Genyad Labs · Research Brief

The Decay Curve: how fast a winning ad actually dies in 2026

We pulled together the frequency, CTR, and hook-rate thresholds now scattered across a dozen platform reports and agency studies, and lined them up against one timeline. The pattern is consistent even when the source isn't: most creative is dead within three weeks.

Fig. 1 — Composite CTR decay, days since creative launch
Observed CTR index Baseline (day 1 = 100)
wk1: −15–20% wk3: cliff, −45–70% wk5+: −38% avg. vs. peak Day 0 Day 7 Day 21 Day 35 100 75 50 25
Indexed composite, CTR = 100 at launch Sources: AppsFlyer, Meta internal analytics, AdEspresso, Mako Metrics — see §5
01 / Headline numbers

The five figures every media buyer should have memorized

None of these are new data — they're published across a fragmented set of platform docs and agency blogs. What's new is seeing them on one page.

15–20%
CTR decline in the first two weeks of a creative's life, before the drop accelerates
2.5
Weekly frequency on Meta prospecting where performance decline begins
45%
CTR drop measured after a fourth exposure to the same creative, per Meta's internal research
9
Days for food & beverage ads to lose 40% of CTR — the fastest-fatiguing vertical measured
3–5×
Longer campaign lifespan for brands shipping 15–50 creative variants a month vs. quarterly refreshers
8–20
Live variations a typical active campaign needs in rotation to stay ahead of the decay curve
02 / Platform benchmarks

Fatigue looks different depending on where the ad runs

Meta's Andromeda ranking generation weighs creative-level signals more heavily than its predecessor, which compresses the fatigue window on Reels specifically. TikTok and YouTube tolerate slightly different frequency ceilings, but the direction is identical everywhere: novelty decays fast, and short-form vertical video decays fastest.

PlatformMedian video CTRMedian hook rateFrequency ceiling (prospecting)
Meta Reels1.78%28%2.5
Meta (feed avg.)1.62%28%2.5
TikTok0.84%33%~3.0
YouTube in-stream0.42%22%3–7 / wk (recall floor–ceiling)

Composite of platform benchmark studies covering 12,000+ ads; see sources.

03 / Fatigue speed by vertical

Some categories burn through creative in days, not weeks

Food & beverage and fashion sit at the fast end — high purchase frequency and high visual saturation compress the decay window. Considered-purchase categories (finance, B2B SaaS) tend to tolerate longer runs, mostly because their audiences see the ad less often per capita.

Food & beverage
9 days
Fashion
12–14 days
Beauty / DTC
~18 days
Electronics
~21 days
B2B SaaS
~28 days

Days to a 40% CTR decline from peak. Ranges synthesized from vertical benchmark reporting; see sources.

04 / What this means for creative volume

Throughput, not talent, is the actual bottleneck

Every study above points at the same operational conclusion: the number of unique creative concepts a team can ship per month predicts campaign longevity better than the quality of any single ad.

Brands producing 15–50 new variants monthly extend campaign life three to five times longer than brands refreshing on a quarterly cycle. That gap isn't a creative-talent gap — it's a production-capacity gap. Each additional variation has traditionally required a new script, a new edit, a new voiceover, and a new export per platform, which is why most teams under-produce relative to what the decay curve actually demands.

This is the specific gap Genyad is built to close — turning existing footage into new scripts, voiceovers, and platform exports without a reshoot, so the 8–20 live variations a campaign needs are a production question, not a budget one. See how the workflow compresses this →
05 / Methodology & sources

How this report was built

This is a synthesis, not a primary study: Genyad Labs aggregated publicly reported benchmarks, platform documentation, and agency research on ad fatigue and creative decay published between January and August 2026, normalized to a common day-since-launch timeline, and cross-checked figures that appeared across multiple independent sources. Where sources disagreed, we report the range rather than picking a single number.

Have a dataset that confirms, refines, or contradicts these ranges? We'll update this report and credit the source — reach us via the contact link in the footer.

Your creative is decaying right now. It's just a matter of how fast.

Genyad turns the footage you've already shot into the variation volume this curve demands.

See how it works