What technology are you using to A/B test your banner ads?
This question seems almost absurd in today’s online advertising and especially in direct response advertising. We do not use the standard methodology of A/B testing for banner ads at all. Instead, we use an ad server which profiles that traffic for us and matches the creative (banner ad) that is most likely to engage the particular user (or profile). It matches this creative from a pool of available creatives, and it keeps evaluating the performance of each creative separately, comparing it to the performance of the other creatives in this campaign. A question arises from this:
Why do we treat the serving of landing pages differently from the serving of ad banners?
I do not know the answer to this question. The only speculation I can provide as for the main reason that the serving of banners is far more advanced than the serving of landing pages, is the development of the online advertising industry. The industry evolved from offline advertising, in which all the focus is given to the ad. The call-for-action is much less accessible and thus less focus is placed upon it. This resulted in technologies being initially developed for ad serving and so the market was left without technologies to support sophisticated and optimized landing page serving; this resulted in less attention given to this part of the value chain of the campaign, as there was no technology to deliver the landing pages in this way.
What I do know for certain is that we should not treat the serving of landing pages any different from the serving of ad banners. We should employ the methodology described above, regarding creatives, for landing page serving. Currently we are not using optimized methods to select what landing page to serve. I would argue that we need to use superior methods and devote larger parts of our budget to the optimization of the serving of landing pages and its content; the landing page is the place where the user makes an actual commitment as opposed to just clicking a banner. In addition to that, we have a much larger asset at our disposal as a landing page occupies the entire screen as opposed to a banner that occupies only a small fraction of it.
As can be seen from the analysis generated by Traffiliate, a post click optimization platform (see screenshot below), users coming from different places (either GEO or website) and having reacted to a different banner, engage differently with each landing page. They convert in a different rate and like different pages. What can also be seen is that we managed to predict (based on historical data) which landing page will perform better for each segment, which means that the parameters used to profile the users have a cause-and-effect relationship and can be relied on, for future optimization. The number of options and user profiles that can be generated during a regular campaign are numerous and thus neither A/B testing (which is done on a limited amount of traffic in a limited span of time) not a manual process of matching landing pages to profiles/target audience can be effectively employed. Only an automated system perform this complex and tedious task of mixing and matching of the different landing pages to the different profiles.
All that needs to be done to optimize the serving of landing pages is:
- Create several landing pages for an offer
- Load/Link them to a landing page serving platform
- Link your banners to this platform and watch it optimizing the serving of your landing pages