Guessing which ad will work is unreliable, even for experienced marketers. Testing replaces opinion with evidence. Over time, a testing habit builds a library of insights about your audience: which hooks, formats, benefits and styles they respond to. The Academy's glossary explains creative testing.
What to test first
Start with elements that make the biggest difference:
| Element | Example variations |
|---|---|
| Concept or angle | Price, quality, convenience, social proof |
| Hook | First three seconds or first line |
| Format | Video, carousel, static image |
| Style | Polished versus user-generated look |
| Offer | Discount, free delivery, bundle |
| Call to action | Shop now, learn more, book |
Test big ideas before small details. Changing a button colour rarely matters as much as changing the main message.
Writing a hypothesis
Before testing, write what you expect and why: we think a customer testimonial video will beat a product demo because reviews show people trust other customers. A hypothesis makes results easier to learn from, whatever the outcome.
Testing methods
- Split testing (A/B): the platform divides the audience so each person sees only one version. Cleanest results.
- Multiple ads in one ad set: the platform spends more on early winners. Fast, but less controlled.
- Dynamic or flexible creative: you upload several images, videos and text options, and the platform combines them. Good for finding combinations, harder to learn specific lessons.
Use split tests for important questions and multiple ads for ongoing everyday optimisation.
Keeping tests fair
Change one main thing at a time, run versions simultaneously, use the same audience and budget, and give each enough time and results to judge. Avoid ending tests after a day because one version looks ahead early.
How much budget and time
Tests need enough results to be meaningful. For conversions, that may mean dozens of actions per version. Small budgets can test higher-volume metrics such as click-through rate or cost per landing page view first. Run tests for at least a week to cover daily variation.
Reading results
Judge on the metric that matters for the objective, such as cost per purchase for sales campaigns. Look at supporting metrics too: hook rate for video, CTR, conversion rate. If results are close, the difference may be noise; run again or test a bigger change.
Scaling winners
When a version wins clearly, move more budget to it, use it as the base for the next test and apply the lesson to other ads. Keep testing, because winners fatigue over time and audiences change. The Academy's guide to ad fatigue explains why.
Platform considerations
Creative that suits each platform tends to perform better. TikTok's documented organic signals are completion rate and rewatches, shares, comments, likes, interest signals from watch behaviour✓, so ads that feel like native TikTok videos and hold attention usually beat polished commercials there. The Academy's guide to social media ad creative covers what works by platform.
Recording learnings
Keep a simple test log: date, hypothesis, versions, result and lesson. Over months, the log becomes a playbook for future campaigns and helps new team members avoid repeating old tests.
Testing on small budgets
Small advertisers can still test. Run two or three ads in one ad set, compare cost per result over a few weeks and keep the best. Use organic posts as a free testing ground: content that performs well organically often makes a strong ad.
Hook rate and hold rate
For video ads, two measures help diagnose creative. Hook rate is the share of impressions where people watched the first few seconds; hold rate is the share who kept watching further. A low hook rate means the opening is not stopping the scroll, so test new first frames, text or movement. A good hook rate with a poor hold rate means the middle loses people, so tighten the edit or bring the main point forward. The Academy's guide on how to hook viewers in the first 3 seconds covers openings in detail.
Creative testing with creators
Creator-made content often brings fresh angles your team would not think of. Brief several creators on the same product, test their videos against each other and against your own creative, and keep the winners. Make sure usage rights cover paid use and that paid partnerships are labelled.
A worked example
A meal kit company tests three concepts: saving time, eating healthily and trying new recipes. Each is a short video with the same offer. Saving time wins clearly on cost per order. The next test compares three hooks for the saving-time video. The best hook cuts cost per order further. The company records both lessons and builds its next month's ads around time-saving messages.
Testing landing pages too
An ad can win on clicks but lose on sales if the landing page lets people down. When a creative test shows strong click-through but weak conversions, check that the page matches the ad's message and loads quickly on phones. Sometimes the best next test is the page, not the ad.
Common mistakes
- Testing tiny details before big ideas.
- Changing several things at once.
- Ending tests too early.
- Judging on the wrong metric.
- Never recording what was learned.
Learning to test
The Academy's paid social module in the third level covers creative testing, campaign structures and optimisation, and the analytics modules cover reading results. The free assessment shows where you would start.
