A swipe file can contain hundreds of visually different ads, yet many teams still test the same product image with a new headline, colour or button. That is design iteration, not a meaningful creative test. The 30 formats below work better as a hypothesis library. Choose a small set that fits the buying situation and test why someone would stop, trust and act rather than producing every format at once.
Group 1 Hooks that frame the problem
- Clear action hook Show the next useful action immediately
- Question hook Open with a question the intended buyer genuinely asks
- Did-you-know hook Use a surprising fact only when the evidence is traceable
- Myth versus fact Correct a misconception that changes the buying decision
- Data hook Include the source, date and measurement context
- Alternative breakdown Compare options fairly and explain who each one suits
Group 2 Founder and team-led content
- Founder introduction State the audience and problem in one useful sentence
- Journey to result Show the starting point, decisive actions and limits of the outcome
- What I wish I knew Share a genuine change in judgement without inventing a lesson
- Walking or in-car monologue Best for a short point delivered naturally
- Whiteboard explainer Turn a complex service into three to five visible steps
- Behind-the-decision Explain why the team kept or rejected a product choice
Group 3 Concrete social proof
- Narrative testimonial Preserve the customer's meaning and secure permission
- Customer-call excerpt Publish only material approved for public use
- Comment response Build a complete answer from a real question
- Recurring DM Anonymise repeated questions rather than fabricating screenshots
- Street interview Explain the sample and avoid treating a few answers as market proof
- Before and after Keep conditions comparable and disclose the timeframe
Group 4 Educational formats that reduce decision friction
- Knowledge drop Solve one defined problem at a time
- Ingredient or feature breakdown Explain function, limits and suitability
- Thirty minutes versus thirty seconds Use the long version for process and the short version for one conclusion
- Expert explanation Verify credentials, quotations and regulatory boundaries
- Checklist Give the viewer something they can apply immediately
Group 5 Scenario and entertainment formats
- Point of view Recreate a recognisable moment in the user journey
- Two-person dialogue Let a real objection and answer unfold naturally
- Challenge format State the rules upfront and do not manipulate the result
- Faceless footage with voiceover Useful for workflows, tools and demonstrations
- Story-led sketch Base the tension on a problem buyers genuinely encounter
- Months later Show the longer-term change and what happened in between
Group 6 Change the production language, not the facts
- Green-screen breakdown Explain a page, product or reference while it is visible
- Collage Combine details that establish a real use context
- UGC unboxing Disclose gifts, commissions and commercial relationships
- Simulated podcast Treat it as a creative format, not a fake third-party interview
- AI avatar Useful for multilingual testing when pronunciation and disclosure are reviewed
- Message-thread or news format Label dramatisation clearly and never invent users or media endorsement
How to structure one testing cycle
Start with one audience problem and design three to five meaningfully different hypotheses. Keep the objective, landing page and major delivery conditions reasonably stable. Record each creative's opening, message angle, proof mechanism, format and call to action. Review attention, clicks, landing-page behaviour, qualified leads and sales quality together. A large asset count teaches little when every asset tests the same idea.
WEPR helps brands connect creative research, scripting, Meta activation and conversion review into a repeatable testing process. We first assess the offer, market, asset rights and measurement readiness, then decide which formats are worth producing.
Should a brand test all 30 formats?
No. Select three to five based on the audience problem, buying stage, production constraints and compliance requirements. Expand only after a useful signal appears.