Google Tiptoes into Behavioral Targeting

Does a tactic by any other name not raise the same stink? Google claims that its new approach to serving ads in recent weeks is not behavioral targeting, or at least not “traditional behavioral targeting.” Analysts say otherwise. What’s going on here?

To understand what is meant, we need to define behavioral targeting. I first wrote about behavioral targeting a little over a year ago. In its “traditional” form, to use Google’s terminology, behavioral targeting takes advantage of cookies to find out a web surfer’s interests and preferences. Then, when that person is visiting a neutral site – say, reading the news – he or she might see ads that are related to their own interests rather than the site’s content.

For example, someone who is interested in football might be reading a story about a hurricane, and see ads for tickets to watch his local football team play. To coin a stereotype, his wife would see ads for the local mall if she viewed the same web page on her computer. His teen-aged kids might see ads for new video games or the latest releases by their favorite musical groups.

For advertisers, this is the stuff of which dreams – and fat bottom lines – are made. Even more than with contextual ads, you can be certain you are reaching people who are genuinely interested in what you have to offer. This kind of advertising could potentially reach its height with the search engines. Imagine being able to display a mortgage ad to someone you know is interested in getting a mortgage even when they’re searching for a new golf bag!

For searchers and web surfers, however, behavioral targeting can send chills down the spine. It’s a matter of privacy, really; in effect, your personal information is being shared with the third party who is displaying the ad. Most of us have seen enough “spills” of personally-identifiable information (notably with AOL last summer) that we’re not entirely comfortable with this.

Additionally, when we notice that we’re seeing ads that have been targeted to our online behavior, it can be a little bit creepy. I have browsed news sites that were not local to my state and seen ads aimed at Fort Lauderdale residents. Granted, this is geo-targeting rather than behavioral targeting, but I wasn’t exactly comfortable with the idea that someone knows where I live and is taking advantage of it online.

{mospagebreak title=How is Google’s Targeting Different?}

Susan Wojcicki, Google’s vice president of product management for advertising, tried to explain that this was not the approach the search engine giant was taking. She tried to emphasize that Google is very careful about its users’ privacy. Instead, Google is focusing on improvements to its ad targeting that last for just one session.

“We believe that task-based information at the time (of a user’s search) is the most relevant information to what they are looking at. We always want to be very careful about what information would or would not be used,” she said.

How might this work? If a user searches for “Italy vacation” in Google he might see ads that show cheap flights to Europe. If he then types in “weather,” some of the ads may be tied to weather conditions in Italy. Or someone who separately types in “vacation” and then searches for “tennis” might see ads for vacations with a focus on tennis.

Ben Murphy, writing in his blog Ben Murphy Design, says that he has already seen this in action. At one point he typed in a search for an accountant in Akron, and then searched for a New York vacation. One of the top sponsored links was for a “Good NY CPA.” It seemed a little strange to him. “On one hand, it’s kind of cool…but as a user it feels, I don’t know…spammy? I told you what I was interested in before, now I’m telling you what I’m interested in now. Why distract me?”

By paying attention to a session or set of searches only during that one session, Google hopes to avoid raising privacy issues. “What we are very careful about is traditional behavioral targeting,” said Wojcicki. “Nothing is stored, nothing is remembered. It all happens within that session.”

Google might also improve the relevance of the ads it serves by keeping track of only one session. Wojcicki used the example of someone searching for “video games.” Is that person really a gamer? Maybe – or maybe she’s a grandmother looking for a birthday or holiday gift for her grandson. In that case, she probably won’t want to see video game ads after she finds the right gift, until she’s shopping for them again.

{mospagebreak title=Other Forms of Google Targeting}

Google is not engaging in traditional behavioral targeting. Let’s take Google’s word for that, and turn to a different service it offers: Personalized Search. When registered Google users are logged into their accounts, this is the default form of search. It deals with organic search results rather than sponsored links, but it works in a rather interesting way.

For those who are not familiar with Personalized Search, the service lets users maintain a history of searches; it looks through this history to understand what keywords mean when entered by particular users. So it would know that a musician searching for “bass” probably wants information on guitars rather than fish, because on all of his previous searches using that word he clicked on guitar-related links. Does this sound like behavioral targeting to you? It should – because it is.

The big difference is that it applies to organic search results, not sponsored links. So the information goes no further than Google; it is not shared with third parties. This is not “traditional behavioral targeting,” in other words.

But wait a minute – don’t the organic search results affect the sponsored links that appear? So doesn’t that mean that the ads will become more relevant? When you combine that with the single session-based behavioral targeting that Google now practices, you’d expect the search engine to be telling all of its advertisers about the improvements, right?

Only it isn’t. Anna Papadopoulos, interactive media director for Euro RSCG4D, didn’t hear anything before Google’s announcement, at which time the search engine had been experimenting with single-session behavioral targeting for weeks. Her agency has a number of large purchases with Google, “and I wouldn’t be surprised if some of our campaigns, especially automotive, were part of this new serving strategy.” Still, it’s galling not to be told about these things. “They’re not very forthcoming with innovation as it relates to the search engine. A small business is one thing, but when you’re working with agencies and big advertisers, I thin you owe it to them to [convey] any changes in algorithm that would affect your advertising.”

Why would Google keep a low profile about this? The most obvious reason, of course, is DoubleClick. U.S. investigators are looking into the search engine’s purchase of the online advertising company, and there are concerns about how well Google will guard its users’ privacy now that it has this kind of temptation to tap into a major source of ad revenue. If Google wants to be permitted to close the deal with DoubleClick, it is going to have to make sure that anything it does concerning user information will not trip the alarm bells of privacy advocates.

{mospagebreak title=Google and Behavioral Targeting: Patented Proof}

Google may be moving slowly, but it can’t afford to pass up on behavioral targeting completely. Its major rivals are starting to use behavioral targeting of various kinds in their advertising programs. Just look at Yahoo’s Smart Ads.

Truth to tell, Google has been interested in behavioral targeting for quite some time. Companies may protest all the time that you can’t tell what they’re going to do simply by looking at their patents, but those patents are proof that someone at the company thought that area was worth investigating. And Google has filed for patents that relate to behavioral targeting.

The patent applications were filed back in October of 2005. One was titled “Determining ad targeting information and/or ad creative information using past search queries.” This patent is for a technology that generates both keywords and an ad creative for an advertiser based on the content of the advertiser’s web site or landing page. That’s a pretty innovative use of past search information; it definitely goes beyond the standard form of behavioral targeting. If it is effective, it could even be more profitable by creating ads that are more likely to generate click-through traffic.

The second Google patent application was titled “Results-based personalization of advertisements in a search engine.” It was published in October of 2005, but actually filed in 2004. With that title, it sounds very much like Google plans to use its Personalized Search service for conventional behavioral targeting. To quote from the abstract: “Personalized advertisements are provided to a user using a search engine to obtain documents relevant to a search query. The advertisements are personalized in response to a search profile that is derived from personalized search results. The search results are personalized based on a user profile of the user providing the query. The user profile describes interests of the user, and can be derived from a variety of sources, including prior search queries, prior search results, expressed interests, demographic, geographic, psychographic, and activity information.”

Correct me if I’m wrong, but I can’t imagine how this can be seen as anything other than the “traditional behavioral targeting” that Google emphatically denies it is interested in doing. Judging from Google’s current moves, though, it isn’t a matter of if, but when. Count on hearing more announcements from Google that sound like baby steps toward behavioral targeting. Google will insist they aren’t, of course, but advertisers and searchers should know better. If Google can be trusted to handle the personal information of its users with all of the care it deserves, this could be a win all around. If not, we could see more AOL-style information spills in the future. 

[gp-comments width="770" linklove="off" ]