Hook, format, and angle mapped into a sequenced plan — with sample sizes and decision rules set before launch.
Build a creative testing plan for Meta next quarter. We can produce about 6 new assets a month.
Nine tests across the quarter, hooks first since they are cheapest to produce and move the most. Each test needs nine days at current conversion volume. Two combinations were dropped because the account cannot power a three-way split.
A creative testing template has to stop you from testing everything at once. This one lays out the hook, format, and angle combinations worth running, sequences them so one variable moves at a time, and sets the sample size and decision rule before the first dollar is spent.
A matrix of the creative variables you could change — hook, format, angle, offer, proof — turned into an ordered test queue. Each row states the variable under test, the control, the budget, the duration, and what result triggers which decision.
By holding everything constant except the one thing being tested. If the hook and the format change together, a win teaches you nothing transferable — and most creative testing produces exactly that kind of unusable result.
Until each variant has enough conversions to distinguish signal from noise, which for most accounts means a week or more rather than two days. The template computes the requirement from your conversion rate and declines to schedule tests you cannot power.
By keeping a queue. When a winner starts decaying — rising frequency, falling click-through — the next test in the sequence is already briefed, so replacement is scheduled rather than improvised under pressure.
The agent reads current creative performance from Meta or Google Ads to establish the control and the conversion rate used for sizing, then builds the test queue. Each scheduled test can hand off to the creative brief template for production, and results flow back into the fatigue analysis in the platform audits.
The agent reads current creatives and conversion rate to set the control and sizing baseline.
Only include hooks, formats, and angles you can actually make this quarter.
Each test shows its variable, control, budget, duration, and decision rule.
Hand row one to the creative brief template; the rest of the queue waits behind it.
Isolates one variable per test so results transfer
Computes sample size from your data and refuses underpowered tests
Decision rules are set before launch, not negotiated after
Maintains a queue so fatigue replacement is scheduled
One variable, one control, a pre-computed sample size, and a decision rule written before launch. Changing two things at once produces a result you cannot apply to the next round.
As many as your conversion volume can power — often two to four variants against one control. Splitting budget across ten variants usually means none reaches significance.
Hook. It typically moves performance most and is cheapest to produce, so it earns the earliest learning. Format and angle come after you know which hook lands.
When it has enough data to conclude it loses to the control, or when its cost per result has climbed steadily at rising frequency. Both conditions are written into the decision rules up front.