Why good creatives stop working
A creative that worked for six weeks does not become bad overnight. It becomes expensive — and the account notices only after the budget has already paid for the discovery.
Fatigue is an auction problem, not a taste problem
The common explanation — people got bored — is only half true, and it is the less useful half. What actually happens is that the pool of users who respond to a specific message gets consumed first. The auction keeps serving the ad, but each additional impression reaches someone slightly less likely to convert. Frequency rises, the marginal buyer gets more expensive, and the average conversion rate sags long before anyone on the team feels tired of the video.
This matters because it tells you what to replace. If the pool for one message is exhausted, producing a prettier version of the same message will not open a new one. You need a different claim, a different objection, or a different audience belief — not a new colour grade.
The three decay curves worth tracking
Fatigue is not one metric moving. It is a sequence, and the order tells you what is wrong.
- Hook rate falls first. Three-second views over impressions drops while conversion rate holds. This is attention fatigue — the opening is no longer novel, but the argument still convinces the people who get past it.
- Conversion rate falls next while hook rate holds. This is audience fatigue — you are still catching attention, but the remaining audience does not want the offer as framed.
- CPM rises with both stable. This is competitive or seasonal pressure, not fatigue at all. Replacing the creative here wastes a working asset.
If you only watch CPA, all three look identical — and two of the three fixes will be wrong.
Measuring half-life instead of guessing
Give every concept an ID that survives every edit, cut and aspect ratio. Then chart conversion rate against cumulative impressions rather than against calendar days. Days are contaminated by budget changes; impressions are not. What you get is a decay curve per concept, and after a dozen concepts you get something far more valuable: a house average.
Once you know that concepts in a given market typically hold for, say, 1.4 million impressions before conversion rate drops ten percent, replacement stops being reactive. Production is scheduled against the curve, and the next validated winner is already sitting in the testing block when the incumbent starts to slide.
Refresh, rebuild, or retire
Three responses, and choosing correctly is most of the saving.
- Refresh — same angle, new execution. Correct when hook rate decayed but conversion rate held. Cheap: new opening, same argument, same proof.
- Rebuild — same audience, new angle. Correct when conversion rate decayed. The claim needs to change, which means going back to the angle map, not the edit suite.
- Retire — the angle is spent for this market. Archive it with its performance history so nobody rediscovers it in nine months and pays to learn the same lesson twice.
The structural fix
Accounts that never suffer acute fatigue have one thing in common: a permanent testing block with a fixed budget share, running whether or not there is a crisis. It looks inefficient in a good month because a portion of spend is deliberately going to unproven creative. It is the cheapest insurance in the account, because the alternative — scrambling for a replacement while CPA climbs — costs more in a single bad fortnight than a year of disciplined testing.
Fatigue is not a failure of creativity. It is the predictable cost of a message being successful, and like any predictable cost it should be planned for rather than reacted to.
Takeaways
- Track hook rate and conversion rate separately — the order they fall in tells you what to fix.
- Chart decay against cumulative impressions, not calendar days.
- Keep a permanently funded testing block so the replacement exists before the incumbent dies.