Ads

How to test ad creative: find out what works with A/B testing

Steps to set up image, copy and call-to-action tests correctly in Instagram and Facebook ads: one variable, enough data, a success metric and interpreting results.

By Sercan Horkal · · 6 min read

What affects ad performance most is often not the target audience but the creative: image, headline, copy and call to action. But you cannot find what works by guessing; you need regular, properly set up tests.

Why test?

It is hard to predict what people will click. Whether a message or an image works is only visible in data. Accounts that test systematically learn over time which message, format and tone work with their audience and carry that into the next campaign.

Rules for setting up a test

1. One variable at a time

If you change the image and the copy together, you cannot tell which made the difference. Change one thing per test: only the image, only the first sentence, only the call-to-action button.

2. Write a clear hypothesis

“A product-in-use image brings a higher click-through rate than a studio shot.” Writing the hypothesis in advance makes the result easier to interpret.

3. Decide the success metric beforehand

Which indicator will you look at? In a sales campaign, click-through rate is not enough; conversions or acquisition cost should be the basis. In upper-funnel tests (awareness), measures such as view rate can be used.

4. Run under equal conditions

The same audience, the same budget, the same period. Platforms’ A/B test tools do this by splitting the audience and preventing overlap; in a manual test, comparing two ads in the same ad set may not be fair because the algorithm steers budget to one.

5. Wait for enough data

Do not decide after a few clicks. The test’s duration and budget should be enough to see a meaningful difference. A test closed too early gives a wrong result based on chance.

What can you test?

  • Image: Product alone or in use, with or without a human face, colour and composition
  • Format: Single image, carousel, video, short video
  • First sentence and headline: Benefit-focused, question, number, problem-solution
  • Call-to-action button and offer: “Learn more” versus “Get a quote”, free shipping versus a first-order advantage
  • Video: The first three seconds, length, captions
  • Landing page: The same ad, a different page

Interpreting the results

  • Scale the winning creative, but also note why the loser lost.
  • Record what you learn. Conclusions such as “creatives with customer reviews work better” are the basis of the next production.
  • Do not overstate small differences. A difference of a few percent may not be statistically meaningful.
  • Watch for ad fatigue. Even a winning creative loses effect over time; performance falls as frequency rises.

A routine of continuous testing

Plan at least one or two tests a month. A simple way: each month, try a new idea against the current winner. Keep a test log: date, hypothesis, result, conclusion.

Common mistakes

  • Changing many things at once.
  • Deciding too early.
  • Looking only at click-through rate. An ad that gets clicks but does not sell is wasted budget.
  • Not recording test results.

For creative production and a test plan, see our Meta Ads service.

Want help with this?

Let’s talk

Get a quote

Tell us about your project.

Write your goal in a few lines and we will reply within 1 business day.