Competitor Ad Teardown

Competitor ad libraries are public, and almost everyone reads them wrong. They screenshot a few ads, note that "they use social proof", and produce a swipe file that changes nothing.

The value is in two things people skip: using how long an ad has been running as a proxy for whether it works, and turning the pattern into an opening for a different position rather than a copy of theirs.

Where the data comes from

All public, all first-party disclosure — no scraping of private data, no impersonation:

• Meta Ad Library (facebook.com/ads/library) — every active Meta ad, searchable by advertiser, with start dates. This is the richest source because of those dates.
• Google Ads Transparency Center (adstransparency.google.com) — active Google/YouTube ads by advertiser and region, with date ranges.
• TikTok Creative Center, LinkedIn Ads Library — same idea, varying detail.

Ask the person for the competitor's exact brand name or page, their market/region (libraries are region-filtered and results differ), and their own offer and positioning — without that last piece you can describe what a competitor does but not what it means for them.

If a browser tool is available, look the ads up directly. If not, ask the person to open the library and paste what they see, and tell them precisely what to capture: ad copy, format, start date, and the landing page it points to.

How to read what you find

Longevity is the signal

This is the core technique. Ad platforms let advertisers run anything; only ads that make money survive.

• Running 30+ days, and especially 90+ → validated. Someone is paying for it every day because the math works. Treat these as the competitor's proven positions.
• Running under 2 weeks → a test. Interesting for direction, but no evidence it works. Don't build strategy on it.
• Many near-identical variants running simultaneously → an active testing program and usually a well-funded account. Note which element varies between them: whatever they're testing is what they believe drives results.
• One ad running for a year → their control. This is the single most informative asset in the library; it represents the message they've never managed to beat.

Always report the start date alongside every ad you cite, and lead the analysis with the long-runners.

Then classify the angles

For each proven ad, extract: the core promise, the angle it uses to make that promise credible, the offer mechanic (discount, trial, bundle, guarantee, financing), the format, and the audience it implies (the language and imagery tell you who they think they're talking to).

Angles cluster into a small number of families — price/value, speed/convenience, status/identity, risk reversal, fear/loss avoidance, social proof, novelty, authority. Tally which families they occupy heavily and which they leave empty.

Find the opening

The output people actually need is a gap, and gaps come in a few shapes:

• Unoccupied angle — they compete entirely on price; nobody in the category is claiming reliability.
• Unserved segment — every ad speaks to enterprise buyers; SMB is unaddressed.
• Weak proof — they claim a result but never evidence it; you have case studies.
• Offer friction — their conversion path demands a demo call; a self-serve trial undercuts it.
• Format absence — everything is static; the category has no one doing short-form video well.

A gap is only real if you can credibly fill it. Check each against the person's actual offer and say so; an opening you can't service is just trivia.

What to produce

COMPETITOR TEARDOWN — [competitor], [market], [date checked]
[n] active ads · [n] running 30+ days · oldest running since [date]

PROVEN POSITIONS (long-runners — what's paying for itself)
  [start date · days live] [angle] — [the promise, in their words]
  …

ACTIVE TESTS (recent, unproven)
  …

ANGLE MAP
  Heavily occupied: […]
  Absent: […]

YOUR OPENINGS
  1. [gap] — why it's open, and the specific claim you could make instead
  …

Close with two or three ad concepts for the person's own offer that exploit the strongest opening — each as a positioning brief (promise, angle, proof, offer), not finished copy they'd paste unchanged.

Judgment worth applying

Longevity is evidence, not proof. Big brands run ads for reasons unrelated to direct response, and a long-running ad may simply be neglected. Weigh it, don't worship it.

You are not seeing their results. The library shows what runs, never what it costs or returns. Never state or imply you know a competitor's performance — say "still running after 6 months, which suggests it works", not "their best performer".

Don't copy — position against. Beyond the legal problem with copying creative, running the same angle as an incumbent with a bigger budget is a losing trade. The point of the analysis is differentiation.

Small samples mislead. A competitor with 4 ads tells you little. Say when the sample is too thin, and consider widening to 3–4 competitors to see genuine category patterns.

Check the landing page too. The ad is half the message; where it goes reveals the offer, the funnel, and often the price. Note when you couldn't check it.

Keep it honest. Use only publicly disclosed library data. Don't attempt to access private account data, don't impersonate a competitor, and don't reproduce their creative assets — describe them.