Metric patterns that locate where an ad breaks — low thumbstop with OK CTR (fix the hook), high watch with low CTR (stronger CTA), clicks but short watch (shorten), high CTR with no conversions (landing page/offer), rising CPM/CPA on a winner (fatigue), no spend at all (similarity) — and how to turn each into a testable hypothesis.

Meta ad diagnostic cheatsheet

TL;DR

Pattern	Likely break	Fix to test
First-frame retention < 90%	thumbnail / first frame	new opening frame¹
Low thumbstop, CTR in line with account	hook	borrow the first 3 s of a top-hook ad, keep the body²³
High completion / hold, low CTR	CTA	stronger, more prominent CTA at the end⁴
Low completion but good CTR	length	shorten, move CTA earlier⁵
Good hold + CTR but weak hook rate	hook not tailored to persona	re-hook for the persona⁶
High CTR, poor conversions	landing page / offer mismatch	congruent page, not the homepage⁷⁸
Former winner: CPM ↑, CPA ↑	fatigue	iterate / refresh / rotate⁹
Many ads, 1–3 get spend	similarity clustering	genuinely different concepts¹⁰

Always anchor on spend + bottom-line metric first: obvious winners (high spend, strong returns) need no diagnosis — make more like them.¹¹
Write the fix as a hypothesis (“raise thumbstop while holding CTR → more site visitors”) so the team knows what is being validated.¹²

Details

Scope the analysis to one funnel stage — e.g. prospecting (cold) ads only — before comparing ads.¹³

Hook fix in practice (Evan Lee, Motion). He isolated an ad with a weak thumbstop but average CTR, paired it with the account’s best-hooking ad, and briefed the team to mimic the winner’s first three seconds — a “mash-up” test.³¹⁴

Video drop-off. Read the 25/50/75/100% view funnel: a big drop after 25% is fine if CTR is strong (people are leaving to click); the same drop with weak CTR is “mission critical”.¹⁵¹⁶

Not every problem is creative. An ad set with ~5% CTR but poor sign-ups was sending people to the homepage; the fix was a dedicated page whose headline continued the ad’s message.⁷⁸ Zach Stuck: if offer and landing page are good and performance is flat, it’s probably the ads.¹⁷

Zoom out. Group ads into buckets (product, creator, SKU, hook tactic, messaging theme) to set weekly resource hypotheses, e.g. “spend more on product X → we need more creatives for X”.¹⁸¹⁹

Statics. Fewer metrics exist; analyse on-image text/headline, persona and messaging instead (Josh Brampton).²⁰

Disagreements / caveats

Dara Denney warns CTR is an unreliable success signal under broad targeting (lower CTRs are normal); use it as a diagnostic, not a KPI.²¹
Evan’s example data was small-spend demo data — as a viewer noted, you’d want more spend before acting.²²

Diagnosing why a Meta ad isn't working (metric-pattern cheatsheet)

Meta ad diagnostic cheatsheet

TL;DR

Details

Disagreements / caveats

See also

Diagnosing why a Meta ad isn't working (metric-pattern cheatsheet)

Meta ad diagnostic cheatsheet

TL;DR

Details

Disagreements / caveats

See also

Footnotes