Test three to five distinct creative concepts at a time, each with two or three hook variations. That is enough for Meta's delivery system to find a winner without splitting your budget so thin that nothing exits the learning phase.
We supply test creative to DTC ad accounts every month, so this question lands in our inbox constantly. Here is the structure that keeps working.
A concept is a genuinely different idea: a customer testimonial, a product demo, a problem-led piece to camera. A variation is the same idea with one element changed, almost always the opening seconds. The distinction matters because concepts teach you what message converts while variations squeeze more from a message you have already proven. Accounts that only make variations plateau; accounts that only make concepts never compound their learnings.
Because creative is the targeting now. Meta's delivery system decides who sees your ad largely from the creative itself, which means the ad is the audience selection. More distinct concepts give the system more ways to find pockets of buyers. In practice, creative quality and variety drive the majority of account performance, far ahead of bid strategies and audience settings.
Fatigue is the decay in performance as the same people see your ad repeatedly: frequency climbs, CTR slides, CPA creeps up. For most SME budgets it arrives within four to six weeks of scaling a winner. The fix is planned succession: new variations of the winning concept ready before the decline, not commissioned after it.
Build a simple grid: concepts down one side, hooks across the top. Five concepts with three hooks each gives fifteen assets from what might be a single shoot day. Launch concepts in one campaign with even budgets, kill clear losers after enough impressions to judge, then promote winners to scaling campaigns and start cutting variations. Document what won and why; the pattern across cycles is your real asset.
A workable floor is enough spend for each asset to gather meaningful data within a week, which for most UK SME accounts means concentrating tests rather than running everything at once. If budget is tight, test three concepts instead of five but never test one. A single creative is not a test; it is a hope.
This is the real constraint. Fifteen assets per cycle sounds impossible if you produce videos one at a time at £300 each. It becomes routine with batch production: one planned shoot capturing multiple concepts, each with several scripted hooks, exactly how our Content Sprints are designed. Pair commissioned UGC-style creative with edited cuts of existing footage and the testing matrix fills itself.
If your account is stuck recycling two tired ads, we can fix the supply problem. Talk to us about building your testing pipeline.
Give each creative enough impressions to be statistically meaningful, typically a week at moderate spend or roughly 1,000 impressions minimum per asset. Judging on day one rewards luck rather than quality, while letting losers run for weeks burns budget. Set the review date when you launch.
For clean reads, yes. A dedicated testing campaign with even budget distribution stops Meta from starving new creative in favour of established winners. Promote proven assets into your scaling campaigns once they have earned it.
Judge top of funnel with hook rate (3-second views over impressions) and hold rate, mid funnel with CTR, then bottom line with cost per acquisition. A creative with strong watch time but no clicks has a message problem; strong clicks with no conversions points at the landing page or offer.
Meta's own guidance suggests keeping ad sets to around six or fewer active creatives so spend is not spread too thin. In practice three to five well-differentiated assets per ad set balances delivery learning with meaningful data per creative.
.avif)
How much content will I get?
What content can you create?
Where can content be shot?
What if I need revisions?
Never done a Content Sprint?
Unhappy with the content?
Need models, locations or props?
.svg)
.svg)