When Google came out with their optimize for clicks/conversions ad rotation option, it was a great alternate to manual, lengthy ad copy testing. At OPM, we are maniacal about reporting; ergo, we decided to take an in-depth look into this. While there are many other variables when looking at this, we took a crude example of just the two variables AdWords was allowing to optimize for: clicks and conversions
Let’s assume you are running three ads for an E-Comm client and a conversion is worth $25.00. Let’s compare 10,000 impressions with varying click-through-rates (CTR) and conversion rates (CR):
In Figure 1, you can see that neither clicks nor conversions are the main drivers for profit, but rather, a healthy combination of both. When considering the conversion process, many other variables are involved for a conversion to happen for a particular ad. OPM recognizes that to truly optimize your ad copy you must consider more than one variable, while controlling your experiment. The number of variables and how you measure them will vary depending on the client. For example: A branding client may focus only on clicks and quality score.
For those of you who are ad copy testing savvy: Why do you find that your control ad copy always wins over your test ad copy? Furthermore, does it make sense to test up to 50% of your impressions on ad copy that is untested and could be performing worse than the current ad copy? These questions should entice you to think about ad copy testing differently.
Designing the experiment for ad copy testing can be similar for many clients by following these steps:
- Always rotate ads evenly
- Don’t let AdWords decide on clicks or conversions for you, rotate ads evenly to ensure you control the variables to test.
- Create a copy of the control
- This ensures that ad history is taken out of the equation and also exposes the unproven test to less costly impressions.
- Gather the right amount of Data
- Don’t make decisions off small amounts of data; you should always gather enough data to be 95% accurate. This is common among statistical experiments and eliminates the occurrence of Type I and Type II errors.
If you would like help analyzing your account or more information on effective ad copy testing, call us at 1-877-676-2010.