A successful information portal in the educational sector, used by 1 in 3 college-bound high school seniors, wanted to increase advertisement response rates by predicting which promotion each unique visitor was most likely to respond to.
The system in place was already extremely successful in optimizing ads, selecting between hundreds of ads using “A-B testing” – or, more precisely, “A-B-C-D… testing”. It measured which ad was most popular, universally across users, but separately for each web page, which, in this case, closely corresponds to the lifecycle of the user.
This made for a formidable system over which to improve, given that a small number of the ads were much more popular than the rest. This also created a challenge for predictive analytics to ever confidently say, “Hey, this user is much more likely to be interested in this relatively unpopular ad than any of the really popular ones”.
The Predictive Analytics team created 291 models – one per ad. The models were generated over millions of ad renders, observing which user was served a specific ad, and whether that ad resulted in a click-through to conversion – in this case an “opt-in” for subsequent communication. Once complete, it was time to put the models to work – providing a refined ad targeting solution.
Once deployed, the results of the new Predictive Intelligence based ad targeting solution were outstanding: A 25% increase in response rate beyond that generated by the existing system, which translates into an estimated $1 million of ad revenue every 19 months.
The increase in ad revenue, coupled with increased visitor engagement provided outstanding ROI for this predictive analytics effort.
Case study provided by Prediction Impact, Inc.