WEBVTT

1
00:00:00.100 --> 00:00:13.240
[on-hold music] You're actually going to be at the trial. I am. I'm a 53-year-old former executive who is now going to be sitting in a pew pretending to be a journalist nine to five, taking notes.

2
00:00:13.400 --> 00:00:20.620
You're out there, you're doing... You're, you're working the beat. I wasn't trained by Jason in finding out the truth. That's true. Do you have a, do you have a notebook?

3
00:00:20.720 --> 00:00:27.590
Do you have, like, one of the, the reporter's notebooks? I even have this... I'll show the, the audience can see one of these legal envelope things. I went to Staples- Oh, yeah...

4
00:00:27.640 --> 00:00:45.500
and put down my credit card and bought some real h- heavy supplies for this whole effort. Yeah. It'll be good. [on-hold music] Welcome to the Rebooting Show. I am Brian Morrissey.

5
00:00:45.600 --> 00:00:56.750
This week marks an important moment in the history of digital advertising as the US Department of Justice presses its case against Google for what the DOJ believes is a monopoly in ad tech.

6
00:00:57.220 --> 00:01:11.580
The seeds of this case were planted all the way back in 2007 when Google bought DoubleClick, which was a critical piece of internet advertising infrastructure that was widely used by both advertisers and publishers in running and targeting and measuring their ad campaigns.

7
00:01:12.120 --> 00:01:31.980
Now, with DoubleClick in the fold and programmatic advertising really taking off, Google methodically grew to dominate all phases of digital advertising by piecing together a full stack solution for ad tech, supplying the tools that were used by both the buy and the sell sides, as well as the exchange used for transacting on ads.

8
00:01:32.440 --> 00:01:41.080
Google was also the biggest source of demand for the ad exchange. The go-to comparison of this situation is if Goldman Sachs owned the New York Stock Exchange.

9
00:01:41.550 --> 00:01:59.260
The government's complaint unearths plenty of embarrassing details for Google, as these things inevitably do, and basically that it was using its position, its dominant position, to its own advantage, which is kinda sensible, and in ways that ended up disadvantaging both publishers and advertisers, not to mention its competitors.

10
00:01:59.760 --> 00:02:20.290
Now, the justice system works slowly, and this case is probably just the beginning of a process, but it could lead to, and I think it will lead to, a reshaped ad market, particularly if Google is constrained in its practices or even forced to relinquish what DoubleClick became, and for Google to no longer have its weird role on all sides.

11
00:02:20.290 --> 00:02:25.540
So on this week's show, I spoke to Ari Paparo about this case. Ari is an ad tech veteran.

12
00:02:25.580 --> 00:02:34.560
He worked at DoubleClick, actually, at the time of, uh, its sale to Google, and he went on to start his own ad tech company, Beeswax, that was bought by Comcast FreeWheel.

13
00:02:35.080 --> 00:02:43.560
Ari's now the founder and proprietor of Architecture, a vertical media company focused on ad tech with podcasts and a newsletter, and I think they'll be doing events too.

14
00:02:44.040 --> 00:02:55.800
We take a walk down memory lane to 2007 and how the Google DoubleClick acquisition was so pivotal, and what impact Google has had on the evolution of the digital ad market.

15
00:02:55.840 --> 00:02:59.060
It's always hard to do counterfactuals, but they're also kind of fun.

16
00:02:59.480 --> 00:03:08.360
And we also talk about why one of the defining things about this case, at least I believe, is that it will, it will unearth unflattering aspects of Google's business practices.

17
00:03:08.640 --> 00:03:23.800
The leverage on these things is always in the discovery and going through the emails and messages that are inside companies, because plenty of things are gonna come out and are highlighted in this complaint that does not exactly paint Google in the best of lights.

18
00:03:24.320 --> 00:03:35.520
So here's my conversation with Ari. [on-hold music] All right, Ari, let's get into it. Okay. Friday afternoon.

19
00:03:35.620 --> 00:03:51.020
I can think of no better way heading into a, a weekend, it's still technically summer, than to talk about dynamic allocation, header bidding, and various other ad tech stuff, because coming up on...

20
00:03:51.060 --> 00:04:01.020
It starts Monday, right? Monday at 9:00 AM. Monday at 9:00 AM, the, the seminal antitrust case is going to- the DOJ is gonna start... kick off its case against Google.

21
00:04:01.440 --> 00:04:13.130
This, this one is around ad tech, and it is, to me, it's fa- it's gonna be fascinating for a few reasons, but one of the big ones is this is very intertwined in the history of internet advertising.

22
00:04:13.150 --> 00:04:22.460
It used to be called internet advertising, then it became digital advertising as they got and became bigger. Let's start by going back, 'cause you, you had a front row at this because you were at DoubleClick.

23
00:04:22.870 --> 00:04:33.800
Was that your first job in ad tech? It was, more or less. I was a customer of DoubleClick during the dot-com bubble, but my first job in ad tech was in 2004 when I started at DoubleClick. Okay.

24
00:04:33.840 --> 00:04:43.740
So at that time, internet advertising, after the dot-com craziness stuff, right? It, it, it was just a... It was a very nascent industry, and there was this separate industry that was thriving. I...

25
00:04:43.880 --> 00:04:54.560
To me, it was, like, almost separate, like search, right? I mean, search almost operated in its own world, and the people who were in what was, you know, slinging banner ads, [chuckles] they didn't know.

26
00:04:54.570 --> 00:05:04.580
'Cause originally, the search result pages were, were, like, the least valuable pages on the internet, which is kind of funny to think back then, because they were using just dumb, like, display ads on them.

27
00:05:04.660 --> 00:05:15.480
But, like, Google built this massive business, it went public, and people were like, "Oh my God, this business is that big? I thought this was just, like, kind of the bad Yellow Pages."

28
00:05:15.980 --> 00:05:27.300
But, like, the display ad business was operating kind of separately, and that business was... I don't... Explain, h- how did the business operate before what is now known as programmatic?

29
00:05:27.360 --> 00:05:37.240
I mean, it was very basic business. Sure. It was an offshoot of journalism and the traditional media companies. Many of the traditional media companies made the jump into digital in the '90s.

30
00:05:37.600 --> 00:05:41.320
By the mid-2000s, they were... had active sales groups that were selling ads.

31
00:05:41.750 --> 00:05:51.680
They were selling ads usually directly, you know, over steaks and martinis, and the ads were banner ads or takeovers, but they weren't really... There wasn't really much video 'cause the bandwidth wasn't there.

32
00:05:51.720 --> 00:06:00.100
There's faxes, right? Wasn't fax- Yeah... faxing was, was- A little fax. There was some faxing going on, some, some bios, some alcohol, and everyone was happy.

33
00:06:00.460 --> 00:06:03.720
And there were a couple of big players, but Google wasn't one of them.

34
00:06:03.860 --> 00:06:16.020
So at the time, the market was dominated by names you may wanna forget, like AOL, Yahoo, and MSN, and they were so dominant that people called them the Big Three- Yeah... uh, if you can imagine that.

35
00:06:16.060 --> 00:06:25.890
So- That was the big concentration. It was, like, every IAB report, I would have to, like, be like, "And the Big Three are now... have gone up from 73% to 76%." [chuckles] Yeah.

36
00:06:26.020 --> 00:06:31.400
Find a millennial and ask him to guess who the Big Three used to be in advertising, and that would be a fun, a fun podcast episode.

37
00:06:31.880 --> 00:06:47.180
So the Big Three were out there, and what was happening from the very dawn of the digital advertising market was that because it's soFragmented and distributed, and there's such a long tail of audience coming to your site there were always ad networks.

38
00:06:47.360 --> 00:07:04.320
Ad networks fulfilled the role of aggregating supply and enabling buyers to buy them in one spot and that was because even on a major website like the New York Times, you would have someone visit you from Bulgaria, and your sales force had no way of selling an ad to a Bulgarian advertiser, so you had to outsource that.

39
00:07:04.680 --> 00:07:12.180
So ad networks were very important from the start, much more important than they were in other mediums like TV, where there are ad networks, but they just aren't as important.

40
00:07:12.759 --> 00:07:37.520
And the big three operated ad networks, so you can monetize your site with the help of Yahoo or AOL or people like that, and these ad networks also operated maybe with a little more tech than journalists do or journalists, journalistic websites did, but the tech was still a, a outgrowth of what happened over the steaks and martinis and the direct relationships between buyers and sellers.

41
00:07:37.860 --> 00:07:46.040
It was just a matter of how that executed across thousands of websites- Yeah... where there was maybe a little bit of tech involved. Yeah, there was like, there was like advertising.com.

42
00:07:46.100 --> 00:07:50.300
They had like a boiler, boiler room in Baltimore of people optimizing banner ads, right?

43
00:07:50.840 --> 00:08:02.260
Yeah, if you ever wonder why so many advertising people are from Baltimore, it's exclusively because of one company, advertising.com and their boiler room. Okay. And then Right Media. I remember Mike Walrath, I have...

44
00:08:02.300 --> 00:08:08.640
Maybe it's like 2005. I, I think I'd just joined Adweek, and he walked me through what they were doing.

45
00:08:08.840 --> 00:08:22.240
And at, at the time, I think it was initially positioned as ad networks would be bidding for, for inventory, but I think of Right Media as the, the pioneer in the space. Just like Google didn't invent search advertising.

46
00:08:22.470 --> 00:08:24.170
That was done by Bill Gross and [audio cuts out].

47
00:08:24.260 --> 00:08:35.540
Yeah, I think the Right Media story is very interesting and isn't really told that much, where Right Media was a s- a technical startup here in New York, and what they did was they offered an ad server for ad networks.

48
00:08:35.600 --> 00:08:47.950
So they let people use their tech to get a bunch of orders from advertisers and place them on many, many publishers' websites, and that was a really interesting niche because the other ad servers, like DoubleClick, weren't really meant for that exact use case.

49
00:08:48.500 --> 00:09:01.520
And then the CTO of Right Media, a guy many of you know, named Brian O'Kelley, he went off into a room with a bunch of pizzas and some Russian programmers, and three weeks later came, came back with the world's first ad exchange.

50
00:09:02.200 --> 00:09:12.960
And what that did was allow Right Media customers in real time to swap impressions between each other, meaning that if a...

51
00:09:13.000 --> 00:09:30.530
Let's say you have ad network one that has a lot of advertisement for people who are interested in cars, and you have ad network two that doesn't have any ads for cars but happens to have their ad tag on a, let's call it a, a Vogue article about cars, 'cause it's a women's ad network.

52
00:09:30.960 --> 00:09:51.700
It would let ad network one with the car ads win the auction and show the ad on ad network two's placement, thereby ad network two would make money as a commission on an ad they couldn't otherwise sell, ad network one would be able to sell more car ads to people who are interested in cars, and everyone would be better off.

53
00:09:51.740 --> 00:10:00.460
It created liquidy- liquidity in really the entire advertising market for the first time in history of advertising. Yeah. And, and it solved an essential problem.

54
00:10:00.490 --> 00:10:08.940
I mean, even back then there was something like a million ad-supported websites, and it's impossible to do that in a direct manner, no matter what anyone says. And so...

55
00:10:08.960 --> 00:10:18.090
But, but this sort of intersects with DoubleClick, 'cause DoubleClick was a... I, I remember my first job as a, a Silicon Alley reporter, and DoubleClick was one of the s- Wait, hold on. I... Hold on.

56
00:10:18.140 --> 00:10:28.280
We have to stop there. Your first job was at Silicon Alley Reporter working for Jason? Yeah. I didn't know that. Yeah, I'm his protege. Um- Oh my God. I may have some copies in my, in my filing cabinet. I need...

57
00:10:28.320 --> 00:10:35.860
Next time I see you- Yeah, and I wrote some good-... I need some signatures. I wrote some good cosmo.com stories, but I covered DoubleClick, and I was like, "I have no idea what the hell [laughs] this company is."

58
00:10:35.900 --> 00:10:44.440
But then I was s- I was like, they're just do these boring plumbing. I was like, there's a, Cosmo is so much more interesting pseudo and all these different, like over the...

59
00:10:44.460 --> 00:10:59.969
There was some crazy ideas, but I was like, this is a dull company. They, like, charge people, like, a couple cents per CPM to, to serve banner ads, and then they, they just did some back-end sort of plumbing work.

60
00:11:00.260 --> 00:11:08.180
That's the way I thought about it. Ar- arguably, DoubleClick was the only New York company in the s- to survive out of the dotcom era. Yeah, which is, which is, I think, telling.

61
00:11:08.240 --> 00:11:18.990
That's, [laughs] that was sort of my point. It's like the boring companies usually are the, the ones. But, like, ad serving itself was, I don't know if it was a commodity market, but, you know, it, it pretty much was.

62
00:11:19.040 --> 00:11:28.640
I mean, there was other players that were in the, in the ad serving market, and they, you know, you, you need it from both ends. You need it from the, the advertiser end and the publisher, correct?

63
00:11:28.670 --> 00:11:29.640
And they had other tools.

64
00:11:29.940 --> 00:11:44.520
DoubleClick had a whole bunch of crap under its umbrella, but the two premier products were DFP, which was called Dart for Publishers, and DFA, which was Dart for Advertisers, and they were both software products, dull software products that were very valuable though.

65
00:11:44.530 --> 00:11:50.120
They were dull but valuable, so it's hard to say whether they were a commodity or not, and they had very high market share.

66
00:11:50.180 --> 00:12:03.800
In particular, DFP, we know from some filings that we'll talk about later, had approximately 60% of the publisher market share, so 60% of publishers used it, and the ones who didn't were largely the big three, AOL, Microsoft, and, and Yahoo.

67
00:12:03.840 --> 00:12:06.460
Everyone else besides those used DoubleClick to serve its ads.

68
00:12:07.040 --> 00:12:15.060
Yeah, and there was a few other competitors to it, but they, they were the dominant one in the market, and, uh, they also, by the way, we should do, like, a side...

69
00:12:15.080 --> 00:12:26.860
We should do an internet history podcast, because the Abacus, [laughs] when they, when they went to buy Abacus, I think that was a sign of what was to come with all the privacy stuff that, again, we, we- Side quest...

70
00:12:26.900 --> 00:12:35.990
we can't go, [laughs] we can't go down, can't go down the Abacus rabbit hole. But DoubleClick, I remember, came out with a, an ad exchange really right...

71
00:12:35.990 --> 00:12:45.140
'Cause I, I m- basically what was happening then w- I think what, w- with Right Media, Yahoo invest- invested in Right Media, and it was like, "Okay, let's see what it is, then we're gonna buy."

72
00:12:45.150 --> 00:12:48.860
It was a strategic investment, and they bought in, in 2007, I believe.

73
00:12:49.480 --> 00:13:01.051
And then Microsoft bought a smuck- smaller one in AdECN, and I think-That was Jeff Green was running that Yeah, that's how Jeff Green got in the business. I mean, my God, this is all the people [laughs]

74
00:13:01.332 --> 00:13:09.612
And then he g- went on to do The Trade Desk and DoubleClick... Explain then the importance of, of the- Yeah... exchange that DoubleClick was.

75
00:13:09.752 --> 00:13:21.572
It w- wasn't like it was this massive thing, but because of their position as the default ad server, quote-unquote, they had an advantage- Exactly... to build an ad exchange.

76
00:13:21.872 --> 00:13:34.232
So yeah, so one thing to note is a common misconception that DoubleClick sold ads, and at the time it did not. In the '90s it did, but it spun off that division. So by the mid to late 2000s, it was not selling ads.

77
00:13:34.262 --> 00:13:40.392
It was only selling software that allowed other people to sell ads, and as a result, you s- used the word commodity earlier.

78
00:13:40.452 --> 00:13:45.532
It wasn't commodity, but it w- the pricing was going down, and the market maybe had been tapped out.

79
00:13:46.032 --> 00:13:55.412
And so this idea of an ad exchange comes in where you don't really have to sell ads, but you could still make a commission if you enabled the sale of the ads, and that was incredibly attractive to the company.

80
00:13:55.452 --> 00:14:00.552
It was also a big threat because if somebody else did it really well, they may- Yeah... not need the ad server anymore.

81
00:14:00.672 --> 00:14:10.852
So under the leadership of two guys, Michael Rubenstein and Scott Spencer, both of whom are kicking around the market in different ways, they launched the first DoubleClick ad exchange.

82
00:14:10.872 --> 00:14:38.192
It was called the DoubleClick Ad Marketplace, I believe, and the idea was that because they had all these publishers using their ad server, they could run an exchange that would only bid when the bid was higher in price than what otherwise would have served, which would give publishers effectively, the way they thought about it, free money, because otherwise they would be losing that opportunity to deliver an ad that is higher priced.

83
00:14:38.212 --> 00:14:51.352
And you could only do that at the time if you were built into the ad server, and that was called dynamic allocation, and this is the first, you know, fence post along a very long path that we'll talk about when we talk about some legal problems.

84
00:14:51.452 --> 00:15:01.132
Right. Okay, so at this time, Google was... Look, Search was a home run. I remember people would say to me, "They're a one-trick pony." It was a hell of a trick. Hell of a trick. Hell of a trick.

85
00:15:01.752 --> 00:15:11.062
That's greatest history of business. The greatest money-making machine [laughs] in history. But they, they expanded beyond Search into, with, with AdSense, and then Google Image Ads, I think they called them. Yes.

86
00:15:11.072 --> 00:15:24.072
Which is... They, they hated... I remember the, the actual very first story I ever wrote was Eric Schmidt talking at an IAB conference in which he was saying, "Stop scaring users with your crappy flash ads."

87
00:15:24.132 --> 00:15:31.221
[laughs] And then I got the guy from the Flash Ad Association sending me an angry email, but I'm like, "I didn't say it." [laughs] And by the way, nobody liked flash ads.

88
00:15:31.292 --> 00:15:39.792
I know that you were the VP of Rich Media, but nobody liked them. At the time, you know, Google was... Yeah, they did Google Image Ads. They were like static banner ads.

89
00:15:39.832 --> 00:15:51.392
They weren't really a big player in this part of the internet advertising market at all. No. They wanted to be. So they realized that... Philosophically, let's give Google some, some credit, some, some...

90
00:15:51.432 --> 00:15:53.562
Let's em- empathize with the way they thought.

91
00:15:53.932 --> 00:16:04.772
They thought ads were a form of content, and we can make ads more useful, and that was their way they built their Google Search ads 'cause they felt that the Search ads were useful. In many cases, they were.

92
00:16:04.812 --> 00:16:11.512
They gave you a better answer than the organic search when you were looking for something, and they wanted to apply that to the wider world of ads.

93
00:16:11.992 --> 00:16:22.992
So like you said, they created AdSense, which was the publisher product that allowed you to make money by putting text ads on your site, and then they wanted image ads to happen, and image ads were really not happening for Google.

94
00:16:23.432 --> 00:16:27.712
They were very strict about not allowing third-party cookies on any of their products.

95
00:16:28.192 --> 00:16:40.372
So third-party cookies being the lingua franca of digital advertising meant that really any major advertiser was hesitant to put their image ads through Google's system. They wouldn't get the reporting they're used to.

96
00:16:40.792 --> 00:16:49.472
They wouldn't get anything they were used to. In addition, they just kinda didn't have the vibe. I don't know what else to say. They didn't have the rizz when it... that it takes to do image ads.

97
00:16:49.552 --> 00:17:00.492
They [laughs] just thought of it in a very technical way, and that's not how image and video ads work, you know? Uh, uh, Madison Avenue, it's sexy. People want exciting ads that do exciting things. And it...

98
00:17:00.712 --> 00:17:11.032
Google just wasn't there, and they were pushing a big rock uphill, and they were competing with the big three, and the big three did have the rizz. You know, Yahoo really knew how to sell ads.

99
00:17:11.452 --> 00:17:22.932
You had leaders like Wendell Omar, who's, you know, legendary in the industry. You had a aggressive sales force. They would buy you jeans if you were an agency player, pl- planner. They would run really rich media ads.

100
00:17:23.332 --> 00:17:34.072
So you have three entrenched competitors who are really good at display ads, and you have Google, who's kind of like the nerdy kid in the corner trying to invent something from scratch that is just absolutely not working.

101
00:17:34.472 --> 00:17:42.462
Yeah, I mean, they were saying... I mean, again, just to go back to what, what Eric Schmidt was saying, it was like, "You, you... The way you guys are doing things is completely anti-user," and I think it's...

102
00:17:43.652 --> 00:17:52.052
Going back, I think that ethos was really, really within Google at, at those... I mean, it was, it was soon after the, the IPO.

103
00:17:52.132 --> 00:17:57.652
I mean, I, I remember, I mean, I, I was still talking with people there who had their, like, first name @google.com, you know?

104
00:17:57.662 --> 00:18:05.702
[laughs] Like, I mean, it, it was a different company back then, and, and it was, it was a different vision of, of what... how ads should, should work.

105
00:18:06.332 --> 00:18:12.692
So DoubleClick got bought by, I think it was Hellman, Hellman Friedman. Was it the- Yeah. It was taken private off the public markets.

106
00:18:13.072 --> 00:18:19.832
It was very bloated, uh, had all these divisions that didn't make money, and so it really needed to be taken private to clean up the act.

107
00:18:20.332 --> 00:18:34.472
Okay, and then with this ad exchange and with what was going on, I, I guess competitively, from what I remember, in the market, it was a good time to, to, you know, to, to find a bigger partner, I guess it would. Yeah.

108
00:18:34.752 --> 00:19:00.212
Uh, I, uh, reta- recently detailed this on a, aBlog post, but the company effectively started a process to sell itself after having cleaned up the mess and made some really good spin-outs that helped pay for the debt of the deal, and they realized that their DFP, their monopoly position in ad serving, would be incredibly valuable to one of four companies, and you could guess who they are.

109
00:19:00.292 --> 00:19:07.352
They're the big three plus Google. And they- It's YMAG, right? YMAG. So Yahoo, AOL, Microsoft, Google spells YMAG.

110
00:19:07.412 --> 00:19:21.952
And so the leadership of DoubleClick with the cooperation of Hellmann and Friedman created this thing called the YMAG model, and what that was was a way of calculating how much money you would make if you owned DoubleClick, if you were one of those four acronyms.

111
00:19:22.432 --> 00:19:23.892
And the model basically worked like this.

112
00:19:24.372 --> 00:19:43.552
You have X billions of impressions running through DoubleClick on a daily basis, Y price, multiply those together, then you say, what percentage of it could you win if you had the ability to bid on it, maybe twenty, thirty percent, and then what would you make on each of those impressions if you bid on it?

113
00:19:43.632 --> 00:19:47.052
And therefore, how much money would you make on a monthly basis?

114
00:19:47.072 --> 00:19:59.132
And the answer was, 'cause I have the actual deck and I published it on my website, the answer was two point six billion dollars a year was the conservative estimate for how much money you could make if you bought DoubleClick and you were one of those companies.

115
00:19:59.712 --> 00:20:08.652
Which was, you know, a lot of money at the time. Yeah, and it's still a lot of money [chuckles] for some of us. But it, it's important to... I mean, that was basically with...

116
00:20:08.692 --> 00:20:18.932
DoubleClick was not as valuable as an independent, neutral... 'Cause i- if you think about an exchange, I would think, I mean, you know, the, the analogy is always made, you know, the, the...

117
00:20:19.012 --> 00:20:22.152
It's like Goldman Sachs owning the New York Stock Exchange.

118
00:20:22.652 --> 00:20:34.492
You wanna sit, you want a neutral party, 'cause I think this judge is gonna be looking at this, and other markets operate in different ways, and sometimes people can get, get so wrapped up into how it's done in their own market.

119
00:20:34.512 --> 00:20:45.112
But it doesn't seem, and it never seemed like it made sense for a company to operate on multiple sides of the market and to also control the exchange for the applications. Yeah.

120
00:20:45.392 --> 00:20:56.072
I, I think if you think of it in terms of supply and demand, DoubleClick had this supply position that was very strong, and the exchange sort of made it liquid. So that was pretty exciting.

121
00:20:56.092 --> 00:21:03.301
DoubleClick did not have a real demand position. Even though it had the advertiser products, they weren't, they weren't bidding in it. It was...

122
00:21:03.322 --> 00:21:08.592
So there wasn't really a buy side at the time that would fulfill that supply.

123
00:21:08.652 --> 00:21:18.952
And the first thing Google or anyone else would have done would be to take the demand they already had and just point it at this exchange, and that's what Google ended up doing.

124
00:21:19.012 --> 00:21:43.252
They took the Google Ads demand, that used to be called AdWords, now it's called Google Ads, and they exclusively pointed that demand at the exchange, thereby creating enormous amount of value for publishers, which is good for publishers, and taking margin on both sides, being able to take margin from advertisers who now were buying more ads and take margin for publishers by transacting on the sell side.

125
00:21:43.792 --> 00:21:57.612
And even if they had done that on a fair basis without any manipulation of the auction, it would've been a pretty big grab for market share in this emerging, you know, exchange marketplace. Yeah. So let me ask you this.

126
00:21:57.692 --> 00:22:09.042
Let's just actually go... Is all of this basically, [chuckles] and a lot of antitrust seems to be this, is, is this just going backwards to like 2007 and doing a job that maybe should've...

127
00:22:09.042 --> 00:22:19.452
'Cause nobody had a crystal ball then in that, I think over the years, and we'll talk then about like how Google used this, this position, but it responded to the incentives like any other company I think would.

128
00:22:19.492 --> 00:22:25.432
It's not like it's some sort of moral, morality play. I- I mean, it just seems like it's going backwards in time.

129
00:22:25.472 --> 00:22:36.872
We're getting, going back, what is it now, 17 years ago, to try to write something that, you know, I don't know, maybe people... Nobody was paying attention, [chuckles] that's for sure, at that time. Well, yeah.

130
00:22:36.912 --> 00:22:41.452
So in tw- 2008, I guess it was, the acquisition was approved.

131
00:22:41.492 --> 00:22:50.072
The government did not interfere with the acquisition of DoubleClick by Google, with the exception of some search considerations which turned out to be totally meaningless.

132
00:22:50.132 --> 00:23:01.432
And the argument that they should have known is, is interesting, but you have to realize the exchange had virtually no presence at all at the time of the acquisition.

133
00:23:01.472 --> 00:23:10.892
The DoubleClick Exchange, according to the document that I published- It was like a noun. It wasn't really a thing, was it? Yeah. The, the revenue was $800,000 a year at the time of the- Yeah... acquisition.

134
00:23:11.252 --> 00:23:20.552
So you would have to do like, you know, a future crime sort of [laughs] approach, I'd say. [laughs] We need... Lina Khan needs some precogs, like in the basics. Yeah, some precogs.

135
00:23:20.772 --> 00:23:33.432
Like, there will be a giant market for display ad liquidity, and the winner of that market will be Google because they will crush their competitors who don't yet exist, and, you know, that's a little bit hard to, hard to manage.

136
00:23:33.462 --> 00:23:39.872
Yeah, that's true. Right. But it does go back to that, and then Google basically said they bought DoubleClick and then Microsoft made...

137
00:23:39.892 --> 00:23:49.851
Well, depending on, on, on which side you're on for that, made the, either the worst M&A deal in history [laughs] or- The worst... if you were, if you were the seller, it was, it was a great deal.

138
00:23:49.912 --> 00:23:58.712
Six point one billion dollars just down the toilet in one fell swoop. They bought Aquantive, which was a distant number second with an Atlas Ad server.

139
00:23:58.992 --> 00:24:10.592
It's, by the way, I love it as like Aquantive was like basically the best agency model because they, they, they spun out the technology. [laughs] It's like the agency was just kind of written off by Microsoft.

140
00:24:11.032 --> 00:24:22.132
But yeah, Google then did what anyone would do, right? It bought, it bought different pieces, it built different pieces, and bought Invite Media for the demand side. It bought Admeld.

141
00:24:22.212 --> 00:24:29.972
I think that was mostly about relationships, right? But it built this, this position within, within the ad ecosystem.

142
00:24:30.332 --> 00:24:44.172
But embedded within this is, is this idea of, of dynamic allocation, and basically that just gives either a fair or an unfair or just an advantage to Google, correct? Yeah.

143
00:24:44.232 --> 00:24:55.508
Effectively,Dynamic allocation let Google win auctions that competitive DSPs and ad exchanges couldn't possibly win because they never even got a chance to bid.

144
00:24:56.028 --> 00:25:09.468
The-- Think of the ad server, the, the ad logic that a publisher is executing as being in sort of layers or strata, and mostly by price, so highest price first, then it goes all the way down.

145
00:25:09.988 --> 00:25:29.768
And so there was some level at which all of the other players, the remnant players, like at the time, people like Rubicon or advertising.com or BlueLithium, all these, all these folks with different business models who were ad networks or ad exchanges, they wouldn't even be called if the price was below, say, three dollars.

146
00:25:30.248 --> 00:25:42.948
And when they were called, they would only be called one at a time, and the first one who won would win, and the ones below them wouldn't be called. So it was a very serial process. I mean that, like with an S, serial.

147
00:25:43.348 --> 00:25:48.348
So one at a time in a row. Meanwhile, Google had a parallel process that went all the way to the top.

148
00:25:48.768 --> 00:25:59.548
So if Google had a twenty dollar ad they wanted to buy and the publisher had a nineteen dollar and ninety-nine cent ad, Google would win, and no one else even knew that ad existed. It was a huge advantage.

149
00:25:59.628 --> 00:26:05.808
It was called dynamic allocation, and then later it was called enhanced dynamic allocation. I see. [laughs] And they went through... Yeah.

150
00:26:05.868 --> 00:26:27.088
They went through some real efforts, and this will all come out in trial, they went through some real efforts to keep that advantage at the, at the adverse interest of their customers, the publishers, who wanted everyone to be able to bid as freely as possible because basic auction theory would say the more liquidity you have, the more optimal the outcomes will be.

151
00:26:27.148 --> 00:26:32.068
Right. Yeah. So what are you, what are you... You're actually going to be at the trial. I am.

152
00:26:32.408 --> 00:26:40.898
I'm a fifty-three-year-old former executive who is now going to be sitting in a pew pretending to be a journalist nine to five, taking notes. You're a journalist. You're a journalist. You're not pretending.

153
00:26:40.898 --> 00:26:48.848
You're, you're, you're out there, you're doing, you're, you're working the beat. I wasn't trained by Jason in finding out the truth. That's true. Do you have a, do you have a notebook?

154
00:26:48.968 --> 00:26:54.577
Do you have, like, one of the, the reporter's notebooks? I even have this... I'll show you. The, the audience can see one of these legal envelope things. Oh, yeah. That's good.

155
00:26:54.577 --> 00:27:03.988
I went to Staples and put down my credit card and bought some real heavy supplies for this whole effort. Yeah. It'll be good. What, what do you- What if I plugged my w- a newsletter? Can I plug my newsletter?

156
00:27:04.068 --> 00:27:12.448
I mean, people should s- Of course you can. And I, I, I'm gonna plug it in the beginning. Okay, okay. Sorry please. But if everyone doesn't, I will, I will give it... This is a, a genuine, a genuine endorsement.

157
00:27:12.578 --> 00:27:21.748
"Marketecture" is definitely a must read because I think- And if you're obsessed with the trial, I have a separate newsletter called "The Monopoly Report" that- Yeah... just covers the trial.

158
00:27:21.808 --> 00:27:34.788
But I do think that, I think one of the more fascinating parts, and we've talked about this, of, of the evolving media landscape are people like yourself who are completely steeped in it, in ways people like me, frankly, are not.

159
00:27:34.908 --> 00:27:44.428
So what are the big... What, what are you th- expecting? This is gonna... Look, Google did not want this, this, this thing to go to trial at all. No. It, it's gonna be embarrassing. It is.

160
00:27:44.708 --> 00:27:53.908
Um, but what are you expecting... What are gonna be... What is this going to hinge on? What are gonna be the big issues that come out? Because I think the header bidding one we didn't get to.

161
00:27:53.948 --> 00:27:59.128
The, the header bidding thing is gonna be, is gonna be massive. But what else? Yeah. So this is...

162
00:27:59.388 --> 00:28:03.728
If you read the allegations from the Department of Justice, the, they have a long document, it's a hundred and thirty pages.

163
00:28:03.748 --> 00:28:18.508
It's, uh, quite a good read, and it goes through step by step from approximately two thousand and eight to approximately, like, two thousand and, I don't know, twenty, step by step, Google manipulating the auction to their own benefit.

164
00:28:18.808 --> 00:28:27.508
Let's... Allegations. Put in the word alleged every time I say anything. So [chuckles] allegedly manipulating the auction to their own benefit and hurting...

165
00:28:27.568 --> 00:28:32.487
The theory of harm is that they hurt both advertisers who had to pay more and publishers who made less.

166
00:28:33.068 --> 00:28:47.328
And it's going to feel, I think, to observers, many of whom are-- work at big publications, journalists, broadcasters, it's gonna feel a little bit like a betrayal because Google has been saying very vocally since the two...

167
00:28:47.448 --> 00:28:52.288
late two thousands, that they're the friends of journalists and publishers. They're the only ones we could count on.

168
00:28:52.368 --> 00:29:04.518
While Meta is very flighty and doesn't care about news and, you know, TikTok, et cetera, works with the Chinese government, you know, just fill in your narrative. Google's been by your side, your trusty ad server.

169
00:29:04.528 --> 00:29:11.548
The biggest check you make every month comes from Google, so we're your friend. And what they're going to see is that that was not always the case.

170
00:29:11.588 --> 00:29:17.788
There are emails saying things like, "This isn't to the benefit of our clients." Like, "We shouldn't be doing this." Those...

171
00:29:18.028 --> 00:29:30.468
There's some, there are some pretty damning conversations that are going to happen that show that Google was not always working to the benefit of the people it said it was working to. And then there's some wild cards.

172
00:29:30.528 --> 00:29:44.728
So first of all, you have a couple people testifying, Neal Mohan, who's currently the CEO of YouTube, who could be interesting testimony 'cause he ran, at the time, before he was the CEO of YouTube, he ran all of the adver- He ran this whole thing, the whole ad tech stack.

173
00:29:44.738 --> 00:29:53.018
Mm-hmm. And then another wild card is the relationship between Google and Meta, which is called Jedi Blue, is a crazy name, where- I know...

174
00:29:53.048 --> 00:30:06.548
wherein it's alleged that Facebook at the time dropped its plans to compete with Google to get access to publisher inventory in exchange for a sweetheart deal that gave them access to everyone's inventory. Oh my God.

175
00:30:06.728 --> 00:30:10.188
That's just... I mean, negotiate that on Signal. Don't, don't do that in email.

176
00:30:10.208 --> 00:30:19.488
[laughs] There's, there's an email from a Facebook executive that says, it quote, "It, it's very clear that Google wants to kill header, header bidding," unquote. Yeah.

177
00:30:19.568 --> 00:30:29.588
So we-- This is a good time to go back, 'cause I, I remember with the header bidding that they... I mean, this is a workaround for, for publishers. Explain to- Yeah... to the, the laypeople header bidding.

178
00:30:29.728 --> 00:30:42.712
Well, the one rule of the internet is that if you have a problem, you need to solve it with JavaScript.And, and header bidding is- JavaScript also, JavaScript gives, but JavaScript also takes away too, right?

179
00:30:42.792 --> 00:30:53.172
It's, it's- Right. So what was the problem? The problem was what I was saying earlier about the r- the stratas in the ad server. The problem was that if you were a outside source of demand, like...

180
00:30:53.352 --> 00:31:04.892
And by demand, I mean, like, money, you, you're Rubicon Project, Magnite is now, is what it's called, you wanted to bid into the, into the publisher's ad stack at every level of the strata, and Google wouldn't let you.

181
00:31:05.392 --> 00:31:11.772
So what do you do? You run your entire auction before it ever gets to Google, and you do that using JavaScript.

182
00:31:12.292 --> 00:31:29.912
And so this invention called header bidding allowed a, an auction to take place on a website before the ad server ran, figure out who the winning ad is and what they were willing to pay, and then shove that data into the ad server whether it liked it or not, like, kinda like stuffing a turkey.

183
00:31:29.972 --> 00:31:37.091
Like, two dollars, this is the bid, take it. And then the ad server didn't know what was happening, and it would spit out the ad if it was the winning ad.

184
00:31:37.532 --> 00:31:48.312
And this is the way the entire digital advertising world works now- Yeah... is with this massive JavaScript stuffed turkey approach, all because Google wasn't willing to let people compete in a more intelligent way.

185
00:31:48.852 --> 00:31:58.032
And in the trial, what's going to come out is Google's, I would call it, obsession with killing this. Like, they did everything they could. They hated it so much.

186
00:31:58.052 --> 00:32:06.572
There are these emails like, "Oh God, I can't believe people are doing this. This is [chuckles] this is the worst. We have to kill this." [laughs] And, and it wasn't just competitive reasons.

187
00:32:06.692 --> 00:32:12.812
I think there was an aesthetic aspect to it, that they just hated it so much. It went against all their principles of good engineering.

188
00:32:13.272 --> 00:32:25.111
But it is literally the thing that holds the digital advertising world together today. Okay, so who... Like, where's the harm? 'Cause they're gonna have to show the harm, right? Yeah. So I mean, publishers were harmed.

189
00:32:25.152 --> 00:32:38.912
A- agencies got paid off. But w- were advertisers hurt? Did they... They ended up having to pay more? So it appears that advertisers w- on Google Ads were hurt. So the advertisers could come in from one of three sources.

190
00:32:39.242 --> 00:32:50.312
They could come in from an external DSP like The Trade Desk, they could come in from Google's DSP, known as D- DV360, or they could come in from Google Ads, which was kind of the self-serve platform.

191
00:32:50.832 --> 00:33:04.472
And there was a lot of manipulation, but it appears that the main potential victim, or maybe think of it as like the patsy, are the advertisers on Google Ads, where whenever Google wanted to manipulate the auction, they would say, "Eh, let's just charge them a little more.

192
00:33:04.512 --> 00:33:16.102
They won't notice." And in various ways, they would be charged a little more and wouldn't even notice. So it's, I'm really simplifying this. Yeah. Google could probably argue against this in a million ways.

193
00:33:16.112 --> 00:33:20.572
They'll bring out their economists and say I'm wrong. But that's, that's kind of on the advertiser side what happened.

194
00:33:21.012 --> 00:33:31.572
On the publisher side, it's more of a theoretical argument, which is that there were all these cases where Google made the auction not an honest auction.

195
00:33:31.652 --> 00:33:40.252
They took money from one customer and gave it to another customer so they could win more. They changed the rates that they were agreed to so they could win more. It was always about winning more.

196
00:33:40.292 --> 00:33:44.972
They wanted to win more auctions. And so- Well, it's easy... I'm sorry, I keep going back to this.

197
00:33:45.352 --> 00:33:57.052
If you're, if you're holding the auction and you're participating in the auction, is it, is any other market operate this way? I think, I think the Goldman Sachs dark pools is sort of in that category.

198
00:33:57.092 --> 00:34:07.812
Like, if you place an order with Goldman Sachs, and they have a different Goldman Sachs customer who has what you're buying, they, they can manipulate those things, but it, it does seem pretty corrupt. Yeah.

199
00:34:08.452 --> 00:34:15.692
I don't- I mean, maybe we need, like, an SEC for ads. I, I do... I actually think that's actually more of the problem here, is that we were...

200
00:34:15.752 --> 00:34:25.432
The advertising industry jumped into a, frankly of, an auction environment with zero regulations, zero oversight. It's not even a code of conduct.

201
00:34:25.972 --> 00:34:33.732
There was nothing illegal or against even the contracts people signed when people manipulated their auctions because there were no rules.

202
00:34:33.772 --> 00:34:40.772
It was all being made up on the fly by people like me in product management roles who were incentivized to make as much money as possible. Yeah.

203
00:34:40.802 --> 00:34:48.692
Um- Well, I mean, this is also, this could, this could fit into the sort of allergy against any kind of regulation, and I think you're seeing this with the, the privacy stuff.

204
00:34:48.732 --> 00:35:03.392
You know, for, for so long, I, I keep going back to, I was just constantly told by people like the IB that, you know, it's, there's no PII and the direct mail guys are sketchier, and that's, that's our story, and we're sticking to it.

205
00:35:03.852 --> 00:35:06.812
And- You know, I once got heckled by a Swedish person on this subject.

206
00:35:07.212 --> 00:35:17.372
I was giving a little presentation on how to manipulate your auction u- using AppNexus, and this fellow just had enough, and he just stood up in the audience and says, "What you're doing is wrong. Well, you're wrong.

207
00:35:17.732 --> 00:35:27.212
You're lying." Yeah. So that was my little role- That's Europe, baby Yeah, that was my corruptions Privacy is a human right. [chuckles] I, I just spent the la... I just spent two weeks closing those consent windows.

208
00:35:27.712 --> 00:35:35.892
[laughs] But yeah, the upside, the gelato was amazing, so. [laughs] Yeah, just... Yeah, I picture you just smoking cigarettes and closing consent windows- Yeah... as your vacation.

209
00:35:36.372 --> 00:35:44.082
I also speak with an [chuckles] European accent when I'm there. So yeah, okay, so when you think of... I mean, [chuckles] might as well play Matlock on a Friday afternoon.

210
00:35:44.512 --> 00:35:57.872
It seems to me like this entire setup is, is just doesn't, it shouldn't necessarily exist, and, and it might even be... It does, it's not like even Google completely needs it, I don't think. Yeah.

211
00:35:58.372 --> 00:36:09.832
I mean, could this jumpstart, like, a more dynamism in the market? 'Cause I think one of the other, obviously, I think Google's competitors in ad tech were absolutely harmed by this.

212
00:36:09.892 --> 00:36:19.972
I don't know how you could sh- show it. Yeah. Yeah, the competitors were clearly harmed, and there's lots of competitors who want to testify, and the list of, the list of them is pretty long.

213
00:36:19.992 --> 00:36:26.702
And actually, Google has been trying to squash their testimony and not let them testify. So- But who the hell else is gonna testify about, like, ad tech stuff?

214
00:36:26.812 --> 00:36:36.812
They want, you know, these economists who don't know anything. So, um- Is a cool marketer gonna, like, take the stand? [laughs] Like... So the DOJ remedy that they've asked for is very clear.

215
00:36:36.912 --> 00:36:48.744
They want Google to divest the ad server and the exchange.That's full stop. So that's GAM and AdX, they want them to divest, and it would be pretty clean.

216
00:36:48.804 --> 00:36:57.204
Those are products that exist that are-- they're intertwined with other products, but those intertwined natures could be untwined.

217
00:36:57.214 --> 00:37:06.124
They're not nearly as complicated as some of the remedies discussed in the search case, which is probably beyond the scope of this podcast. Yeah. But yeah, they could be spun out. And what would happen?

218
00:37:06.224 --> 00:37:07.204
What would be interesting here?

219
00:37:07.544 --> 00:37:21.144
Well, it would be really interesting if Google Ads, which is the largest supp-- demand source for advertising in the world, in the history of the world, could no longer exclusively deal with Google AdX, the supply side.

220
00:37:21.494 --> 00:37:28.534
What if Google Ads, when they wanted to run a banner ad or a video ad, had to go out there into the marketplace? Think about that. Think of that. It's crazy.

221
00:37:28.864 --> 00:37:34.804
Go into the marketplace and pay proper prices for their inventory they wanna buy for their advertisers. It'd be great, right?

222
00:37:35.564 --> 00:37:47.683
So that would be really interesting, and it would probably be really good for Google because these two divisions, the, the net-- what's called the network business, has been declining, actual revenue declines.

223
00:37:48.084 --> 00:37:57.384
It's the lowest profitable, the lowest profitability of any Google division. It's a massive pain in the ass legislatively and from, you know, privacy and everything else.

224
00:37:57.904 --> 00:38:06.244
And, you know, Google's not in a position where they have a lot of bandwidth to do new things. They feel like-- The company feels like it's sluggish, like they need to focus more.

225
00:38:06.784 --> 00:38:08.244
I think it'd be the best thing that ever happened to them.

226
00:38:08.644 --> 00:38:22.724
Spin it out, put some restrictions on the spun-out company so it can't recreate the monopoly, get that Google Ads demand bidding into every publisher's exchange that, that is clean and works according to their requirements, and see the CPMs rise across the board.

227
00:38:23.344 --> 00:38:29.064
So has this harmed innovation, do you think, in, in digital advertising? I think it has. I think it has.

228
00:38:29.224 --> 00:38:41.644
If you have competitors who literally can't sign up customers because they don't have the Google Ads demand, how could they compete? How could they innovate? Yeah. So I guess this, like, new, a, a new landscape would...

229
00:38:41.724 --> 00:38:47.824
I mean, look, these things are always tied up forever. It's not like, you know, at the end of two weeks [chuckles] we're gonna have a new digital ad system.

230
00:38:48.064 --> 00:38:59.144
I hope so, or else I'm gonna have to sit on those benches for forever. [laughs] Any bombshells that you, you expect in testimony? I d- I can't imagine there'd be any bombshells.

231
00:38:59.344 --> 00:39:08.324
I think that I don't know of any disgruntled former Google employees. I, I guess I'm the closest to a disgruntled former Google employee. Are you? You're disgruntled. But I'm not even disgruntled. I like the company.

232
00:39:08.464 --> 00:39:20.424
I have, I have Gmail open right now. And I think most of the folks on the Google side who are testifying are either current or former employees who will be relatively positive in the way they try to spin things.

233
00:39:20.944 --> 00:39:24.004
I'm not expecting anything that's a breakthrough. Yeah.

234
00:39:24.304 --> 00:39:34.444
And I guess, like, for publishers, I mean, a lot of this, as you said, like, a, there is, like, this sort of factor, like, you know, o-of, you know, Google was always, you know, partners with publishers.

235
00:39:34.684 --> 00:39:41.624
A-and at the same time, it was a bit of a fraught relationship [chuckles] on a lot of different levels because Google is just so powerful.

236
00:39:41.864 --> 00:39:50.784
Like, I can always remember, you know, it would always be, like, a reporter was asking someone who was doing anything in, in the digital ad market, "Well, why won't Google do this and just kill you?" [laughs] Right.

237
00:39:51.864 --> 00:39:54.864
They would always say, "Well, Google's the environment. They're not the, the competition."

238
00:39:55.024 --> 00:40:09.444
And I think that's, you know, part of the, the issue that a lot of this antitrust is, is dealing with, is a lot of these markets are incredibly concentrated, and that is, you know, a fact of life in this market, right?

239
00:40:09.784 --> 00:40:09.944
Yeah.

240
00:40:10.084 --> 00:40:26.304
I think that if you took the counterfactual, and Google had never done any of this, and they had operated, you know, more even-handedly, and, and the market had evolved in sort of a different way, I'm not sure that publishers' lives would be that different.

241
00:40:26.384 --> 00:40:33.324
Maybe it'd be a little more choice of monetization. Maybe there'd be more choice of technology, a little innovation.

242
00:40:33.844 --> 00:40:44.964
But short of someone like Meta going all in on publishers, which they very much have said they don't want to do, I don't see anything radically changing for publishers.

243
00:40:45.004 --> 00:40:57.273
They would still be on the wrong side of declining audience, of more anonymity, of more commoditization, of, you know, consumer interest moving to social. Those are all factors that- Yeah...

244
00:40:57.304 --> 00:41:03.884
you know, you can't pin on Google. Well, they lost to UGC at the end of the day. That is a big part of it, yeah. You know, I mean...

245
00:41:04.464 --> 00:41:13.284
And that's, you know, look, people voted with their attention at the end of the day for better or for worse. And it's not like they didn't have their chance. I mean, News Corp owned MySpace and- That's true.

246
00:41:13.364 --> 00:41:23.424
We could do an episode on that too. We should do it. I already have a name for it. What's the name? The Four Sixty-Eight by Sixty. That's the name of your... That's the new podcast called Four Sixty-Eight by Sixty? Yeah.

247
00:41:23.454 --> 00:41:30.734
[chuckles] It's just, like, me and you and four other old dudes we're listening to? [laughs] Talking about... We could do an abacus episode. Yeah. We got all sorts of ones.

248
00:41:31.044 --> 00:41:41.024
I would prefer, like, a Mad Men-style TV show, but taking place entirely in 1996 in Flatiron District of New York, where everyone is just like, "Have you heard?

249
00:41:41.104 --> 00:41:47.844
I got a new Compaq laptop, and I could get four hundred baud on it." And they're like, "Oh, man, this could be the future. Governments are gonna fall, man.

250
00:41:47.884 --> 00:41:50.484
When I can email my representative, it's gonna change everything."

251
00:41:50.723 --> 00:42:00.704
[laughs] Well, I for one am very excited for this trial to lead to a little bit of a walk down memory lane 'cause we're [chuckles] gonna have to go into the, the guts of the history of this stuff.

252
00:42:00.764 --> 00:42:12.904
But thank you for walking us through this. And again, a big endorsement of Architecture. I'm a, a loyal listener and reader, so thank you, Aaron. Thanks for having me. I'm going into this with my eyes open.

253
00:42:12.964 --> 00:42:27.264
I'm gonna do my best to pretend I'm a journalist. Okay. You'll do fine. All right. Thanks, Aaron. [outro music]