–

Revealing the true identity of unknown users

MM: What other techniques-either innovative or just standard-issue data analysis used in a new way-do you use in identifying the true identity of this inferred-unknown user?

I think maybe the only difference or uniqueness would be like tiers of knowing someone. Clearly, when somebody comes into the site and log in, they’re a member or they’re registered. We know immediately who they are.

Maybe the next step down or a lower level would be a cookie-based approach. Then I think probably beneath that — in terms of the level to which we know an individual… We go from trying to identify an actual John Smith at 1234 Main Street, to identifying individual patterns of engagement.

It’s not so much about identifying an individual’s name and address. That’s obviously ideal. We’d love to have that level of knowledge. But when we don’t, the point is that our approach at Targetbase and — again, very consistent with what Alterian does and the way they think… “Let’s know what we can and let’s infer the rest.”

Let’s say somebody comes to the site and there’s not a recognizable cookie on a machine. We have no idea who they are. But as soon as they start doing stuff — as soon as they start engaging with us — we start to get an understanding of who they look like, if you know what I mean. In terms of their pattern of behavior.

Again, back to the idea that we discussed earlier — of segmentation. What group of similar patterns of engagement would we lump this person into, based on the information we have, so far?

The other idea maybe that I would throw out there is  “retroactive identification.” When an individual comes to a site, they allow a cookie to be placed on their machine. Let’s say they engage with us over a period of a few weeks or months. Then they decide to register. Then they decide to sign up.

One of the things from a technology standpoint as well as from an analytics standpoint that we do that may seem simple and straightforward, but you’d be surprised at how many don’t think this way and don’t do this. We can connect that previously anonymous activity to this known person now.

MM: All that unknown data. Yes.

All their previously unknown data.

MM: Sure.

Now that’s rolled into it.

Here’s a brand new registrant — but now we actually do have longitudinal data on them.

MM: In fact, that would also give you a whole set of metrics or insights in terms of how many clicks — how many particular content-consumption cycles occurred by type or class that it took to get somebody at the register.

Absolutely.

Idea of behavioral targeting

MM: Does Targetbase pull any data in terms of these inferred-unknown people or users from ad networks?

Yes. We’ve certainly used that. Depending on the circumstances and the client need.

You get into the whole idea of behavioral targeting. That’s certainly very viable and will become increasingly so. A very viable approach to targeting — certainly ads, and perhaps other things.

One of the criticisms that I have of the standard network approach — behavioral targeting approach — is it’s such a limited view of an individual’s online behavior. Even with really large network properties online. You’re getting a potentially very skewed view into an individual’s behavior — which can still be very powerful from an ad standpoint, depending on your objectives.

To answer your question, we have — in a couple of instances — mapped in some of that data, where it made sense for our clients on a particular ad network or property that would gain additional insight into the consumers on the database, in terms of their online behavior.

I would say that’s more the exception. Simply because when we’re [looking at] that, we typically lean more toward panel data. Matched panel data — more like your [COM] scores. Nielsen Net Ratings, we’ve used. And/or we’ve conducted primary research, where it made sense.

–