A Creative Testing System
Creative wears out and new winners are rare, so successful accounts test constantly. A simple, repeatable system finds winners faster and keeps testing from disrupting campaigns that work.
Separate testing from scaling
Run new creative in a testing campaign with its own modest budget, often using ad set budgets so each concept gets spend. Move clear winners into the main scaling campaign. This protects the main campaign’s learning.
Test concepts, then variations
- Test three to five different concepts (angles, formats, messengers).
- Take the winning concept and test variations: hooks, length, text, thumbnails.
- Promote winners; retire the rest.
Deciding a winner
Decide the metric before launching: cost per lead, cost per purchase or ROAS. Give each concept enough spend to collect meaningful results (as a rough guide, at least two to three times your target cost per result) before judging. Use Meta’s A/B test tool when you need a clean, statistically valid comparison.
Spotting fatigue
Rising frequency, falling CTR and rising cost per result together signal fatigue. Refresh the hook or angle before results collapse.
Pro tips
- Plan a steady flow of new concepts each month.
- Name ads with the concept and hook so reports are readable.
- Keep a library of winners and why they won.
Common mistakes
- Testing new ads inside the main scaling campaign and resetting its learning.
- Calling a winner after a few hundred rupees of spend.
- Running one winning ad until it collapses.
Test concepts in a separate campaign, then variations of winners, decide on a pre-set metric with enough spend, promote winners and refresh before fatigue.