Meta Creative Fatigue in 2026: The Andromeda Thresholds

Meta creative fatigue in 2026 is governed by a different delivery system than the one most refresh rules were written for. Since Meta rolled out Andromeda — its ads retrieval engine announced in December 2024 — delivery leans harder on the creative itself, and operators report decay arriving sooner than their old rotation calendars expect.

This post covers four things: what Andromeda actually changed, which fatigue thresholds are backed by real data in 2026, which widely-shared numbers are folklore, and how to turn the signals into a Monday action list.

The short answer

• Andromeda is real and verified. It's Meta's personalized ads retrieval engine, announced December 2024, which selects ads from tens of millions of candidates and delivered a +6% recall improvement and +8% ads quality improvement on selected segments, per Meta's engineering team.
• "Creative is the targeting" is an industry thesis, not Meta's claim. Triple Whale's analysis argues your creative "now serves as the primary input the platform uses to decide what the audience should be." Meta's own post doesn't say that — but the thesis matches how delivery behaves in practice.
• "Fatigue now hits in exactly 2 weeks" is folklore. No tier-1 or tier-2 source publishes that number with a methodology. What is citable: Triple Whale's fatigue framework recommends a 7–10 day minimum evaluation window and flags decline signals on a weekly cadence — fatigue is now measured in days-to-weeks, not months.
• The thresholds that hold up: frequency above ~2.5 on cold audiences, week-over-week CTR drops of 10–15%, and CPM drift above ~18% over two weeks. Details and sources below.

Why this matters

Fatigue is an efficiency leak that compounds quietly. Triple Whale estimates that 15–25% of monthly ad spend is typically wasted on already-fatigued creative. That's their estimate, not an audited figure — but even at the low end, a $50K/month account would be paying $7,500 a month to ads that have already stopped working. That's plain arithmetic on their number, and it's the right order of magnitude to care about.

The macro backdrop sharpens the point. Per Tinuiti's Q1 2026 benchmark report, Instagram CPMs posted their first year-over-year decline since 2023 — after years of double-digit inflation — driven by a surge in available impressions. Cheaper inventory doesn't save a fatigued ad, though. When delivery is creative-led, creative efficiency is the lever, and fatigue is the main thing that erodes it.

What Andromeda actually changed

Strip away the hype and the verified facts are these. Andromeda is a retrieval engine: the stage of Meta's ad system that picks a shortlist of candidate ads from tens of millions before the final auction ranks them. Meta's engineering post reports a 10,000x increase in model capacity and says the system was built anticipating that the number of ad creatives will "grow significantly."

The practical interpretation — and this is Triple Whale's framing, not Meta's — is that the system evaluates historical engagement, ad copy, format, and visual signals before it considers advertiser-defined audiences. In their words: "The creative is the targeting."

What that means for fatigue, mechanically:

1. More creatives compete for retrieval. Your ad isn't just wearing out an audience; it's losing retrieval slots to fresher candidates.
2. Performance signals decay the ad's candidacy. Falling engagement doesn't just lower CTR — it makes the system retrieve the ad less, on worse inventory, at worse prices.
3. Refresh cycles shortened across the industry. Not because someone published a new fatigue constant, but because creative volume went up and retrieval rewards novelty of signal.

So when operators say fatigue "feels faster" in 2026, the mechanism is plausible and the direction is right. Just don't let anyone sell you a precise number of days.

The 2-week claim: what's verifiable and what isn't

Search for Andromeda and fatigue, and you'll find confident claims that fatigue now sets in within 2–3 weeks, down from 4–6+. We looked for the source. There isn't one — those numbers circulate on AI-tool vendor blogs with no dataset, no methodology, and no consistent figure (one puts it at 3–5 days). Treat them as marketing, not measurement.

Here's what's actually citable in 2026:

• Triple Whale's Creative Fatigue Framework sets a minimum evaluation window of 7–10 days before judging a creative, and runs its decline checks on a weekly cadence.
• Meta's own Creative Fatigue recommendations trigger off cost-per-result, with no published timeline at all.
• The honest summary: nobody credible has published a "fatigue now takes N days" study. What changed is the cadence of detection — weekly reads are now the floor, because by the time a monthly review catches decay, you've paid for two to three weeks of it.

That reframing matters for how you operate. The question isn't "how many weeks do I get?" It's "how fast do I notice?"

The thresholds that hold up

These are the fatigue signals with named sources behind them. Two or more moving together over a week is a fatigue call; any single one alone is a question, not an answer.

| Signal | Threshold | Source | |---|---|---| | Frequency (cold audiences) | Above ~2.5, performance decline tends to follow | Triple Whale | | CTR trend | Week-over-week drop of 10–15% or more | Triple Whale | | CPM drift | More than ~18% increase over a two-week period | Triple Whale | | Meta's "Creative limited" status | Cost per result elevated above your historical baseline, but under 2× | Meta Business Help Center | | Meta's "Creative fatigue" status | Cost per result at least 2× your historical baseline | Meta Business Help Center |

Two notes on reading that table.

Meta's native statuses are cost-based, not frequency-based. A common misreading is that Meta flags fatigue at some frequency cutoff. It doesn't — the Delivery column statuses key off cost per result versus your own history. By the time "Creative fatigue" shows, your cost per result has already doubled. The third-party thresholds exist to catch decay before that, which is the whole game. For the funnel-by-funnel frequency bands (DTC vs lead gen, the 2.5 vs 3.5 debate), see our creative fatigue frequency thresholds guide.

The classic decay curve is older than Andromeda. AdEspresso's 500-campaign study — from 2018, so label it accordingly — found CTR down about 23% and CPC up about 68% by frequency 4, and CTR roughly halved with CPC up 161% by frequency 9. The exact numbers are pre-Andromeda, but the shape of the curve — decay compounds with each exposure, and costs rise faster than clicks fall — still matches what operators see.

The counterargument: not all repetition is fatigue

Motion's recent piece argues the opposite corner, and it's worth taking seriously: drawing on Les Binet's wear-in research, they contend that "Message saturation is not the problem. If anything, it should be the goal" — repeated exposure builds recall, and some ads improve with repetition.

Both things are true, in different jobs. Brand campaigns optimizing for recall can wear in over repeated exposures. Direct-response campaigns optimizing for purchases wear out, because the system needs fresh engagement signal to keep retrieving the ad onto good inventory. If you're running DR on Meta — which is most readers of this blog — the fatigue thresholds apply. If an ad's frequency is high but CTR and conversion rate are flat, you may be looking at wear-in or a tight retargeting pool, not fatigue. Don't refresh a working ad on a frequency number alone.

A Wednesday-morning walkthrough

Illustrative numbers — the pattern shows up repeatedly in high-spend Meta accounts. A DTC account, three prospecting ads, checked against the table above:

| Ad | Frequency | CTR (WoW) | CPM (2-week) | Meta status | Read | |---|---|---|---|---|---| | Video A | 3.1 | −19% | +24% | Creative limited | Three signals confirm. Fatigue. | | Static B | 2.7 | −6% | +9% | — | One soft signal. Early-warning, not action. | | UGC C | 2.2 | +4% | +3% | — | Healthy. Leave it alone. |

The tiered call: Act today — pause Video A and ship the replacement brief; waiting for Meta's "Creative fatigue" status means waiting for cost per result to double. This week — prep a variant for Static B and re-check Friday. Monitor — UGC C runs untouched.

That triage takes a senior buyer ten minutes per account — when the deltas are already computed. Across an agency's 20 accounts, it's a half-day of pulling exports, unless the diagnosis arrives pre-done. That's the job an Account Health Score and an urgency-tiered action list exist for: the analysis is already finished before you open anything.

Common mistakes

• Refreshing on a calendar instead of on signals. "Rotate every 2 weeks because Andromeda" burns creative budget on ads that were still working. Rotate on threshold breaches, not folklore timelines. Our refresh cadence guide covers trigger-based rotation.
• Waiting for Meta's native flag. "Creative fatigue" status means cost per result already doubled, per Meta's definition. It's a confirmation, not an early warning.
• Treating one metric as a verdict. Frequency 3.0 with flat CTR and CVR isn't fatigue. CTR down 12% with frequency 1.4 probably isn't either — check creative against landing page before you blame wear-out.
• Applying DR fatigue rules to brand campaigns. Recall-objective campaigns can legitimately wear in with repetition, per the research Motion cites. Match the rule to the objective.
• Quoting vendor folklore as benchmarks. "Accounts lose $8–15K/month to fatigue" and "fatigue hits in 14 days" have no published methodology behind them. If a number has no dataset, it's a pitch.

FAQ

Did Andromeda officially shorten creative fatigue timelines? No official statement says so. Andromeda is Meta's ads retrieval engine, announced December 2024. It made delivery more creative-driven, and operators widely report faster decay — but no tier-1 source has published a before/after fatigue timeline. Detect by threshold, not by calendar.

What are the creative fatigue thresholds to watch in 2026? Frequency above ~2.5 on cold audiences, week-over-week CTR drops of 10–15%, and CPM rising more than ~18% over two weeks, per Triple Whale's framework — plus Meta's native "Creative limited" / "Creative fatigue" statuses, which trigger when cost per result is elevated or at least doubled versus your history.

Is Meta's built-in fatigue detection enough on its own? It's a useful floor and a late one. The statuses are cost-per-result based, so they confirm fatigue after the damage is priced in. Third-party thresholds catch the slide one to two weeks earlier, and a creative fatigue tool turns the flags into a ranked list of which ad to rotate first.

How often should I check for fatigue now? Weekly, minimum. Triple Whale's framework uses a 7–10 day evaluation window, and weekly is the cadence at which the CTR and CPM deltas above become readable. Monthly reviews concede two to three weeks of paid-for decay. A weekly Meta Ads report that computes the deltas for you makes the cadence sustainable.

Does rising CPM always mean fatigue? No. Market-wide CPMs move — Tinuiti's Q1 2026 data showed Instagram CPMs actually declining year over year on surging impression supply. Fatigue is CPM rising on your ad specifically while CTR falls and frequency climbs. Always read the signals together.

---

Creative fatigue in 2026 isn't a new disease — it's the old decay curve running on a faster, creative-led delivery system, surrounded by more confident folklore than ever. The fix isn't a magic refresh interval. It's weekly threshold checks: frequency, CTR trend, CPM drift, cost per result versus baseline — and a ranked decision about what to rotate first. If you'd rather skip the spreadsheet and get that as a pre-diagnosed, urgency-tiered action list every Monday, see how Good Morning's creative fatigue detection works — $50/mo per account, read-only Meta access. Action items, not analysis.