22 Sep 08 Ad Creatives Impact on Multivariate Tests
Inspiration Point: How to use ad creative as part of a multivariate test
Ian Lurie suggests that it is imperative to incorporate your creatives into your MVT matrix. I do believe this can be a good idea if you are running a high budget PPC campaign. The only problem I see with this is I usually like to build a relatively high number of landing page specific creatives. It could be a huge pain to incorporate all your creatives into your multivariate test. You may also run into bidding problems going this route.
For example, if you are paying $.25 per click for creative A and $.10 per click for creative B and creative A results in 5% higher conversions…is it really worth it to push creative A and drop creative B? Maybe the keywords for creative A drive 100x the search volume as creative B. I still don’t see that as justification to drop creative B. Instead this would be an indicator that you should try to optimize creative A to see if you can create a cheaper or more cost effective variation.
Now given my above example what would you do if you are targeting creative A and creative B without plugging them into your MVT matrix? Well you would probably make the exact same conclusion even if you think creative A might be driving 5% fewer conversions because cost and volume are factors you shouldn’t overlook. In PPC it is good to maintain a high volume of landing page specific creatives targetting a wide array of long tail keywords.
I’m not suggesting Ian is wrong by any means. The more data you have the better you are equiped to make the best decisions when it comes to optimizing your landing pages. I just think you need to factor in the setup time and the size of your campaign before feeling obligated to use creatives as test elements.
Leave a Comment
