How to Research Competitor Facebook Ads in 2026

Meta Ad Library is the public-source-of-truth for competitor ads. It tells you which ads a brand is currently running. But it doesn't tell you the things that matter for creative strategy: what hooks the videos open with, how they emotionally position the product, which products get which formats, or how the strategy varies by audience.

Format mix needs the same care: work it out from what each creative actually shows. In a 30-day sample, 55% of SKIMS ads are image creatives and 45% are video, so the brand invests in both formats at scale.

To do real competitor research, you need to layer on top of Meta's data: hook classification from video transcriptions, emotional scoring of every creative, format-to-product mapping, and platform-strategy patterns.

Active ad count is your first sanity check. It tells you the magnitude of their creative testing budget. Brands like SKIMS run thousands of active ads, while focused DTC brands like Poppi or Olipop sit at a few hundred. Marketplace brands like Temu run 50,000+: different game entirely.

If your competitor is running 1,000+ active ads, they have a serious testing pipeline. If under 200, your creative depth can compete head-to-head.

Sample 100-250 of their ads from the last 30 days and classify each by actual content type, image or video, from the creative itself. Glossier runs a roughly 50/50 image-to-video mix. Liquid Death is the surprise: despite a brand built on loud video, 81% of its sampled ads are static images, with video saved for collections like best-sellers.

Format mix tells you whether the category rewards visual product-shot ads (fashion, beauty) or longer-form storytelling (wellness, beverage).

This is where most competitor research stops too early. The first 1-3 seconds of a video ad determines its CPM and engagement. Pull transcriptions of all video ads, classify the openings, and you'll see five patterns dominate every DTC category: sensory demo ("Look how soft this is..."), transformation ("I never thought I'd say this..."), problem statement ("My skin's been breaking out..."), concept reframe ("What if your sleep is actually..."), and authority ("Doctors recommend...").

Examples: Huel leans on transformation + authority. Dr. Squatch mixes problem + humor. The Ordinary uses authority + sensory demo. See our DTC video ad hooks guide for the full taxonomy with verbatim examples.

The link URL on every Meta ad reveals which products and collections the brand is pushing through paid channels. More importantly, you can map format-to-product: SKIMS sends 60% video to "everyday cotton" but 100% image to "best-sellers"; that's a strategic decision about which products need motion to sell vs. which need a clean product shot.

Group ads by landing-page slug, count formats per group, and you'll see the strategy. For data-rich brand analyses with format-by-product splits already computed, see SKIMS, Oura Ring, or Headspace.

Pick a category. Pick 5-10 brands in that category. Run the same 4-question analysis on each. Compare results.

For example, if you're operating in fashion, study SKIMS, Gymshark, Lululemon, and Shein side-by-side. In beauty, look at Glossier, Rare Beauty, Fenty Beauty. Wellness and nutrition have their own deep playbooks.

Pattern-match across the 5-10 brands and you'll find the conventions of the category. Then identify what's missing (the angle nobody is testing), and that's your wedge.

• Guessing the format mix: sample the actual creatives and classify them yourself.
• Looking only at active ads: running history matters. Long-running creatives (90+ days) are the workhorses. Short-running ones are tests.
• Ignoring transcriptions: the first sentence of a video ad is the most copyable signal in competitor research. Most teams never extract them.
• Studying brands outside your category: emotional drivers and hook patterns vary dramatically by category. Compare beverage to beverage, not beverage to fashion.
• Stopping at "what they're running" without asking "why": every format/landing-page choice is a strategic decision. Reverse-engineer the strategy, don't just clone the format.