Guide
You found a competitor's ad in a spy tool and it looks great. But "it's running" and "it's working" are not the same thing — and copying the wrong ad is how you burn a budget on someone else's mistake. Here's how to actually judge it.
A spy tool or the Meta Ad Library shows you an ad exists. It doesn't show spend, ROAS, conversions, or targeting. So everyone reaches for proxies — and the most popular proxy is also the most misleading.
Long run time feels like proof, but it isn't. Ads sit on autopilot budgets nobody reviews. Big brands run creative for awareness, not direct response. Some advertisers just never turn losers off. Plenty of long-running ads are quietly mediocre — and plenty of genuine winners get rotated out early to beat fatigue. Run time is a hint. Treat it as one.
These are better than run time, but still circumstantial. None of them look at the ad itself.
The one thing you can measure directly is the creative. An attention model scores any publicly viewable ad on whether it holds a viewer — its Hook Strength, the exact second attention drops, and a percentile against 76 top-performing ads. A competitor ad that scores high and has run long in many variants is a real "worth modeling." One that scores low is telling you its success — if it has any — isn't coming from the creative, so copying the creative won't transfer it.
See it in practice: our Facebook ad library scores real, live ads pulled from the Meta Ad Library — with the hook strength, drop-off, and rank for each — so you can see exactly which are worth studying.
Once you know which competitor ads are genuinely strong, model the angle, build your version, and pre-test it against the same benchmark before you spend. That's how competitor research becomes an edge instead of a guess — you launch knowing your ad out-hooks the one you learned from. Full workflow: how to spy on competitor ads.
Score any public ad for attention and see if it actually holds up. First analysis free.
Score an AdYou can't see a competitor's spend or sales, so you infer from signals: how long the ad has run, whether they've made many variants of the same angle, and whether it keeps reappearing. These hint at success but aren't proof. The more reliable read is the creative itself — whether it holds attention — which you can measure directly.
Not necessarily. Long run time is the most over-trusted signal in ad research. Ads coast on autopilot budgets, run for brand reasons, or simply never get turned off. A long-running ad is a hint worth investigating, not evidence you should copy it.
Judge the creative, not just its run time. Score it for whether it holds attention — its hook strength and where engagement drops — against a benchmark of proven ads. A high score plus repeated, long-running variants is a strong 'worth modeling'; a low score means the ad's success (if any) probably isn't coming from the creative.