Facebook ad testing guide

November 11, 2020

How to test your ads on Facebook

Look at any great work of art. Then look a bit deeper. Under the paint of the Mona Lisa there’s a whole other painting – a first draft that didn’t test well with Da Vinci’s patron. Shakespeare’s manuscripts are full of crossings out and rewrites. Charles Dickens even changed the entire ending of ‘Great Expectations’ when he realised that what he had planned was not what his audience wanted.

The point is, the greatest, best-performing works in history came about after a lot of testing. These writers and artists tested out their drafts, then edited what didn’t work, then tested again, until they got a product good enough to withstand the ultimate test – of time.

Obviously we’re not all Dickens and Da Vincis, but testing works just as well for marketers as it does for artistic geniuses. In the next in our series on Facebook ads, we’ll teach you all about Facebook ad testing.

Facebook ad testing

By testing your Facebook ads, you take out a lot of the risk associated with them. That’s important, because margins for Facebook ads are very small. You don’t want to pour a whole load of time and money into ads that people don’t even notice as they scroll on by.

If you test your Facebook ads properly, your campaign will hit the ground running. You’ll be able to optimize your ads, increasing performance with every test and tweak. You’ll give your campaign the very best possible chance, from the get-go.

That being said, testing Facebook ads can be confusing. There are a lot of variables to consider, and at times the testing process can seem like a long, hard slog.

How can you get the best out of Facebook ad testing? What ad testing strategies are efficient, effective, and will give you the clear results that you need?

Let’s take a look:

Building the perfect Facebook ad testing strategy

1 – Pre-optimize your ad sets

Optimizing through testing is effective, but it also costs time and money. You can slash your spend by optimizing before testing. There are lots of tools and platforms out there which can make sure that you’re only putting the best to the test.

For ad creatives, Datasine takes away the guesswork.  Our AI-based platform combines detailed creative data with both Facebook’s and your own campaign insights, to accurately predict which images will perform best with your audience. In a recent experiment we ran, Datasine’s AI predicted the best performing Facebook ads with 99.9% accuracy.

By using these tools and platforms while creating your ads, you can make sure that you’re only bringing the very best performing ads to the table. You can then spend your ad-testing budget on perfecting rather than fully optimizing your Facebook ads.

2 – Work out an ads testing budget

Your Facebook ads testing budget will depend a lot on your overall Facebook campaign budget, and the calculations you put into working it out. You can read more on coming up with an accurate Facebook ads campaign budget here.

As a general rule, testing is around 10% of a campaign budget. However, this will vary depending on your campaign specifics.

Check back with your goals, objectives, audience data, and past performance insights to work out how much to spend on Facebook ad testing.

3 – Work out what to test

To draw insights from testing, you need to have solid variables to study. There’s no point in discovering that one ad set performs better than another if you don’t know why.

This means that you need to establish some variables to test against one another. 

For example, you might pit two nearly identical ads against one another – one with a red CTA and one with a green CTA. If green performs better, you can draw some confident insights about why this ad was more popular than the other.

Variables to test may include:

  • CTAs
  • Logo placement
  • Text fonts
  • Copy tone
  • Keywords
  • Audiences (which audience does a particular ad perform better with?)
  • Ad placement

4 – Split or multiple testing?

Facebook will let you either split-test your ads, or test multiple variations at once:

Split Testing

We’ve gone into split-testing in detail here, but basically it involves pitting variables against one another in a controlled way.

When split testing, you specify a variable to test (CTA) for example, and an audience to test it with. Facebook will then divide that audience into two random groups. Your ads will be shown evenly to the two sections of your audience (with no overlap). These ads will be identical, apart from the variable you’re testing. For example, one audience will see a red CTA, and the other will see a green one.

Facebook will then measure the performance of each ad, based on your campaign’s objectives.

Split testing is great for making solid decisions on important variables – but it does take a lot of time. If you’ve optimized as far as you can, and aren’t getting statistically significant insights from split-testing, it might be time to try out multiple testing.

Multiple testing

With this method, you set up several creatives in one ad set, each with variations in things like ad copy, imagery, and so on.

You can then use Facebook’s ad analytics to determine which ads in the set are performing best, and steadily delete lower performing options.

This option is not as ‘scientific’ as split-testing. Facebook will prioritize higher-performing ads, so you don’t get even distribution of each variable across your audience. However, it will give you a reasonable read on which ads are working for you, and which are not.

This method works best for super-tight optimization. It’s not great for serious insights, but it will help you to keep your ad budget in line, and to ensure that your ad sets only contain the very best of the best.

However, neither method will be as effective as it should be if you’re testing with the wrong people. Which brings us to Step 5…

5 – Identify your test audience

If you’re at the testing stage of your campaign, you’ve probably already identified your target audience. However, it’s worth double-checking that you’re showing your ads to the right people. 

It sounds obvious, but it’s worth saying: you should be testing your ads with the audience you want to see the ultimate campaign. You can’t learn anything about how your target audience will respond to your ads if you’re not involving them in the testing process!

We’ll be delving into how you can use Facebook to identify and target audiences in the next in this series – so stay tuned!

6 – Run your tests – and reap the rewards!

You’ve got your ads together, you’ve established a testing budget, you’ve set up your variables, you know how you’re going to test, and you know who you’re testing with. Now, it’s time to let Facebook do its thing and run your tests.

Facebook will test your ads for as long as you tell it to, and it will present results as clearly as it can based on your stated goals and objectives. Your task is to draw insights from the testing data it gives you, and tweak your ads according to those insights.

To get the very best out of your Facebook ad testing, here are some tips:

  • A/B test two versions at a time. Facebook does give you the option to A/B test multiple ad versions at once. That’s tempting – especially when you’ve got a lot of variables to get through. However, if you’re trying to draw the kinds of insights that split-testing gives, we strongly recommend that you stick to two versions and one variable at a time. This makes things a whole lot clearer when it comes to your final analysis.
  • Give yourself enough time. Work out how much testing time you’ll need for statistically significant results. This will depend on things like your testing budget, your sample size, and your overall goals. As a general rule, we recommend running tests for at least 7 days.
  • Test old ads as well as new ones. Ads, like everything else, have a sell-by date. People get bored with seeing the same ad over and over again. Make sure to regularly test and re-optimize (or retire) even your highest-performing ads to avoid ad-fatigue.

Start as you mean to go on with Datasine

You can do a huge amount of testing in one go with Datasine. Datasine’s intelligent platform combines data and creativity to predict the very best performing creatives for your audience.