Beyond Keywords: New Dimensions In Search Behavior
Behavioral Insider: What do you see as the major gaps or limitations in paid search targeting as it's currently practiced?
Brad Bostic:The challenge of search targeting has been to go beyond just using keyword search data. As rich as conventional keyword data seems to be at first glance, in actuality the keywords people type in their computers don't always tell you much -- even about what subject matter they're interested in, much less what kinds of products and services they're most likely to be interested in. Commons sense tells us that keywords are still very ambiguous. The obvious example people give is is when people type in the keyword "apple." Does an advertiser know whether their ad is going to be seen by someone who wants an iPod, or someone who wants to buy an apple orchard or is looking for a Fiona Apple CD?
BI: How will what ChaCha's doing change that paradigm?
Bostic: In our case, the way we can see making search targeting more relevant is the same way we're working to make search itself more relevant, by adding the value of human intelligence. We are using, in addition to computer search algorithms, tens and soon hundreds of thousands of human guides whose job is to interact with search users to make sure they're able to find what they're really looking for by delivering search results customized to each user's needs.
In terms of ad targeting, the first thing customized search accomplishes is expanding targeting beyond keywords to real subject matter. That's the first big value proposition -- but when you're dealing with a truly human interactive augmentation of the search experience, then there's no reason targeting needs to be limited just to a static subject category, either. The point of enhancing search with human intelligence is to learn not only what subject areas keyword searchers are hoping to get information about, but also more broadly what their interests are. Another job for guides is to sift through queries and make sure they are being categorized into the right subject categories and sub-categories.
BI: How does that differ from what we know as contextual targeting today?
Bostic: Contextual advertising just targets users based on the general subject content of the material they are looking at in a given moment. It's badly misnamed in that it's not based on any understanding of what the actual context is of that user's interests. To actually deliver contextual relevance means developing an understanding of how unique searchers search, and what core interests really motivate their search.
BI: So that's where the behavioral dimension comes in?
Bostic: We can, over time, learn more about search user behavior in terms of how specific searchers use their keywords to explore the subjects they're interested in, by tracking how their stated interests relate and evolve. So over time we're constantly refining the ability to understand the user's querying behavior in the context of what we know their main interests are.
BI: Will your targeting be based entirely on the pay-per-click paid search model?
Bostic: Another big limitation in how advertisers can target search is in the types of ads they can deploy, which are very limited so far. We offer a pay-per-click traditional model but also display ads, especially rich media and video, Traditional search interactions take only a matter of seconds -- but interactive search with guides takes on average a few minutes. During that period, as searchers are working with guides to tailor their answers, display ads can be targeted by subject matter and user interest.
So imagine that someone is interested in athletic shoes. Not only do you base targeting on their keyword search but, based on the kinds of questions they've asked in the past, you learn what kind of athletic shoes, whether they're a middle-aged mom interested in taking up jogging or buying a gift for their teenage son. During the guided search session you can target short, video-rich media based on what their needs and interests are.
BI: What are your goals for targeting going forward?
Bostic: Human interactivity and intelligence in the form of guides, and user community interaction, is at the heart of what we do -- and we are only beginning to introduce new ways of augmenting the search process. As we do, the system is designed to constantly learn more about how search behavior relates to personal interests. The goal is to continually improve the ability to customize the user search experience for greater and greater relevance. Delivering advertising that directly relates to what the user is looking for is part of that process.