How to Test Your Creatives Right in 2019

How to Test Your Creatives Right in 2019

Testing is an essential and probably the hardest part of Facebook Ads Management.

We spend so much time on looking for the right setup in our campaigns and adsets – interests to choose from, optimization strategy, CBO/Adset level budget placement…

Yet, when it comes to what our targeted viewer sees, IT’S ONLY OUR CREATIVE that decides whether the potential buyer will click on the ad.

In our work, we devote tons of time and resources to CREATING/TESTING/IMPROVING our clients’ creatives.

There is no “secret sauce” how to ALWAYS make things work for a product, so you must test your ads and find winners that will bring you the money

So how should you test your creatives?

If you want to do that methodically and reliably, you need to dig deep, spend some time and resources, and get much data to choose from.

Here are the testing options that you can choose from depending on your available creatives:

1. Different call-to-action buttons (e.g., “Sign Up” or “Learn Now”).

2. Same video with different headlines.

3. Same video with different texts.

4. One image vs. one video.

5. One video vs. another video.

6. A square video vs. a full-size video.

7. One video vs. a carousel with 2-5 videos.

Here, you should eventually get a clear hypothesis based on the tests’ results that will show what exactly your target audience likes to see the most.

Don’t forget to check what your competitors are running! We recommend using AdSpy as probably the best tool for this purpose.

SO, we’ve got the testing options. How do we approach the testing itself?

1. Manual testing – 2-5 ads per adset.

This is a somewhat conventional yet proven method to test your creatives – find winners, kill losers, enjoy better performance.

The idea is well known to you – you pull all your creatives for the product and put them into one adset. Then, you see which one shows best results and scale your campaigns with only this ad on.

With this type of testing, you totally rely on Facebook allocating your budget correctly and serving the really BEST creatives to your BEST traffic.


– Facebook will optimize and allocate higher budget to the ad with HIGHEST CTR.

“Great” you will say.  Well, not always. Sometimes, it will be your CHEAP products with really low margins that we get you hardly over breakeven, comparing to your more expensive items, which ads might have lower CTR.

– Facebook might not always allocate budgets evenly at the beginning, hence getting “fake winners” for itself. This may leave your better ads unnoticed – so always check the metrics for ads that even didn’t have any conversion. Maybe they just need to spend a bit more.

How to optimize ads using manual testing?

a.Turn off ads with low ROI, even if other metrics such as CTR and CPM are good – monitor the statistics closely and regularly, because you LOSE THE MOST on not killing poor performers during the test phase;

b. Make the number of ads you test small – Facebook moves towards consolidation and simplification, so they really are not fans of overly complicated creatives’ tests.

If you are sure that your ads are of good quality – test 2-3 creatives. If not, you might go up to 4-5, but not more;

c. Change one feature at a time – headline, text, thumbnail, call-to-action. Do not run completely different combinations of these things, as it will make harder for you to understand which feature exactly performs the best (imagine you would test audience interests, with having more than 1 interest in one adset);

What is the recommended structure then?

While, again, there is no universal “secret sauce”, this method is the most reasonable we have come up with in terms of efficiency:

– Video 1+Text 1;

– Video 2+Text 2;

– Video 1+Text 2;

– Video 2+Text 2.

2. Facebook Split Tests.

Honestly, we do not use Facebook split tests often, but many of our clients have reportedly shown great results with them.

Indeed, when it comes to testing creatives specifically, it’s quite efficient.

In a nutshell, during the test, Facebook splits your traffic into several groups of audiences that do not overlap. In case of creatives, you may choose an option to cross-test different images, vids, thumbnails, etc in this test run.

Then, every adset’s results are evaluated by Facebook based on the objective of your campaign, and you get the creatives with best results.


– Your audience ABSOLUTELY NEEDS to be large. Otherwise, the results of your tests will not be significant;

– You have to pay for Facebook A/B testing;

– Since Facebook splits the viewers’ pool into several baskets, you pay x-that many baskets for your testing comparing to when you run just 2-3 ads for the whole audience.

How do you optimize ads during Facebook Split Tests?

a. Again, test ONLY one feature/variable in one run – keep it simple so that you could interpret the test results correctly;

b. Limit your split test to A/B testing Although you will have an option to get up to about 5-6 different adsets, testing ONLY 2 has proven to be the most effective option in terms of results/spend;

c. Think about the required sample size and time needed to get significant results. You should know your desirable conversion rate already, so you might know how much you can afford to spend on testing to get significant results. For getting the required sample size, you can use this web tool.

d. Although you should test one feature at a time, differences in samples must be significant. Say you want to test different headlines – do not just change one word, change the whole thing.

For more info about retrieving and interpreting your results, you may find this Facebook page pretty useful!

3. Dynamic Creatives.

This is, so to speak, an automatic A/B/C/…/Z test run AUTOMATICALLY by Facebook machine learning.

You only need to put in several videos/images, headlines, texts, and Facebook will do the work for you.

I am sure you’ve all heard of it and probably tried as well. So you might tell me now “Hey! DC is not testing!” Yes, not exactly – you don’t retrieve any results from it. But Facebook does and it uses it, as it promises, better than you would.

Yet, we are limited to some of the parameters for dynamic creatives.

For those of you who have yet tried dynamic ads, this Facebook guide will definitely help you.


– You are completely reliant on Facebook and do not see the results of tests – only the final performance of the whole thing.

How do you optimize dynamic creatives?

a. Use your common bidding strategy – there is no need to allocate more to dynamic creatives for proper testing. Machine learning will test it on the fly.

b. Stick to broad targeting – let Facebook machine learning enough data to optimize.

c. Your texts/videos/headlines should be universal. As you know, Facebook mixes everything up, so you must be sure that all the variations are substitutable.

d. Monitor the performance – Facebook will want to run combinations with the highest CTR, which might though not give you good CPA.

To Wrap It Up

So, the three methods to use for testing your creatives are:

1. Manual testing – with 2-5 ads in an adset.

2. Facebook Split Tests.

3. Dynamic creatives.

We want you to remember that EACH CASE, EACH PRODUCT, EACH CAMPAIGN should be approached as a unique one. What is working best for one thing might be useless for another. So, it’s always good to have as many tricks as possible under your sleeve.

So what are your thoughts on this one? Have you been using all three methods in your practice? Which one do you like the most? Can you think of other cons or optimization recommendations for them? Tell us about it!

Let’s share our ideas and boost our sales together!

Every month I choose a few entrepreneurs to personally help them scale their Facebook ads and businesses.

Watch the free training and apply here: