Behavioral Insider: What do you think is the most misunderstood thing about behavioral data by media owners and advertisers?
Roy Shkedi: The challenge of BT is to truly scale data. Think of all the hundreds or thousands of sites that are out there on the larger ad networks which they're trying to monetize but of which they really have no idea who their visitors are. Social networks, which represent a large portion of the kind of remnant inventory ad networks sell, are a good example of what I mean. In theory they have terabytes of data. But if you drill down, most of that data is irrelevant for advertisers. What does it matter, for instance, who consumer X is friends with?
BI: Explain what you mean when you refer to post-search data.
Shkedi: What's relevant is data that illuminates purchase intent. We believe a general search engine is just the starting point, if at all, in the research process a consumer goes through when considering to buy a product or a service. As Web users get savvier they don't need the general search engines to find what they want. If they are looking for a flying ticket or a hotel for example they will go directly to a travel or airline site without going through the general search engine first. If they are looking for a digital camera, some might still start the search with a general search engine but very fast they will find themselves on a price comparison shopping site or on a product review site.
While the general search engine only knows that the consumer is looking for a 'digital camera,' the vertical site knows which kinds of cameras, what main features and what price range the consumer is looking at. An ad targeted to the consumer wherever he is on the Web based on the vertical site information would therefore be so much more effective than an ad targeted to the consumer based on the general search site information. In other words, as Web users get more savvy they become far more focused on finding exactly what they want and are spending far more time on vertical sites. Post search data is where our unique focus is. We have accumulated relevant purchase intent data on tens of millions of US/UK users that are in purchasing mode for products and services.
BI: Once you've selected and aggregated the right data, what are the main challenges you see in deploying it correctly?
Shkedi: Our platform analyzes received profiles in real-time, categorize them and make the data available to the ad networks and publishers that indicated their interest in such a category. The preeminent way to do that is to organize data into deep verticals. Our data has been categorized on an advertiser-friendly and privacy-sensitive basis, and is being made available today to ad networks and large publishers that are seeking to maximize their targeting capabilities. Right now we have 40 different vertical categories of purchase intent data. For example, ad networks who are very interested in insurance-related products would use us to receive data related to insurance consumers so they could cookie them as demonstrating strong behavioral purchase intent for insurance-related products.
BI: I know AlmondNet is quite involved in privacy initiatives. How important is insuring privacy to the future of behavioral data aggregation?
Shkedi: We are very sensitive to privacy. We don't collect personal-identifiable information and we analyze search queries to make sure we don't store any information that can personally identify a person (as happened unfortunately to AOL). From early on we've been adamant about not working with any partners who don't have clear privacy policies in place. For example, policies must explain to users that data is collected and that they have an opt-out. We strongly believe that it is in the best interest of both consumers, and the industry itself, to self-regulate its data usage. In order to achieve this goal I think that all of us need to do our share.
BI: Where do you see the learning and adoption curves for behavioral data going in the near term?
Shkedi: The challenge going forward is that advertisers, once they see how much of a difference post-search purchase intent data makes within larger vertical categories, are looking to achieve deeper granularity within those verticals. So before, an advertiser would be very happy to locate a large number of buyers who have indicated with strong behavioral signals that they are in-market for autos. Now the goal is to be able to find a large number of consumers who can be identified by strong purchase intent data to be in-market for a particular model or a particular brand of auto.
What I think is, the leaders in behavioral targeting in the coming year will be those who can combine deep granularity in very highly defined niches with large scale. So far it's been either/or -- either scale with weak granularity or strong granularity without scale. Media owners, and advertisers, are ready, and frankly need to go beyond that, but to do so means leveraging a far wider universe of data in a much more focused way than they've been able to do till now.