Facebook Ad Testing: A, Please Meet B
For veteran digital marketers who have spent any time in search or display, ad testing is old hat. In fact, many would attest that typically the ad projected to be the clear winner often comes up short. For that reason, it’s imperative to apply this same discipline to a Facebook advertising strategy. While the Holy Grail in online testing is the sophistication of a multivariate approach (MVT), marketers just getting started can yield strong results from even the simplest A/B testing.
What is A/B Testing?
A/B testing is the method of taking at least two versions of an ad and changing a single element to understand that element’s contribution to the ad performance. The following is an A/B test for a headline:
Marketers often want to put forth many “best guess” ads and see which one performs better (thinking they must know their audience the best). While this approach can provide the initial control ad (using the best performing from the bake off), continuing to test in this manner provides no insight into the elements providing lift and doesn’t allow further iteration in a focused manner.
How to Create a Facebook A/B Test
Per the examples above, there are three key elements of a Facebook ad that may be tested:
For the test, pick one ad that will serve as the control. From here, choose which element to test and then create the number of versions of this element to include. For example, if the image is the first to test, create the same ad with at least 2 different images (keep in mind there could be many more versions of the image to test making the test more of an A/B/C/D test but keep in mind that each should only have a different image – all other elements should remain the same test).
Campaign Set Up
Next, set up the test within Facebook. Be sure to set up each ad as as separate campaign so budget for each is unique. This is important because Facebook will optimize spend based on performance of ads within the same campaign. Because this is a test, we want ad spend and serving to be consistent across the ads and creating separate campaigns (with separate budgets) for each will do just that.
Assuming this advertising test is focused on a performance metric, be sure to choose one. What will success mean?
Dude Ranch Dream Vacation
- Engagement: Are you looking to drive clicks to a Facebook page or a landing page hosted on your own site?
- Conversion: Are you looking to drive a conversion event? Perhaps liking your Facebook page, joining your app or signing up for an email newsletter on your own site?
Decide what will determine which ad is the winner before you start.
Budget and Test Duration
As discussed above, be sure to set up the ads in separate campaigns so each ad test will have its own budget. In addition, when deciding on bids, consider starting below the Facebook suggested range. Based on how much traffic you receive, you can always increase them but it’s smarter to start lower and try to get the traffic at the best price. Also, keep in mind that each campaign / test must have the same targeting – if you don’t keep all elements consistent across the test, the results will be skewed. Finally, for the test results to be valid (statistically significant), a good rule of thumb is to accumulate at least 100 clicks per campaign / test. If your control (A) has 10 clicks and your test (B) has 5 clicks, you won’t be confident that those results will hold true if you were to push the test more broadly. And, if there are hour of day or day of week trends in your business, consider using daily budget caps to help spread the traffic over a representative period.
Setting up a quick and dirty testing approach to Facebook advertising doesn’t have to be hard. The growth in ROI will whet your appetite for a more complex strategy but don’t get bogged down in huge factorial manual approaches or the need for a platform vendor right away. Start simple and build the case for more sophistication with the solid results this method can produce.