How much Data is Best for Split-Testing Ad Copy

(Click here to expand)

When trying to decide what ad copy is most effective in moving a viewer to a click, to a conversion each step of the equation offers it’s unique input into what drives a viewer of an ad to convert.

Step 1: Impression to Click

The almighty click through rate, CTR, has been regarded by pay per click marketers across the industry as the primary focus of PPC performance. Since this metric supposedly weighs heavy in the computation of the AdWords quality score it seems many marketers have no problem driving up high click through traffic even if it does not convert.
Impression to click is really the least important factor in determining the success of the bottom line of the purpose of the campaign, to convert visitors to clients. CTR is most useful when determining the strength of headlines, but becomes less useful when testing description text and display url’s.

Step 2: Click to Conversion

Conversion rate is a metric many pay per click marketers just felt they didn’t need to pay attention to. When I first began researching and managing PPC accounts I was contracting for an agency in Boulder, CO and was told not to worry about conversions, that was the job of the Web team. The idea was that it was my job to get the people there and their job to make them convert. What happened is a lot of traffic that was generated was not even qualified, much less actually going to convert. Disconnected data led to bad results.
By taking an integrated approach to Interactive design, Creative content, and Search marketing an environment is created in which departments consult with each other about the data collected and collaborate on implementing the best solution.

First, the importance of setting a baseline for when there is enough data to do the analysis is necessary. I draw the line at a thousand search network impressions. It is important to draw this line in the sand and stick to it. By knowing I have adequate quality data to analyze I can be confident that the conclusions I draw will provide the optimizations needed to continue to improve performance.

Comments are closed.