Doctrine
Your competitors are running your angle experiments for you, in public, for free. The Meta Ad Library shows every active ad a page runs. Two signals separate proven angles from noise: an angle with many creative variations (nobody makes 12 versions of a loser) and an angle that has run for months (nobody pays to keep a loser alive). You are not stealing ads. You are reading the market's already-graded homework, then writing your own answer.
When to use
- Before Angle Writer, to seed it with market-proven directions.
- When your tested angles all plateaued and you need fresh hypotheses.
- Works with no MCP: the Ad Library is public (facebook.com/ads/library). Web access helps.
Run it
Competitor ad teardown for my offer: [ONE-LINE OFFER]
My competitors: [2-5 NAMES OR PAGE URLS. If I only know one, suggest lookups for adjacent players.]
For each competitor, using the Meta Ad Library (facebook.com/ads/library, search their page, country: all):
- PROVEN ANGLES: group their active ads by the underlying message, not the visual. Flag groups with 3+ variations (they are paying to scale that message) and any ad running 90+ days (it is paying them back).
- For each proven angle, extract: WHO it targets (read it off the creative), the PAIN or desire it names, the PROMISE it makes, and the emotional register (fear, status, relief, belonging).
- Their hero message (the one claim their whole account leans on) and their visible differentiator.
- THE GAPS: which of my potential buyers do their angles ignore? Which pains does nobody in this market name? That white space is my opportunity list.
Then give me 3-5 angle hypotheses for MY offer: each one either (a) a proven market angle re-aimed at my differentiator, or (b) a gap angle nobody is running. Never copy their copy; the output is directions, not text.
## Guardrails
- Never reproduce a competitor's copy or creative. Patterns are free; plagiarism is a brand risk and usually a losing move anyway (their angle is fit to their offer, not yours).
- Never cite a competitor's implied results as fact ("they must be printing money"). Longevity and variation count are the only honest signals the Library gives.
- Date-stamp the teardown. Ad Libraries rot; a 6-month-old teardown is a history lesson.
## Good output looks like
A table of competitors x proven angles with the two evidence signals per angle, a named white-space list, and 3-5 angle hypotheses ready to hand to Angle Writer.