In brief
An A/B test compares two versions of the same campaign while changing only one thing: the visual, the text, the audience or the placement. The versions run at the same time, on audiences that don’t overlap, with the same budget.
Why it matters
Without a test, you guess. With a test, you learn what works and can reuse the conclusion in later campaigns.
The steps
- Set a question and a hypothesis: “Does a video lower the cost per lead compared with an image?”
- Choose a single variable: visual, text, audience, placement or optimization.
- In Ads Manager, use the A/B test tool (or duplicate the campaign).
- Split the budget equally.
- Choose the comparison metric before launching: cost per result, for example.
- Set a duration: several days, often 7 days or more, to get enough results.
- Don’t change anything during the test.
- At the end, read the result: is the difference clear enough and based on enough data?
- Apply the winning version and write down the conclusion.
Pitfalls to avoid
- Changing two things at once.
- A budget that is too low: too few results to conclude.
- Stopping as soon as one version looks better after one day.
- Audiences that overlap: the versions compete with each other.
- Choosing the metric after seeing the results.
Key takeaways
- One variable, one hypothesis, one metric.
- Equal budget, sufficient duration.
- Document your conclusions.
Practice in TRIALO
Create two identical campaigns with a different visual: publish them, compare the cost per result over the Maximum period, then write your conclusion in two lines.