A/B testing is the foundation of successful digital campaigns and in all honesty, it’s not as difficult as it sounds. Setting up A/B tests could save you a lot of time and money because sometimes all it takes is a few small tweaks to double sales and increase ROI.
Even if you think you’ve gotten your campaigns down to as good as they can get, Facebook updates their platform often, so it might be worth running a test just in case anyway. Whether you’re just getting started or are a seasoned ad guru, let’s take a closer look at A/B testing on Facebook.
What is A/B testing?
Simply put, it’s comparing two things to see which one performs better.
You can compare anything and everything: visuals, text, emails, conversion funnels, target segments or products.
Facebook allows you to test your creative, audience and placement. They will automatically split your budget evenly and randomly divide exposure so there will be no overlapping audiences.
There are two ways to do this: through Ads Manager or Experiments.
With Mediamodifier templates, you can easily edit the text, colors and backgrounds of the visuals that you want to compare results for.
You can also create mock Facebook posts to see what they’re going to look like together with captions before going live with them. This is especially great for marketing teams or agencies to help decide which visuals or copy to go with.
What to look for?
When conducting your A/B test, Facebook will suggest using Cost per Result, CPC (Cost per Click), Cost per 1,000 People Reached or Cost per Purchase.
In most cases, those will be the things to look for, but you can go so much deeper than that!
You can measure custom conversions on cost per 3-second video view, cost per messaging conversion and so much more. If you have Facebook Pixel installed, you can measure custom conversions as well, for example, how many people visited a certain page on your website.
Option 1: A/B testing through Ads Manager
With a recent update, you need to first set up a control and you can add variants after publishing.
Creating the test
Set up your campaign or ad set as you normally do.
Then, select the campaign, ad set or ad you want to test and click on the measuring glass in your toolbar to set up your test.
Choose your variable
You can now either create a variant or compare existing campaigns. Let’s compare audiences to see who reacts better to a “Recap of 2020” post.
Set a budget, time schedule & key metrics
The budget is up to you and will be split evenly between both options. The more you spend, the more accurate the results, but realistically the difference is only a couple of percent.
Use the amount that you would normally use for a campaign. If you’re not sure, just go with the default suggestion you’re shown.
The length of the campaign can be 1-30 days. Facebook suggests a length of 4 days for the best results, but if you’re running a test for conversions that generally take longer i.e. brand awareness campaigns, you can run it for 10.
Then, choose what you want to measure. You can add up to 10 conversions.
From the drop-down menu, you’ll see the most popular measures, custom conversions and other standard events. Think about your campaign objective and choose the one best suited for you.
Review details and start
Before publishing, review the details to avoid any errors and start the test!
You can track results in Experiments and Facebook will send an email with the results when it’s finished.
Option 2: A/B testing through Experiments
You’ll have the exact same opportunities as creating the test through Ads Manager with the difference of everything being on the same page, including the results.
Once you get the results, you’ll be presented with the winner based on your selected metric with a breakdown based on gender and age as well.
Here is an example based on a low-budget event engagement A/B test where we tested whether people are more likely to engage with a Facebook Event ad (A) or a Post ad with a separate link (B).
Congratulations! You’ve successfully completed your first A/B test on Facebook! You can either run the winning ad straight from the results page on Experiments or continue testing other variables.
Have fun with it and before long, your ads’ cost per result will be lower than you could have ever imagined.