How to A/B test your Facebook Ads

ab testing facebook advertising

Complete the form on this page here to receive a free A/B Testing excel spreadsheet template

A/B testing plays a key part on any advertising platform. Facebook is no exception. There is a reason why Facebook decided to introduce A/B split test in 2017 – all advertisers who were seeing any tangible results were already doing it in a “manual mode”.

Here is what A/B testing means and how you can utilise it to optimise for best possible results on Facebook.

What is A/B testing?

A/B test is an experiment, where you make incremental changes to one of the variables in your campaign and see how that change impacts the performance and results of that campaign.

You can test almost any aspect of your campaign: audience, placement, creative and delivery.

Here is an example: You want to sell pencils to men in their 30’s in San Francisco. You theorise that the colour of the pencils may impact the result. So you create a campaign:

a) Location – San Francisco.

b) Placement – on all platforms.

c) Creative – static image – pencils on the table.

d) Delivery – standard.

You then add 4 ad variations:

Ad A – red pencils

Ad B – blue pencils

Ad C – yellow pencils

Ad D – black pencils

You set your budget and decide that you want to give it about 4-5 days to see which ad performs best. Let’s say your budget is $200, which is about $50/day, and $10/creative.

24 hours after your campaign has been approved, you notice that C & D ads have spent $10 each, and generated 2 clicks each. At the same time, A & B spent $8, and generated 200 clicks. No-brainer. You now see that your audience is not interested in yellow and black pencils, so you can go ahead and switch off C & D creatives.

After another 48 hours, you notice that while A&B are spending your budget at the same rate, A is converting into sales much better. Brilliant. Red pencils for the win!

You can now test other variables in your campaign, using your strongest creative, knowing that this shouldn’t impact your results, and you can accurately experiment with other aspects of your ad campaign.

Of course, it’s never that simple.

A/B testing methodology

Step 1: Begin with identifying variables that are important to you. It is quite straightforward if you are testing one creative (see the example above). However, things get more complicated when you need to test different “sub-variables”. What if you want to test the pencil colour AND different pencil sizes? Well, if your monthly budget permits you, then test all at the same time, otherwise, take one variable at a time, and make sure that your testing is done methodically. Decide which variables would be important to test and why.

Step 2: Create your first campaign. Let’s suppose you are most interested in engagement. Create campaign #1, where you are testing one of the variables that you have previously identified. Make sure at the start to tick the “create split test” option.

Step 3: Select categories under which your variables (that you want to test) would fall. As of 2019 Facebook offers 4 options: Creative, Delivery Optimisation, Audience and Placement.  Let’s suppose you are testing an Ad Creative. After you have created Ad A and Ad B, you can select “Test Another Ad” option at the bottom of the screen, to create Ad C, Ad D, etc.

Step 4: You have finished creating the first campaign, and you have 3 different variants of the variable in question that you are testing. Excellent! Now, as mentioned before, if you have other variants that you want to test, you can create separate campaigns, where you repeat steps 2 to 3.

Step 5: Once all campaigns are ready, and you have enabled them, check on a regular basis, to see when they get approved.

Step 6: Once all campaigns are approved, the testing phase begins. Let the campaigns run for at least 12 hours, before you switch any of them off. Switch them off BEFORE this period expires, only if one of the variants or campaigns begins to spend money like crazy, without bringing in any substantial results. Always keep your ROAS (return on ad spend) in mind.

Step 7: Continue to switch off variants of each variable, until you have top two variables of each variant. (That is unless you were testing just two variables to figure out the best one, in which case just one variant will suffice). These are the variants that bring in the best ROAS for you.

Example) You have 3 variables that you tested. In total, you should have 6 different variations after the test phase ends.

Step 8: Test the combinations. Now is the time to test the combinations. Let’s stick to the example above and let’s say that you are dealing with 3 variables of the same campaign with 2 variations each. This means that you will have the total of 8 different combinations.

Figure 1

A or B – Audience

C or D – Placement

E or F – Ad Copies

Possible combinations: ACE, ACF, ADE, ADF, BCE, BCF, BDE, BDF.

This means that you should have 8 different campaigns running to start off with. Repeat steps 5 to 7 for all campaigns, until you have 3 best running campaigns.

Step 9: Time to expand on your results (this technique here we originally heard of from TSassKickers). Create 3 duplicates of each campaign (if you are following the above example, that should be 9 different campaigns).

Again, run them for the next 3-4 days, and switch off along the way any campaigns that are not bringing in the right ROAS. Creating duplicates help you reach your audience faster, and because Facebook algorithm works sometimes in mysterious ways, sometimes one duplicate will perform 10x better than the original or the other duplicate.

Step 10: Once you have identified the best campaigns from the above (it doesn’t have to be 3 at this stage, just whatever is bringing in the best results for you), it’s time to double down. Double the budget on these campaigns.

Monitor campaigns very closely from this point and onwards. Doubling the budget might mean that your spend can very quickly get out of control, so make sure that it is still worth running these campaigns, and again – kill the ones that are not performing well!

Step 11: While the results are coming in, and you still have a positive ROAS… continue to double your budget every 4-5 days. If this is still working for you, you clearly found a gold mine, and you need to cash in, while it lasts!

Step 12: Once all of your campaigns seem to have reached their limit… take a breather and pause everything. You have done it.

Now rinse and repeat with your next new campaigns.

Things to watch out for:

1. Don’t switch off your ads too early. Ideally, you want them to run for at least 1 day after they are approved. If you switch off the ad earlier, you are not letting it optimise properly, and you would miss out on some significant changes in the results from that ad, that could have come in if you let it run for another day.

2. Don’t test multiple variables at the same time. You don’t want to be guessing which particular variable impacted the performance of your campaign – always pick one to start off with, and work your way through each (if you feel it’s necessary).

3. Make sure that you take into account when all of your ads are approved by Facebook. Some ads may take hours if not days to be approved, so keep an eye out for it – it can radically mess up your data.

Complete the form on this page here to receive a free A/B Testing excel spreadsheet template