Frequent testing of offers, copy and product display (i.e. “ads”) should be an essential part of a continuing website improvement process.  Whether conversion objectives are measured in unit sales, revenue, subscriptions or click-throughs, testing can help verify that a site’s pages are producing conversions at peak performance. The most common format is A/B testing, where visitors view one of two ads, and the ad producing the highest conversion rate is considered the “winner”. The winning ad is then served to all visitors.  Multivariate testing is a variation of A/B testing, using more than two ad variations.

Working with our clients, we have come to realize that conventional A/B and multivariate testing is cutting conversions short. These testing methodologies have been transferred from the direct mail world where the objective is to find the one best ad to mail to each household, and assumes that it will be viewed only one time. The issue is that these methods (and the associated math) don’t take advantage of recurring visits.  Furthermore, in contrast to direct mail, serving ads online has practically no marginal cost.

Reducing ad performance to simple conversion rates tends to obscure and hide details of ad performance, details that are the key to providing additional lift.  Our research shows that each ad has a unique conversion rate. Ads that may convert at high rates the first time they are seen but then quickly decline in productivity (humorous ads which quickly become non-numerous) may appear at the end of the test to perform as well as ads which improve steadily with each ad viewing.  Understanding the performance details of ad saturation and conversion rates by the number of times an ad is seen are essential for getting the most out of an inventory of ads.

Numeric Analytics, a website analytics consulting firm, has developed a testing and monitoring process that takes advantage of these performance details. Based on predictive modeling techniques, this method produces lift over and above conventional A/B testing in a way that compliments behavioral targeting and other optimization techniques.

For additional information, please contact Steve Chidester Business Development Manager

801-456-3750 – Office

801-831-5730 – Cell

stevec@numericanalytics.com

http://www.numericanalytics.com/