The frequency thresholds that signal fatigue
Frequency, the average number of times a person has seen an ad, is the clearest leading indicator. Omniscia flags a warning at a frequency of 2.0, marks an ad fatigued at 3.0, and critical at 4.0. Those bands give you a graded read rather than a single pass/fail, so you can act early at the warning stage instead of waiting for delivery to crater.
Frequency alone is not the whole story, which is why Early Warning pairs it with CTR-decline detection. A creative whose click-through rate is trending down while frequency climbs is fatiguing fast, and the combination is a stronger signal than either number read in isolation.
Survival analysis: how long your ads actually last
On top of the live thresholds, Omniscia models creative lifespan with survival analysis, the same statistical approach used to estimate how long things last before an event. Applied to your account, it learns how long your ads typically run before fatiguing, so the prediction is calibrated to your creatives rather than a generic rule.
That is what turns 'this ad is at frequency 2.4' into 'this ad has roughly this many days left at current pace'. You stop reacting to today's number and start planning the refresh before the decline begins.
how it works
- 01
watch the warning band
Treat a frequency of 2.0 as the moment to prepare a refresh, not a number to ignore until 4.0.
- 02
confirm with CTR trend
Check Early Warning's CTR-decline read alongside frequency; a falling CTR with rising frequency means fatigue is accelerating.
- 03
read the runway
Use the survival-analysis estimate of how long the ad typically lasts to schedule the refresh before throttling, inside the 5-to-7-day window.
- 04
rotate or refresh
Swap in a distinct new hook rather than a near-duplicate, so the replacement doesn't immediately cluster with what it's replacing.
frequently asked
- What frequency means an ad is fatigued?
- Omniscia flags a warning at a frequency of 2.0, fatigued at 3.0, and critical at 4.0. The warning band is the point to start preparing a refresh rather than waiting for delivery to collapse.
- How early can fatigue be predicted?
- Omniscia's Early Warning predicts fatigue 5 to 7 days ahead of platform throttling, using frequency thresholds, CTR-decline detection and survival-analysis modelling of how long your ads typically last.
- Why not just watch CPA?
- By the time CPA rises, the platform is already throttling and budget is already wasted. Frequency and CTR decline lead CPA, so they give you a window to act before the damage shows up in cost.
- Does the prediction adapt to my account?
- Yes. The survival-analysis model learns how long your own ads typically last before fatiguing, so the runway estimate is calibrated to your creatives rather than a fixed industry rule.
Last updated June 6, 2026