• by Steve Smith , Staff Writer @popeyesm, August 3, 2007

How many impressions do you need to serve to a past visitor before you admit diminishing returns? That is the question Andrea Palmer, manager of interactive services, Siquis, Ltd., and her colleagues asked of their retargeting program for Spirit Airlines. When they ran the numbers on reach and frequency and discovered how many times customers were seeing the same ad, a radical reduction in the retargeting spend produced a  surprising effect: none at all. Palmer tells us what they learned.  

Behavioral Insider: What kind of volume are you dealing with at the Spirit Airlines site?

Andrea Palmer: Re-targeting is a really important part of the media plan, and it does very well in terms of converting users who are familiar with the product and have been through the booking engine or at least [are] seeing it on the site. From our media every month we drive about one to two million uniques, and there is about four to five million in [overall] traffic. And that feeds into the retargeting portion. We also found it is more efficient to watch the ratio of retargeting to straight up media. If we are not driving new people to the site, then we have lost our bucket of people to retarget to. If we can't maintain a certain flow of new users and we can't maintain a certain turnaround and a constant changing of the new users on the site, then we're just getting the same customers.

Behavioral Insider: Explain the analysis you did on the natures of the spend and how you recalculated the value of the retargeting.

Palmer: We used Atlas as out server and used their advanced reporting. We ran a few different reach and frequency reports to see what frequencies we were actually getting vs. what we anticipated. We found with our retargeting portions especially that basically 90% were repeat impressions. So each person getting an impression from our targeting was getting 200 to 300 impressions a piece in one month. So that is where we said -- that there has to come a point where we don't have to continue to re-message. We can either consider them converted or consider them not going to convert. If you see an ad 100 times in a month, do you need to see it one more time before you convert -- or are you probably familiar with the product? You're either not flying anywhere, not looking to take a vacation, or not interested in the destination -- or the cost point is not what you are looking for.

Behavioral Insider: So many buyers complain about too little scale in retargeting, and you were finding the opposite.

Palmer: For us it's not about getting enough; it's about not going overboard and getting too much. So we pulled that back and took a chance and cut it by half. We found that it got us the same amount of revenue. By taking half the [amount] we had been spending we got the same results.

Behavioral Insider: Did you reinvest that in another areas?

Palmer: It was enough to let us go into another network or another one or two programs. This was a good amount of money we saved.

Behavioral Insider: Are you also adjusting the nature of the retargeting, or just reducing frequency?

Palmer: Definitely. The initial strategy was to hit people at the home page. But we also have a loyalty program. So we hit people who may be coming through a bookmarked page, not to the home page, or through the e-mail marketing. That is where the creative comes in. We can talk to our different audiences with different creative. So if our loyalty program is being re-messaged, it is in a different way from people we assume have not been at our site before. We can use destination-based targeting. So [if] we know that they fly out of Ft. Lauderdale, we can messages that with flights from there that are on sale this week.

Behavioral Insider: How granular can you get with custom creative and still get ROI?

Palmer: Well of course the pool of users isn't so astronomically huge that we have to have 50 different versions of the creative, but if we can make it just a little bit more relevant, then it is worth it. We don't go overboard. We have maybe a couple of different buckets that I would call substantial. We can certainly identify maybe 20 to 25 different segments we could theoretically have multiple creative for, but it doesn't make sense with all the different ad sizes and build-outs [that] would require. But every little bit that we can do -- a slight difference in the headline -- can work to our benefit. It just makes people feel you are talking to them rather than the masses.

Behavioral Insider: What would be the next level of sophistication and depth after this analysis?

Palmer: We could do a lot more testing in terms of frequency. Is it better to go deep and strong with one or two or three retargeting networks, or is it better to split that up among every single publisher you are using? Publishers will say they can do 100% retargeting as long as your audience and traffic is high enough on the site. Is that the best strategy? I don't know. Publishers seem to think that it is their best tool, and I think there needs to be a more sophisticated approach than just going to retargeting. 

Behavioral Insider: What have you learned about moving the needle on retargeting to another level?

Palmer: We learned how to maximize small amounts of money. How to correctly scale it. How to make sure that we are looking at every possible audience rather than assuming there is only one audience. There is a lot more that goes into it. On the surface retargeting seems like a simple way to get your message down. But I think it has gotten to the point where it can be used as a quick fix or an easy way to make goal. When you get into the data and dig down a little deeper to see the effect of those changes you have made, I think there's a lot to be learned that way.