BT's Next Stage: Custom Creative

Behavioral targeting works well enough when you hit interested customers with generic ads for your brand. But how much better does it convert in-market viewers when the creative itself addresses very specific details of those users’ recent online behavior? Greg Smith, CEO of EchoTarget, Inc., says his hotel and travel clients can see a 200% to 400% lift in transactions when the behavioral network sends the right destination-specific creative to targets. Smith helped develop Boomerang for DoubleClick, which in 1999 was one of the first retargeting products. Now he is exploring how and where the extra investment in custom creative pays off within the scale and efficiency constraints of behavioral retargeting.

Behavioral Insider: Explain how you get more granular data on site visitors so you can retarget them with more specific creative.

Greg Smith: What we call dynamic retargeting can collect information on the fly. We work for several travel advertisers where destination intent is a key driver in response. Many of our travel marketers send us dynamically information as to where that user wants to go when they went to that Web site. So, instead of saying this user went to this airline site and left, we can know that they went to the Web site but searched for a flight to a destination.

BI: How do you create a custom creative plan out of that data?

Smith: For one airline client, five destinations account for 50% of their destination search. We’ve collected that information and then we showed specific banners, not ‘great deals at our airline’ but ‘great deals to this destination’ for each group. We’ve also shown those users the general advertising as well. And we’ve seen a substantial lift in click rates [on the custom creative] of about 200% to 800% ranges. It is something we always thought would work.

BI: Custom creative always raises the problem of efficiency. How many discrete units are you making?

Smith: Right. The biggest thing has been creating the units. We just committed to a hotel client that will test seven different segments, and we will create three different sizes for each of those segments. Now that’s 21 different units to make. Everyone gets excited by this capability, but they don’t have the resources to create the units. So, about nine months ago we came to the conclusion we had to do this ourselves. We hired someone to make these creative units and typically provide it at no charge to the client with certain volume. We don’t want to be an agency, but they do appreciate the help.

BI: At some point the creative load overwhelms the benefit.

Smith: Clients start to ask if they can overlay this destination intent with the origin of the user through an IP address, and then we have specific flights from New York to London. All of a sudden, if you just add one more variable, that 21 units can grow exponentially and get out of hand. So the key in developing these programs with clients is figuring out what are the key variables that drive response. In the case of travel, it’s destination. In auto it is the vehicle type. In financial services it’s looking for home equity line, credit cards, or brokerage. Just finding someone who is interested in financial services doesn’t mean anything. You need to be much more specific and have creative units that match the segment.  

BI: Doesn’t custom creative also involve the ever-present scale problem in BT?

Smith: Absolutely. If an airline has five top destinations accounting for 50% of their searches, it doesn’t mean we don’t want to find the other half. We will just show them a generic offer. So the key thing is, slice the pie as much as you can, but go after the entire pie at the same time. That’s how you keep your scale. We know that static retargeting still is very effective.  The client in retargeting has to have pretty sizable volume as well.

BI:  What incoming traffic justifies something like 21 creatives?

Smith: In that particular case we want to have at least 100,000 users on a monthly basis, and I say at least.

BI: What creative strategies and appeals work well?

Smith: In the case of destinations, using an image is important. The Statue of Liberty for New York or Golden Gate for San Francisco can deliver the message to the user much faster than the words. Rich media definitely helps as well, but we are really in the first inning in terms of combining BT with rich media on a customized basis. Price points definitely work. But it is very difficult to manage. The marketers have to ask themselves, how long will the price be valid. Will it be worth it to swap these things out?

BI: Bottom line. Do you get enough lift to justify the effort? 

Smith: If you aren’t making an order of magnitude change in results, the client isn’t going o get excited about it. No one is going to pay you any more to improve results 25% to 30%. The clients will want the increase in results for free. It’s got to be an order of magnitude better, at least 100% or 200% better. Then people will see the difference. We encourage them to test whenever possible. In a few cases clients provided us with post-click information, which is incredibly helpful, so we can see how the segments perform on a transaction basis. We ran tests using two different banners. We show that user the generic banner in the normal campaign, and we showed a destination-specific offer. Same branding, same look and feel, but it was addressed to that particular market. What we find consistently is improvement on the transaction basis of 200% to 400%. That’s going to make it worthwhile. It does require additional creative and making different incremental trafficking. But that incremental work pays off in spades, and it’s a substantial improvement.