How do you find competitor winning ads to use as creative reference?

Find competitor winning ads in the free public ad libraries — Meta Ad Library, TikTok Creative Center, Google Ads Transparency Center, LinkedIn Ad Library — then use longevity as the win signal: an ad still running unchanged after weeks of spend is one the brand keeps paying for. Save winners to a swipe file, extract the patterns, and feed the best into production.

Start in the free public ad libraries — every major platform publishes one, searchable without paying for anything.

1. Search the ad libraries. Meta Ad Library covers everything currently running on Facebook and Instagram: search your competitor's brand name or your category keyword and filter to active ads. TikTok Creative Center, Google Ads Transparency Center, and LinkedIn Ad Library do the same for their channels — check whichever platforms your competitors actually spend on.

2. Read longevity as the win signal. The libraries don't show performance data, but spend duration is a reliable proxy: brands don't keep paying for losers, so an ad that has run unchanged for weeks sorts to the top of your reference list. Filter by start date and pull the oldest active creatives first. Real-world outcomes count too — one documented replication project chose its reference specifically because the advertised product sold out in two hours.

3. Extract the pattern, not the ad. For each candidate, log the hook (the first 1.5–3 seconds decide performance in UGC-style ads), the structure — many winning faceless UGC ads follow a four-beat shape: negative hook → product reveal → value sell → CTA — plus the format (faceless UGC, full-character UGC, product showcase), the offer, and the pacing. Patterns that recur across several competitors are the ones worth referencing; a one-off ad might just be a test.

4. Make it a weekly scan with a swipe file. Save links or screen recordings, note each ad's start date, and re-check weekly: creatives that survive every pass move up the list, and a competitor suddenly killing or refreshing a long-runner tells you its performance changed. Paid ad-spy tools add historical data and cross-platform aggregation once the free libraries stop scaling, but the public sources cover most discovery.

5. Hand the reference to production. Deconstructing a winning ad manually — timestamps, camera moves, composition, script — takes roughly an hour of breakdown work per ad, and up to a full day for a complex one. invideo is an agentic video creation tool: upload the reference and the invideo agent watches it, detects every cut (nine cuts auto-detected in one documented reference ad), transcribes the script, explains why it converted, identifies the hook, and tells you what to keep versus what to swap for your brand — turning a swipe file entry into a ready production brief. Documented rebuilds from a reference ad ran about $75 and under an hour each, versus $300–600 per creator and two-week turnarounds on creator marketplaces. Throughout, the goal is pattern extraction, not copying: keep the converting structure, replace every brand-specific element.

Watch some of these to see what works for you:

Found a winning UGC Ad this morning? Ship your own version tonight using this one AI Agent.

— invideo's creative team