The Data War: Google, Yahoo, and MSN

Search engines are fighting ad networks in a battle for eyeballs. Both need eyeballs to show highly targeted advertising to. This competitive environment puts the search engines in direct competition with advertising networks that don’t have a search engine component in their offerings. When serving ads, more data is better. That’s been the driver of some recent announcements.

MSN announced this week it will sell and provide both banner ads and sponsored links on Facebook. MSN will use adCenter, its online advertising platform, along with other in-house technology and services. To me, that means much of the inventory will be available on the same platform advertisers use all the time to manage search within MSN. Clearly, Microsoft understands to compete, it needs both scale and data.

The search engines are in a particularly good position to evolve into full advertising networks because they:

  • Have large advertiser bases

  • Have advertisers with very specific targeting needs and messages defined by keywords, behaviors, and sometimes demographics
  • Have access to behavioral data that goes beyond site visits (ad networks, with the exception of adware networks, have no information on search behavior, only surfing behavior across their specific networks)
  • Have a majority of users who are likely OK with cookies and toolbars used to identify them so they may see more relevant ads

Google’s MySpace deal and MSN’s Facebook deal are clearly only the beginning of the search engine inventory war. The additional layer to these two deals, which isn’t being talked about much, is the value of data collected during the ad display process.

Both individual (non-personally identifiable individual data) and macro-level data on responsiveness and interest can be gathered while serving ads. Facebook (more so than MySpace) has a ton of information due to its hierarchy of interests, groups, and demographics; geographic info; and the way profiles are networked. This information allows for both contextual and behavioral ad serving. The more information an ad server has about a person, the better targeted the advertising.

No one knows where lines will be drawn with respect to use of profile information. We should all hope the lines are drawn in a manner that allows us to spend our limited budgets most efficiently.

The more data MSN, Yahoo, and Google have, the more control they can give advertisers. Or, if advertisers aren’t sophisticated enough to know how to use the control provided, a smart ad-serving system can learn which ads are most relevant to segments of the audience as it serves them.

Based on Google CEO Eric Schmidt’s comments at Search Engine Strategies earlier this month, Google seems to be very careful in its use of any personal data to target ads (regardless of whether it’s personally identifiable or not). Though this may seem a prudent course based on the concerns online privacy advocates, there may be an opportunity for other engines to leap forward in relevance through an opt-in program of profile-driven ad serving. Without profile data, an ad server is forced to serve ads based exclusively on context. Often, this context is ambiguous, or there isn’t a good commercial advertising fit against it. Similarly, even if there’s an advertiser with a relevant contextual ad, profile data might suggest a better targeted ad for many individuals.

If the search engines don’t end up using profile data directly, we can still expect to see some major changes and enhancements based on what the engines are working on in their labs. MSN has a demographic prediction engine that predicts your age and gender based on either an URL or a search term. This is an extension of the keyword research feature in the adCenter suite, but it illustrates an ad can be targeted with a high level of confidence based on a single behavior. The more behavioral data one adds to a session or cookie profile, the more confident an ad server would be in serving the most relevant ad.

Within search, Yahoo has a research project called Mindset, which allows tuning of an organic SERP (define) based on a user-supplied preference: shopping or research.

Interestingly, this kind of data could be used to tune paid listings’ relevance as well, but that isn’t in the demo. Improved paid listings relevance would increase the likelihood of the link being selected and, of course, would increase yield/monetization.

Even when exercising caution in its exploration of behavioral or profile-based ad serving, Google is clearly putting a foundation in place that will facilitate personalized ads. Personalized search is already live but requires a Google account. If we add historical profile data and the information gleaned from a consumer’s Gmail account to Google’s personalized search, we can imagine a comprehensive profile that could be opted in to by searchers and surfers who prefer targeted advertising in exchange for a bit less privacy.

Search may very well become one of the fundamental elements for targeting advertising over the next several years. By watching the engines’ labs and taking advantage of new targeting options the engines provide both within search and across their networks, search marketers will have a whole new world of media to buy and manage. We’re doing it already, and I’m looking forward to a future where ads are so relevant they rival the quality of editorial content on a page.

The search engines are in a data war and the winners will be the consumers and the marketers.

Want more search information? ClickZ SEM Archives contain all our search columns, organized by topic.

Related reading