the gap before a real crisis does.

THE BRAND CRISIS RESPONSE READINESS AUDIT

You are a skeptical crisis communications advisor, not a PR partner assuming everything will blow over. Your only job is to find the gaps in how this brand would actually respond to a crisis, using only the answers I give you. Do not soften a gap just to be encouraging.

Ask me these questions one at a time, waiting for each answer before the next:
1. What's the worst plausible incident for your brand this year — a product recall, a data breach, a public complaint that goes viral? Name it.
2. Do you have a written crisis communication plan today? Paste it, or say "we don't have one."
3. Who is authorized to approve and post a public statement, and how fast can they move? Name the role, not the person.
4. Has your brand ever had to walk back or update a public statement after new facts came in? Describe what happened, or say "not yet."

Once I've answered, produce a report with these sections:
- Readiness Score (1-10): Rate the plan's actual coverage, quoting my own answers as evidence.
- The Gap That Bites First: Name the single most likely failure point given the scenario in Q1.
- The Second Statement Problem: Facts often change mid-crisis. Define who updates the public message and how fast, based on Q3 and Q4.
- Channel Checklist: List every channel that needs the same message within the same hour: X, Instagram, press, internal Slack, customer email.
- What This Audit Can't Tell You: State plainly what's missing from my answers and what would sharpen the score.

If I say "we don't have one" for Q2, don't invent a plan. Build the whole audit around that gap instead. If my answer to Q3 is vague, like "we'll figure it out," push back and ask for a named role before moving on. 
Save the output as a Markdown file for a crisis-readiness review.

🔗 Does your website need an llms.txt file? (10-minute read)

Backlinko walks through llms.txt, a text file to tell AI crawlers which pages on a site matter most. No major AI platform has confirmed it yet, and Google has said it won't.

The piece still makes a case for testing it anyway, since setup takes an afternoon and players like Stripe already have one live. Worth reading if you are deciding between a low-risk afternoon experiment and waiting for the standard to prove itself first.

🔗 Why "affordable" brands now sound like luxury ones (6-minute read)

Marketing Brew looks at brands like Josh Cellars and Carbone Fine Food, both positioning everyday products as part of an aspirational lifestyle.

VML's Carolyn Rooke argues social media made this necessary, turning ordinary categories into something people want to be seen buying.

Worth reading if your brand sits in the middle of the price ladder and hasn't picked a lane between accessible and aspirational.

🔗 AI search is making YouTube creators more powerful (7-minute read)

Jellyfish analyzed 27 million AI assistant responses and found YouTube creator content shows up in one in four answers. AI models favor detailed, niche creator videos over content from brands or big-name influencers, mainly anything longer than 10 minutes.

Worth reading if your team still treats YouTube as a brand-awareness channel rather than a source AI is actively citing back to shoppers.

Today's stories were all about who's actually in control of how a brand shows up right now, whether that's on purpose or not. Where does your brand currently have the least control over how it shows up to customers?

Nielsen committed to seven changes in how it measures TV audiences.

It touches on weighting, smart TV recognition, co-viewing, and how Nielsen counts Hispanic households. They are tied to keeping official MRC accreditation through 2026.

The group publishing the details, the Video Advertising Bureau, has spent this year sparring with Nielsen over a separate report it called misleading.

Nielsen is sharing preview data all summer so buyers can see the shift before it lands. The new numbers will set fall TV ad rates.

Each of these seven changes shifts the audience number your media buy gets priced against. Hispanic Universe Estimates and Co-Viewing touch every dataset Nielsen sells.

Nielsen already spent this year defending its neutrality after a separate dispute over how it weighted linear versus streaming viewing.

That fight showed that the same viewing data can tell two different stories depending on the method. Buyers who skip the preview data this summer negotiate fall rates on numbers they never watched move.

Commerce Media stops being a retail-only story 🛒

Commerce media, ad networks built on a company's own purchase and behavior data, no longer belong to retailers alone.

Axios reports that half of the major commerce media networks launched since 2022 come from travel, delivery, and financial services companies.

The bigger question for planning season is which of your customer touchpoints could carry ads before someone else's network gets there first. See the breakdown →

TikTok locks in a multi-year deal with the NBA and WNBA 🏀

TikTok signed a multi-year global partnership with the NBA and WNBA.

The two leagues' combined TikTok accounts already reach 30 million followers, and in-app interest in both leagues is up double digits since January.

TikTok is pitching itself as a discovery-to-broadcast funnel. If you buy sports media, TikTok just gave itself a stronger case for a line item beyond the usual in-feed spend. Read the writeup →

📌 Why one testimonial beat a full-page redesign (9-minute read)

Growbo breaks down a Swiss hospitality school's test: one testimonial placed above a lead form beat every redesign the team tried, lifting form completions by 50%.

The tactic rests on placing real social proof at the exact moment someone decides, paired with an unpolished, real photo rather than a stock image.

Worth a look if your team keeps chasing bigger fixes for a conversion problem.

🛠️ Pomelli: turn your website into a brand profile AI can design from (tool)

Pomelli, from Google Labs and DeepMind, scans your website and builds a "Business DNA" profile of your brand, tone, colors, and fonts included. From there, it generates campaign ideas and on-brand social assets you can edit inside the tool and export.

It's built for small teams without a designer, currently free in the US, Canada, Australia, and New Zealand. Worth a test run if your brand posts happen ad hoc.

📌 Why Webflow stopped trusting aggregate traffic numbers (6-minute read)

Kyle Poyar interviews Webflow's VP of growth on why the company now treats total traffic as a misleading metric. ChatGPT referrals convert at 24%, 6X Google's rate, even as Webflow's overall traffic looks flatter than before.

The piece lays out seven concrete shifts that any team relying on organic or paid search can borrow directly. Worth reading before your next traffic report gets read as bad news.

💡 Prompt of the day

Most campaign briefs describe the campaign.

Almost none define exactly when to stop running it. This prompt forces both, and won't let the model soften a contradiction in your own answers just to keep things moving.

CAMPAIN BRIEF & KILL CRITERIA BUILDER

You are a skeptical marketing operations lead reviewing a campaign before it launches. You are not a cheerleader for the idea, and you do not add enthusiasm to the input that doesn't support. Your job is to force clarity, not friendliness.
Ask me the following questions one at a time. Wait for my answer before asking the next one.
1. What is the campaign, in one sentence, and what channel(s) will it run on?
2. What is the single metric that decides success, and what number counts as a win?
3. What is the budget and the timeline?
4. What would have to be true for this campaign to fail even if it hits its metric? (Examples: wrong audience, cannibalizing another channel, unsustainable CAC.)

Once I've answered all four, produce a brief in this exact format:
- Campaign Brief: Objective: [Restate my answers to Q1 and Q2 as one clear sentence.] Budget & Timeline: [From Q3.] Success Metric: [The single number from Q2. If I gave more than one metric, pick the one that matters most and say why the others are secondary.]
- Kill Criteria: List exactly three conditions that mean this campaign should be paused or killed before the full budget is spent. Each one must be specific and measurable, not vague ("if it's not working"). Base at least one kill criterion directly on my answer to Q4.
- Checkpoint Schedule: Propose two checkpoints between launch and the end date where the kill criteria get checked against real data. Name what data to pull at each checkpoint.
- What I'm Assuming: List every assumption you made because my answers were incomplete or ambiguous. If an answer was missing entirely, say so directly instead of guessing.

If any of my four answers contradict each other, stop and point out the contradiction before producing the brief. Do not smooth over the conflict.
Save the final brief as a Markdown file I can drop into a campaign folder or Project.

🔗 The "Queen of Email" on why open rates are lying (12-minute read)

Email consultant Beth O'Malley argues most email teams obsess over the wrong signal, ignoring whether a send moved someone forward. Her bigger point is that unsubscribes are a healthy signal, and a smaller, more honest list beats a big one nobody reads.

Worth reading if your team still reports opens as the headline number.

🔗 No single channel wins for indie game marketing (7-minute read)

Three indie publishers say there's no reliable formula for marketing ROI anymore, since an algorithm change can undo a strategy overnight. No More Robots spent $50,000 on ads in one launch window to build awareness, treating paid media as a spotlight.

Worth reading with a caveat. Treat it as an analogy rather than a direct playbook.

Some stories today share the same thread: numbers you have trusted for years are shifting under you, and most teams keep reporting them anyway.

Which metric do you still report every month because leadership expects it, even though you've stopped fully trusting it?

Stockholm's Patent and Market Court ordered Google to pay Klarna, formerly PriceRunner, damages across the UK, Denmark, and Sweden.

The court ruled Google's 2017 fix for its EU Shopping case never actually stopped the self-preferencing regulators condemned. Google kept its Shopping Unit in a prominent box up top while its Panda algorithm kept demoting rivals below it.

The UK portion alone totals 950 million pounds, plus 316 million in interest. Klarna sought 4X that. The court built its own damages model instead. Google plans to appeal.

This is the first ruling to test whether Google's remedy actually worked. Google's search dominance never dipped below 82% desktop share.

That dominance, along with data showing 28% of clicks go to the first organic result, led the court to tie lost rankings directly to lost revenue.

For any brand growing through ranking rather than paying Google directly, this is the clearest playbook yet for proving that harm in court.

6sense puts GTM intelligence straight inside Claude and ChatGPT 🤖

6sense launched an MCP Server that puts its GTM intelligence, including account insights and buying-stage data, in any MCP-compatible AI agent. It's live in open beta.

Teams can ask Claude or ChatGPT which accounts turned into qualified buyers this week and get answers grounded in 6sense's own data.

If your team runs AI agents for account research, it's worth testing. Read more →

Publishers say Gemini trained on books without permission ⚖️

Hachette, Cengage, Elsevier, and Scott Turow filed a class action alleging Gemini was trained on millions of books supplied only for Google Books, Play Books, and Scholar.

The complaint also alleges Google pulled books from pirate sites and stripped copyright data from the training set. No court has ruled yet, and Google hasn't responded.

"Trained on it" and "licensed to use it" are different legal questions now. Find more →

Advertisers push back as Google defends its August bidding change 📊

Google Ads Liaison Ginny Marvin pushed back on a thread where a consultant called the update the most frustrating change he's seen.

It only touches campaigns marked "Limited by budget" effective August 17. Later, campaigns beating their stated target will drift back toward the number in the settings.

If any campaigns use Target CPA or Target ROAS, check them now. See more →

📌 How to fix a B2B newsletter (or start a great one) (6-minute read)

This piece breaks down why most B2B newsletters read like links dressed up as marketing. The fix is to send from a real person and write in "you" language.

She backs it with a blunt frequency rule, monthly to start, and treats the pitch itself as a P.S. Worth reading before your next newsletter meeting.

📌 How to beat the new LinkedIn algorithm in 10 steps (7-minute read)

Marketer Logan Gott lays out ten shifts he's seeing work on LinkedIn. His strongest claim is that saves and DM shares now matter more for reach than likes or comments.

He also argues that links in the post body get suppressed, so the workaround is the first comment instead. Useful if your LinkedIn playbook predates video and PDF carousels.

📌 Write 30 days of social content in 90 minutes using AI (9-minute read)

This piece walks through a system for batching a month of social content. It leans on a 40/30/20/10 mix of educational, engagement, personal, and promotional posts.

It's a good refresher if your team still writes daily and burns out mid-month. Pair it with the LinkedIn breakdown above for the fuller content-ops picture.

💡 Prompt of the day

Today's lead story is a court putting a hard number on platform dependency that most companies never calculate themselves. This prompt forces math for you.

THE PLATFORM DEPENDENCY RISK AUDIT

You are a skeptical revenue analyst, not a growth hacker pitching diversification tactics. Size how much of the business depends on one platform you don't control, using only the evidence I give you. Don't soften the finding.
Ask me these one at a time, waiting for each answer:
1. Which single platform, search results, a marketplace, an ad network, or an app store, sends the largest share of your traffic or revenue?
2. What percentage of last quarter's revenue traces back to it? Paste the number, or say "we don't track this."
3. Has that platform changed its algorithm, policy, or fees against you in the last 12 months? Describe it, or say no.
4. Do you have a second channel already producing real revenue, or is diversification still theoretical?

Then produce a report with these sections:
- Dependency Score (1-10): Rate our exposure, quoting the answer that drove the score.
- Exposure Timeline: How fast this could hurt us if the platform changes again, based on question 3.
- What Breaks First: The single most fragile part of the business given this dependency.
- One Diversification Bet to Fund This Quarter: A specific, fundable move, grounded in question 4.
- What I Can't Tell You: What's missing from my answers and what would sharpen the score.

If my answers are vague, especially the percentage, ask for a real number before scoring. Don't invent a figure I haven't given you. Save the output as a Markdown file I can bring into a leadership meeting.

🔗 Automated email segmentation (7-minute read)

HubSpot walks through building segments that update automatically. It covers cleaning data, then linking dynamic lists to automated send sequences.

It's introductory territory. Useful if segmentation is new to your team.

🔗 How to build a cult brand (7-minute read)

This piece breaks down the mechanics behind brands like Patagonia, YETI, and Glossier that inspire fans to defend them unprompted.

Patagonia's "Don't Buy This Jacket" ad is the sharpest example. Worth reading before your next brand review if "friendly and approachable" is still in your guidelines.

🔗 B2B marketing in the 2nd half of 2026 (10-minute read)

ABM consultants argue most "AI-native marketer" posts hide a bubble, citing research that found most firms saw no productivity gain from AI over three years.

Read past the offers for the actual data. Worth it if your team feels pressure to prove it's "AI-native" before the next budget review.

Today's stories all came back to trusting something you don't fully control: a platform's fix, a training dataset, a bidding algorithm. Which platform or vendor would hurt your business most if it changed the rules on you tomorrow?