How many creatives to test and when to kill losers.
Inputs
Test plan
Spend / creative / month
—
Days to decision
—
New creatives / month needed
—
About this calculator
Creative testing is the engine of paid social performance. Without consistent creative refresh, accounts decay. With proper testing — 6-12 concurrent creatives, fast kill decisions, scale winners aggressively — accounts compound winners over time. Most underperforming accounts have a creative pipeline problem, not a targeting problem.
The math: monthly budget ÷ concurrent creatives = spend per creative. Compare against the spend needed for a confident decision (~50 conversions × CPA). If spend per creative is below decision threshold, you\'re testing too many at once and never getting clean signals. Reduce concurrent count or increase budget.
The discipline of decision-making is harder than the math. Operators tend to keep losing creatives running too long ("maybe it\'ll turn around") and scale winners too cautiously. The math doesn\'t support either — kill losers fast, scale winners aggressively, fund the production pipeline that produces both.
Pair with the Ad Fatigue Timeline (when to refresh winning creatives) and the Creative Fatigue Detector (real-time signal that current creative is dying). Most successful DTC brands maintain a 15-25 piece creative library at any time, with weekly production refreshing 20-30% of the library.
Frequently asked questions
How many creatives should I test at once?
For typical DTC: 6-12 active creatives, refresh 2-3 weekly. Below 4, single underperforming creative drags down account. Above 20, dilutes spend across underperforming variants. Sweet spot: enough variety for meaningful testing, not so much that no single creative gets enough spend to learn from.
How much budget per creative for a fair test?
Each creative needs to deliver enough purchase events to reach significance — typically 50-100 conversions. At $20 CPA, that's $1,000-2,000 spend per creative. Below that, results are noise. Test with at least $500 per creative before declaring winners.
When should I kill a losing creative?
After 50+ purchase events with statistically lower CPA than top performers. Or after 7 days of running without conversions if other creatives are converting at expected CPA. Don't kill creatives in the first 3 days — Meta's algorithm is still learning during the initial spend.
How do I scale winners?
When a creative hits 100+ conversions at lower CPA than account average, scale by 20-30% per day. Above 30% daily increase, Meta resets learning phase. Watch frequency — even winning creatives fatigue at 5-7 weekly impressions.
How fast should I refresh creative pool?
2-3 new creatives per week minimum. Faster for high-volume accounts (5+/week). The math: at 3-4 week creative lifespan, you need new winners ready to replace fatigued ones continuously. Falling behind on the production pipeline = account performance degrades.