Creative Fatigue Leading Indicators to Watch Before CPI

This publication made the case back in January, in Detecting Creative Fatigue Before It Shows Up in CPI, for catching fatigue through leading indicators instead of waiting on cost to move. That piece established the principle. What it did not spell out is that the indicators do not all move at once, so a team checking them in the wrong order, or checking only one of them, still catches fatigue later than it needs to while believing it is ahead of the problem. The indicators run in a sequence. Reading them in that sequence is where the time advantage actually comes from.

The sequence, fastest to slowest

Hook rate and early-frame click-through decay first. The audience segment most likely to disengage from a familiar creative is the same segment paying most attention to the opening seconds, the group for whom novelty was doing most of the work. Impressions per mille, or IPM, follows, as an ad's overall pull weakens across a wider audience rather than only among the most engaged early viewers. Conversion rate on click or install lags IPM, because a shrinking but still-committed pool of genuinely interested viewers keeps converting reasonably well even as the top of the funnel thins. CPI moves last of all. It is a blended output of everything happening upstream, diluted further by whatever the algorithmic buying platform does to compensate for a weakening creative, usually by shifting spend toward better-performing audience segments on its own. A team watching CPI alone is reading the last domino to fall.

Why the order matters more than the list

Knowing that four indicators exist is not the same as knowing which one to check first on a Monday morning. A team with limited time for a weekly creative review should read hook rate decay before anything else, because a decline there is the earliest reliable signal available and it buys the longest runway to act, whether that action means a new hook variant, a frequency cap adjustment, or an early flag to the production pipeline that a replacement concept has to move up the queue. Checking CPI first, then working backward to explain a move that already happened, is the same mistake in a different order: it treats the most lagging indicator as the trigger for a decision the fastest-moving indicator could have flagged two or three weeks earlier.

A weekly reading routine

The routine that fits around a normal week is short. Monday means pulling hook rate and three-second view-through rate by creative, flagging anything down more than roughly 15% against its own four-week average rather than against a portfolio benchmark, because decay is relative to a creative's own trajectory. Wednesday is for IPM: check the trend on whatever Monday flagged, since a hook rate dip that has not yet reached IPM may still be noise. Friday is the narrow one. Review CVR and blended CPI only for creative that has declined consistently across both earlier checks, and use that final pass to set timing and priority for a refresh. Detection ought to have happened already.

A worked example

Take a hypothetical top-performing playable concept running for nine weeks. In week five, hook rate drops from a stable 38% to 33%, a decline that sits inside most teams' noise tolerance and gets dismissed. By week six, IPM has softened from 1,450 to 1,290, which confirms the week five signal was not noise. CVR barely moves through week seven, holding near 4.1% against a four-week average of 4.3%. That lag is what makes CPI-only monitoring dangerous, since a team watching CVR and CPI alone would still see a portfolio performing close to plan. CPI does not move meaningfully until week eight. By then the concept has been decaying for three weeks, and the replacement that should have entered testing in week five is only reaching the brief stage now.

Connecting detection to the retirement decision

Detecting the sequence early only pays off if it feeds a real decision, which is the retirement discipline this publication argued for in Creative Variant Retirement Rules for a Live Portfolio. A hook rate decline flagged in week five is no reason on its own to retire a concept, since plenty of creative recovers after a temporary dip tied to audience saturation rather than genuine fatigue. It is a reason to start the production clock on the replacement early, so that a genuine retirement decision in week seven or eight has a tested concept ready instead of a gap where the team flies blind while something new clears review.

What this buys a team

The value of reading creative fatigue leading indicators in the right order is not a better dashboard. It is roughly three weeks of runway between the first reliable signal and the point where cost data would have forced the same decision anyway. Three weeks is enough time to brief and produce a genuine replacement and get it tested, rather than scrambling for one after the fact.

Creative fatigue leading indicators: what to watch next

Watch whether automated bidding systems mask the sequence described here by reallocating spend across audience segments faster than a manual weekly review can track. A sufficiently aggressive algorithm compensating for a decaying creative could flatten the CPI-stage signal even further, which makes the earlier hook rate and IPM checks more important rather than less.

Watch, too, whether any major network starts surfacing hook rate or early-frame retention metrics directly in its own reporting dashboards, since that would cut the manual pulling a team does to run the Monday check described above. And watch whether the three-week runway estimated here holds steady as production timelines shorten with AI-assisted creative tools, because a faster production cycle changes how early a team has to catch the first signal to still ship a tested replacement in time.

Related archive reading

These articles provide related context and remain subject to their stated review status.

• Building a Creative Fatigue Detection Cadence That Works
• Detecting Creative Fatigue Before It Shows Up in CPI