Before you start — what you need
- Access to Meta Ads Manager (or your platform's reporting dashboard with daily breakdowns)
- A creative that has been live at least 14 days with ≥2,000 impressions — below that, data is too thin
- Ability to pull daily CTR, CPM, and frequency for the creative
- Audience breakdown by frequency bucket (1–2 impressions vs 4+ impressions per user)
- A tracking spreadsheet or creative library to log the diagnosis and outcome
The 4 signals of creative fatigue
Fatigue has a specific quantitative signature. The four signals below, checked together, distinguish actual fatigue from day-of-week variance, audience saturation, and underlying creative problems. Each has a threshold and a place in the reporting dashboard where you can pull it.
| Signal | Threshold | Severity | Where to find it |
|---|---|---|---|
| CTR decay | ≥20% drop from week 1 to week 3 | Medium | Meta Ads Manager → trend chart |
| Frequency rise | ≥3.5 in a 7-day window (severe: ≥5.0) | High | Meta Ads Manager → frequency column |
| CPM increase | ≥25% rise from week 1 to week 3 | Medium | Ads Manager → cost trend |
| New-audience health | New audience flat + mature down ≥30% | High (diagnostic) | Breakdown by frequency bucket |
For the full definition of the pattern, see our glossary entry on ad creative fatigue, and our blog post Low CTR Ads: How to Diagnose and Fix for the broader diagnostic context.
The 7-step diagnostic playbook
Pull the 30-day CTR trajectory for the creative
Fatigue looks like a specific curve. In Meta Ads Manager (or your reporting tool), chart daily CTR over the last 30 days for the single creative you're diagnosing. Healthy creatives show CTR that bounces around a steady mean. Fatiguing creatives show a clear decay — CTR was 1.4% in week 1, 1.2% in week 2, 0.9% in week 3. Any decay of ≥20% from the week-1 mean to the week-3 mean is a fatigue signal, especially if the decay is monotonic rather than zigzagging.
- →Look at rolling 7-day averages, not daily spikes
- →Threshold: ≥20% decline from week 1 to week 3 mean = signal #1 active
- →Compare against your own CTR baseline, not just industry benchmark — the decay pattern matters more than the absolute level
This step is done when: You have a 30-day CTR chart and can state whether the decay exceeds 20% (signal #1 active or clean).
Pull the frequency metric
Frequency is impressions-per-unique-user in the reporting window. Meta exposes this directly; most platforms do the same. Above 3.5 in a 7-day window, creative fatigue starts to register as measurable CTR decay. Above 5.0 is the severe zone where CTR often collapses by 40–50% over the baseline. Frequency is the cleanest single fatigue signal because it's causally upstream: over-exposure produces fatigue, not the other way around.
- →Threshold: frequency > 3.5 in a 7-day window = signal #2 active
- →Severe zone: frequency > 5.0 = signal #2 active at high intensity
- →For broad targeting, frequency typically runs lower; for narrow, higher — calibrate against your account's history
This step is done when: You've pulled the 7-day frequency and know whether it exceeds 3.5 (signal #2 active or clean).
Check the CPM trajectory over the same window
CPM tells you what the auction is doing. A creative that's fatiguing triggers two things in the auction: Meta (or whichever platform) serves it at lower rates, and the relevance/quality score often drops — both of which push CPM up. Chart daily CPM for the last 30 days. If CPM has risen ≥25% from week 1 to week 3, and it's happening alongside CTR decay, the fatigue signature is strong. Rising CPM alone (without CTR decline) is usually macro-level competition; the two together are fatigue.
- →Threshold: CPM up ≥25% from week 1 to week 3 = signal #3 active
- →Rising CPM alone = competition pressure or delivery issue, not fatigue
- →Rising CPM + falling CTR = strong fatigue signature
This step is done when: You have a 30-day CPM chart and can state whether the rise exceeds 25% (signal #3 active or clean).
Compare new-audience to mature-audience performance
This is the tiebreaker signal. Segment your creative's performance by audience stage: impressions served to users who've seen the creative 1–2 times vs users who've seen it 4+ times. If CTR on new audiences (1–2 impressions) is flat or healthy while CTR on mature audiences (4+ impressions) is in decay, that's textbook fatigue — the creative still works on fresh eyes but doesn't survive repeat exposure. If CTR is declining on new audiences too, the problem isn't fatigue; it's a creative quality or market-fit issue that no refresh will solve.
- →Threshold: new-audience CTR stable but mature-audience CTR down ≥30% = signal #4 active
- →New-audience CTR also declining = it's not fatigue, it's a creative quality problem
- →This signal is the most diagnostically useful of the four — it separates fatigue from design failure
This step is done when: You've compared new vs mature audience CTR and know whether the fatigue signature is present (signal #4 active or clean).
Score the combined fatigue level — low, moderate, or severe
Count how many of the four signals are active: CTR decay, frequency over 3.5, CPM rise, and mature-audience decline. Zero signals = the creative is healthy; don't change anything. One signal = moderate watch; monitor but don't act yet. Two or three = fatigue is real; it's time for the decision tree in step 6. Four signals = severe fatigue; act immediately. Quantifying the level prevents the common mistake of reacting to a single weak signal with expensive creative rebuilds.
- →0 signals = healthy, no action
- →1 signal = watch, gather 5–7 more days of data
- →2–3 signals = act (go to step 6)
- →4 signals = severe, act immediately and aggressively
This step is done when: You can state the fatigue level (low / moderate / severe) and the number of active signals.
Apply the refresh-vs-kill-vs-expand decision tree
Three paths, each triggered by a specific pattern. REFRESH if the creative has 2–3 active signals, it was previously a top performer, and you have a variant pipeline to draw from. Swap in a new variant of the same creative concept — different frame, different headline — without killing the underlying approach. KILL if the creative has 4 signals and no variant of the concept is working, or if new-audience CTR is also declining (meaning the concept itself is dead, not just over-exposed). EXPAND if the creative is still performing on fresh audiences but fatiguing on mature ones: broaden the audience (new lookalikes, expanded interests) rather than touching the creative at all. Expansion gives a fatiguing-but-good creative more fresh eyes, which resets the fatigue clock.
- →Refresh = new variant of same concept (cheapest, fastest)
- →Kill = pull creative, replace with different concept (when the idea is dead)
- →Expand = new audience, same creative (when the creative still works on fresh eyes)
- →Don't mix: pick one path and execute cleanly
This step is done when: You've named one path (refresh, kill, or expand) and documented the reason.
Implement the action and set a re-evaluation date
Execute the chosen path: swap in the refresh variant, pause the killed creative, or launch the expanded audience — whichever the decision tree selected. Then set a hard re-evaluation date: 10 days out. On that date, re-run the four-signal check. If the action worked, the signals should be clean. If they're still active, the diagnosis was wrong; try a different path from step 6 (for example, if refresh didn't fix it, the concept is likely dead and it's time to kill). Logging both the diagnosis and the outcome builds an account-specific fatigue playbook that sharpens across cycles.
- →Execute exactly one action — don't refresh AND expand simultaneously
- →Set a calendar reminder for re-evaluation at +10 days
- →Log the full diagnosis (signals + chosen path + outcome) in your creative library
This step is done when: The action is live in the ad account, a 10-day re-evaluation is scheduled, and the full decision is logged.
The refresh-vs-kill-vs-expand decision tree
Once you've counted active signals in step 5, use this tree to pick an action. Don't improvise — the tree is what turns diagnosis into a cleaner, faster decision than team debate.
REFRESH
Triggers: 2–3 signals active · creative was a previous winner · variant pipeline available
Action: Swap in a new frame of the same concept. Preserve the underlying hypothesis (headline promise, product angle, CTA structure). Only the visual frame should change. Cheapest and fastest path — expect CTR to recover within 5–7 days if the concept is still viable.
KILL
Triggers: 4 signals active · OR new-audience CTR is also declining · OR previous refresh attempts didn't recover performance
Action: Pull the creative. Replace with a different concept, not a frame refresh. If new audiences are also rejecting the creative, the concept itself is the problem — a different headline, different offer framing, or different visual angle. This is when you go back to the drawing board.
EXPAND
Triggers: Mature-audience CTR is declining but new-audience CTR is healthy · OR frequency > 4 in a narrow audience
Action: Broaden the audience. Add a new lookalike, expand interest targeting, or move to Advantage+ broad. The creative is still working on fresh eyes; you just need to feed it more of them. No creative work required — this is an audience decision, not a creative one.
Common mistakes to avoid
Calling fatigue on a single weak signal
One of the four signals firing is a watch condition, not an action condition. CTR decay without frequency or CPM moving is often just day-of-week variance or macro-level shifts. Wait for at least two signals to align before touching the creative.
Confusing fatigue with creative quality problems
If new-audience CTR is also declining, it's not fatigue — the creative never worked. Refreshing won't help. Kill it, diagnose the original failure (see our fix-low-CTR playbook), and rebuild. The new-vs-mature audience split is the cleanest test for this distinction.
Refreshing when you should expand
If the creative is still performing on new audiences, the concept isn't dead — the impression pool is just exhausted. Expanding audience is faster, cheaper, and preserves a proven concept. Refreshing a still-good creative wastes production effort on a problem you don't have.
Killing too late
A creative at 4 signals active has usually been bleeding ROAS for 1–2 weeks before the team notices. Schedule the four-signal check weekly on creatives running ≥14 days. Catching fatigue at 2 signals and refreshing costs 1 week of production; catching it at 4 signals costs a month of wasted spend.
Treating every refresh as a brand-new hypothesis
A refresh is 'same concept, new frame' — not a test of a new creative direction. Don't conflate a fatigue refresh with a pre-launch variant test. Keep the hypothesis constant; change only the frame. If you want to test a new concept, do it as a separate, documented test.
Frequently asked questions
How long does it take for creative fatigue to set in?
What's the fastest way to spot fatigue?
Is creative fatigue different from audience fatigue?
What's the minimum data to diagnose fatigue reliably?
Where can I read more about creative fatigue?
Related how-tos
Score refresh candidates before you swap
When fatigue triggers a refresh, you don't want to guess which new frame will re-ignite CTR. Upload refresh candidates to GazeIQ and pick the one that beats the tired creative on attention score before it goes live.
No credit card required · 3 free scans included