Build a Meta creative testing process around your business goal
Our creative testing workflow separates learning from scaling, groups variations around a clear concept, and evaluates them against an event that matters to the business. The structure helps organize decisions; its budget and variation counts are operating heuristics to adapt to your account.
By NewForm · Updated
Key takeaways
- Give testing and scaling distinct purposes so new ideas receive attention and established winners can be evaluated at scale.
- Group purposeful variations by concept and allocate a testing budget you can afford.
- Choose a revenue event or a proxy whose relationship to revenue you have checked.
- Treat early screening, budget multiples and variation counts as decision aids rather than guarantees of significance.
Separate testing and scaling into distinct campaigns
Our videos call this division “learn or earn”: use one part of the account to investigate new concepts and another to scale promising assets. The goal is to make it clear which decisions concern learning and which concern current performance.
When an execution shows promise in testing, the described workflow moves it into scaling and evaluates what happens there. Success in the test is evidence to act on, not a guarantee that performance will transfer unchanged.
Structure tests concept-by-concept using ABO
The videos favor ad-set budgets for concept testing so the team can allocate spend deliberately. Each ad set contains a concept and multiple executions, allowing comparisons within a defined message or persona.
This gives the team control over the test budget; it does not ensure equal delivery to every variation or remove audience and auction differences. The recommendation to group variations is an organizational choice, not a claim that a particular structure is required by Meta.
Anchor creative tests to cold audiences and down-funnel signals
The source videos warn about choosing creative winners using a metric that does not match the business goal. A low install cost can be misleading if those users do not subscribe. Likewise, evidence from people who already know the brand may not answer whether the ad can attract new customers.
For prospecting, evaluate creative in the context of acquiring new users. Prefer a purchase, subscription or another revenue-linked outcome when it arrives often and promptly enough to be useful. When using an earlier proxy, check its relationship to the later outcome.
One suggested subscription-app proxy is a trial that remains active after the first few hours. That is a hypothesis to validate against later paid conversions in your own data. It should not be assumed to predict revenue for every app.
Budget heuristics and variation testing
The videos describe a testing budget around ten times the target cost per desired action per day at the ad-set level. For a $20 target, that would mean $200 a day. This is the team’s operating heuristic, not a platform requirement or proof of adequate statistical power.
They also favor multiple executions per concept and discuss roughly ten as a working quantity. The underlying lesson is to avoid rejecting a sound idea because one execution failed. Shared creative and delivery conditions mean those attempts are not independent experiments.
Set limits before the test, use leading indicators to investigate expensive underperformers, and decide whether another variation is justified by what you have learned. Stopping a weak concept is a budget decision under uncertainty, not proof that no execution could work.
Adapted from NewForm’s original videos on creative strategy and paid social.
Explore NewForm Framework →