Level Marketing / Articles / Paid ads
A/B testing ad creatives: how to run it
You can't know in advance which ad will perform best — even an experienced specialist is only guessing. That's why professional advertising runs on A/B tests: several versions launch at once and the numbers pick the winner. Here is what to test, how and for how long.
What an A/B test is
An A/B test compares two or more versions under the same conditions, with only one element different. For example, the same audience and budget but a different opening for the video. Then any difference in results comes down to that element.
What to test
| Element | Example | Impact |
|---|---|---|
| Hook (opening) | Start with a problem or a result | Very high |
| Offer | "10% off" or "Free measurement" | Very high |
| Format | Video or carousel | High |
| Copy | Short or detailed | Medium |
| Audience | Interests or Advantage+ | Medium |
| Button | "Learn more" or "Send message" | Low |
Start with the elements that matter most: the hook and the offer.
How to run it: 5 steps
- Write a hypothesis. "A video that opens with the result will lower cost per lead."
- Pick one variable. Keep everything else identical.
- Launch. In Meta you can use the A/B test (Experiments) tool or run separate ad sets with equal budgets.
- Collect enough data. Wait until each version has a few dozen results (clicks, messages or leads) and at least 4–7 days have passed.
- Conclude and record. Keep the winner, switch off the loser and log the result.
How to pick the winner
The main metric is cost per lead (CPL) and, where possible, conversion to sale. CTR and cost per click are secondary: lots of cheap clicks with no leads means the creative is "interesting" but doesn't sell. More on CPL in our CPL guide.
A test log
Record every test in a simple table:
| Date | Hypothesis | Versions | Result (CPL) | Conclusion |
|---|---|---|---|---|
| Week 1 | Hook: problem vs result | A / B | A cheaper | Opening with the problem works |
After a few months this table becomes a valuable knowledge base about what works for your audience.
Common mistakes
- Changing several things at once.
- Stopping a test after 1–2 days.
- A budget too small for any version to collect enough results.
- Finding a winner and not starting the next test — creatives wear out over time.
Conclusion
A/B testing moves your advertising from guesswork to numbers. At least one test a week is the simplest way to keep improving. For creative ideas, see 7 types of video ads.
