November 14, 2017 – This month’s SEMpdx event focused on ad copy optimization. 3Q Digital’s Caitlin Halpert, Senior Director of Client Services, presented how to run more effective A/B tests on Google, Bing and Facebook. Here are 10 takeaways from the evening:
- There are three factors to consider when testing ads: sample size, statistical significance and practical significance.
- Differences in ad format performance are usually the result of differences in inventory. We can’t compare ad sizes for a fair equivalent. With that in mind, test each format independently.
- When it comes to Facebook ads, each ad has a life of its own. You can’t compare ads on this platform.
- To run a fair A/B test that’s statistically valid, follow the scientific method when appropriate: Formulate a question; create a hypothesis; design a controlled test; analyze the results.
- Ads must be eligible for the same auctions so that results reflect the same audiences on the same sites. They also need to enter on a level playing field, which requires re-setting ad history and re-launching legacy ads when the test begins.
- The output from proper testing will be useful. For example, the outcome will inform future optimizations, results will be practically significant, and there won’t be confounding factors.
- Using the “optimize” setting for your ad rotation can drive 10-20% more conversions than the “rotate evenly” setting.
- Create ads that are meaningfully different variations.
- Differentiate between data-based and data-influenced decisions.
- Get creative with ad copy, ad customizers and extensions.
Caitlin closed by emphasizing the importance of questioning everything. One final tip: Learn what you can about the system, not about ad language.