7 Tips for A/B Testing Your Facebook Ads
When Facebook Ads first started becoming available to digital marketers, it was overwhelming to wrap our heads around, yet glorious to be able explore the vast targeting opportunities for advertising.
Since demographic and interest data is input by users themselves, you are able to reach exactly who you want, when you want, and tell them what you want through Facebook ads. You can target people who like Cinnamon Toast Crunch but don’t like Lucky Charms, or people who recently became single.
Thus, with the robust targeting capabilities available, how do you make sure your ads are resonating with users and getting them engaged with your brand?
The answer is test, test, and test. Oh, did I mention testing yet?
Knowing where to start when it comes to A/B testing your Facebook ads is a challenging yet crucial step. It’s challenging because there are so many directions you can go and it can sometimes be difficult knowing which element is the “right” one to test. You can test everything from ad copy to images and gender to age range. However, it’s important to decide which of these elements you’d like to split test, and letting them mature in order to achieve statistical significance.
Here are 7 tips or best practices that I’ve developed over a few years running Facebook ads.
Start by testing 2 distinct things - make sure the differences are significant!
- For example, test two completely different ad creative or ad copy
- Test Wide Age Ranges (18 to 35 VS 35 to 60)
- Don’t have too many split tests running, or else you won’t be able to really determine what actually performed the best (especially if you have a low budget)
Once you come to a conclusion on the split test, refine it or test more
- Break down the age ranges more
Don’t waste time on things you already know
- For example, if it's a high heel brand, start by advertising specifically to women. Exclude all other genders.
- In future, you could run a split test on your best performing products to men, with messaging catered to that demographic.
Best Practices: Create separate ad sets to get better data
- If you keep them in the same ad set, generally Facebook will quickly pick a favorite (Easier when you have a larger budget to employ)
- Test what the CPC or CPM is before split testing, then multiply that by the number of things you are split testing and by how many conversions you need to have a legitimate test.
- Try doing something like $5 (CPC) X 2 (# of split tests) X 10 (# of conversions needed to be valid) = $100
- If you have a big budget set this budget in one day, to determine the split test faster
- If you have a smaller budget, set budget to $10 a day and give the the split test 10 days to run
- Recommendation of testing until 10 to 20 conversions, leads or clicks (depending on what you are looking for)
- You will probably not need as big of a budget if running clicks, b/c they are cheaper
Make sure to test statistical significance
- Use a statistical significance test such as this link!
- Trackable Links will give you more reliable data
- UTM Tracking Parameters for Google Analytics tracking
- Customized Bit.ly url works as well
Things you can split test:
Campaign Split Tests:
- Ad Type
Easy A/B Tests for Ad Sets:
- Precise Interests
- Age Range
- Mobile OS
- Custom Audiences
- Job Titles
- Relationship Status
- Purchase Behavior
- Education Level
- Bidding (cCPM, CPC, CPM)
Easy A/B Testing at the Ad Level:
- Ad Copy
- Landing Pages
To conclude, there isn’t really one correct way to A/B test your Facebook ads, so being creative and experimental would be an asset. Just make sure you’re accounting for statistical significance and testing just one element at a time!