On Google-DoubleClick: an interview with Microsoft GC Brad Smith

Obviously, the role of antitrust enforcement watchdog is a new one for Smith and for Microsoft, but such ironies won’t blunt the impact of any meritorious argument they might raise. I interviewed him last week about his perspectives on the deal. Below are excerpts. (I’ve edited my questions to make them sound more articulate than they really were. Also, I wasn’t taping, so Smith’s answers are just captured here as best I could using pen-and-paper notetaking.)

Q. For the time being, no one is asking outright that the deal be blocked. Instead, they’re just urging antitrust regulators to make a “second request.” [A second request for information indicates that the antitrust authority–either the Department of Justice or Federal Trade commission–has decided to initiate a full-bore, analysis that will probably take six to nine months to complete. The second request will come, if it comes, in mid-May.] Are people just being conservative?

A. Mostly, they’re probably being conservative. These questions are very novel. Before anybody tries to come to a conclusion, it would make sense to learn a lot more, in terms of having a data-driven analysis. It may well be, after learning more, we’ll be saying it should be blocked.

Q. I’ve heard people say that there’s no problem here, because Google and DoubleClick aren’t direct competitors. [Google’s biggest business is in the market for paid-search ads, which are the text ads that show up alongside search results. According to eMarketer, Google held 75.6% of the U.S. paid-search market as of February 2007. Google also posts contextual ads on third-party sites, where the ad is targeted to relate to the content running near the ad space. DoubleClick, on the other hand, is the leader in providing Web services that advertisers and publishers use to post and manage display ads, including rich media and video. Compared to Google’s contextual ads, these are typically higher-end ads, running on higher-end sites, touting higher-end brands.]

A. The law is opposed to two companies coming together if it’s going to acquire market power that would enable them to raise prices. So a big part of this question is, are they part of the same market and, if so, will they be in a position to raise prices?

Companies typically defend in a merger analysis by saying either, “No, we’re in two different markets,” or “Yes, we’re in the same market, but we don’t have a significant share of it.”

So the threshold question is: What is the market? To answer that you ask: Are two products substitutable for each other. If you raise the price on product A, will they shift to product B. If so, they’re substitutable, and are in the same market. I think [the argument that display ads, which DoubleClick brokers, and contextual ads, which Google handles, are in different markets] is very unlikely to sway regulators.

The thing that differentiates the two is the way they’re generated. A contextual ad scans the context of a page, and then chooses and serves up an ad related to that context. If it sees that the content is about Ford Motor earnings, it might serve an automobile ad.

With a display ad, they look at it through cookies generated when someone goes to other pages. You may be writing this article about Google, but the reader thirty minutes ago went to a Ford automobile page, so it might serve up an ad for a competing automobile.

Web sites rely on both. Are display ads and contextual ads substitutes for each other? They look the same, and they serve the same purpose.

Q. Do they look the same though? I thought contextual ads were typically much simpler than display ads.

A. Nothing in the technology requires that they be simpler. . . . Are they in the same market? It’s a very objective question. If the price of one goes up, will publishers switch to the other? We think the answer’s yes. If these two companies come together, they’ll have 85% of this market place. They’re the two principal competitors. One has the lion’s share of contextual; the other has the lion’s share of display.

Two other questions then need to be considered. How broadly should the market be defined. The narrowest would be by segment: display is one market, contextual is one market. At the other extreme, there’s [what Google CEO Eric Schmidt was reported as saying at the Web 2.0 Expo conference on April 17, which is that advertising is a trillion dollar business and that a post-acquisition Google would only have about one percent of it.] He wants to include all advertising on the planet. That would be the first time regulators have ever defined a trillion-dollar market, except maybe in the oil industry. Is a Web site publisher going to use newspaper ads [as a substitute for display ads]? I don’t see how that would work. [At the conference, eWeek.com also quoted Schmidt as having said: “This is an emergent business with lots of choices: customers have lots of choices, end users have choices and advertisers have choices. These are people [Microsoft and AT&T] who were involved in acquisition reviews as best I can tell and who lost.” -RP]

The last step, once you define the market and figure out the market shares, is you ask, what are the barriers to entry [by new competitors]? Even if the merged company has 85% of the market, if the barriers to entry are low, the regulators might say, we’re not going to worry [because new entrants to the market will prevent the merged company from raising prices to anticompetitive levels]. This is something the regulators need to study. This is a a market where there are very strong network effects with significant barriers to entry.

Q. What are the network effects here? [“Network effects” were famously a factor in the Justice Department’s monopolization suit against Microsoft in the late 1990s. The argument there was that it would have been extremely difficult for a new competitor to enter the market for operating systems when so many thousands of existing applications had already been written to work only on Windows. Few people would want a new operating system, because few applications would exist to run on it.]

A. Everyone [gauges their success in this business] by measuring revenue per something. Revenues per ad impression. Revenues per search. Revenues per click-through. RP-something. What you see today is, Google’s revenue per search or revenue per ad are way higher than its competitors’. They’re double Yahoo’s, and even more compared to Microsoft’s. The reason? It’s how much personal information their sites collect and use. . . . It’s based on how much information you have on users. That’s attractive to advertisers. They keep aggregating [data on] all the searches you do, all the web sites you visit, and use all that data to serve up an ad. That generates more revenue per search. That makes it harder for other people to break into the market. Whereas in the 1990s, people focused on the applications barrier to entry, this is basically the privacy barrier to entry, or the advertising barrier to entry, or something. Within the next six months, I guarantee you a new term will emerge for how difficult it is to enter when you have somebody with economies of scale from owning so much of the personal information on the Internet.

If the kinds of factors I’ve described are real, what you’ll probably see with consolidation is the profitability of the company that serves the ad will continue to go up at the expense of the the Web site publishers on the one hand and the ad agencies on the other. The company in the middle will own all the personal information and derive all benefits. Talk to ad agencies, publishers, content creators. This is why they’re worried.

Well, readers? Do you find Smith’s arguments convincing?