Here is how to use the Meta Ad Library for competitor research in one sentence: search a competitor's page at facebook.com/ads/library, filter to active ads, sort by the new impression ranges, then read the start dates. Any ad in a high impression bucket that has been live 60+ days is a proven winner, and you can reverse-engineer it. That is the short version. The rest of this post is the full operator's method, step by step, including the 2026 features most people still haven't noticed.
The Ad Library used to be a window with the curtains half drawn. You could see every ad a competitor was running, but not which ones mattered. A brand with 400 active ads gave you 400 guesses. That changed in January 2026 when Meta rolled impression data out to every ad in the library. A transparency tool quietly became a competitive intelligence weapon, and it still costs nothing. No login, no spy tool subscription.
Impression buckets tell you what a competitor is scaling. Run time tells you it earned the budget. Together they hand you their test results for free.
What changed in the Meta Ad Library in 2026?
Three updates matter for competitor research, and they compound.
1. Impression ranges on every ad. Every ad now shows an impression bucket: under 1K, 1K to 5K, 5K to 10K, 10K to 50K, 50K to 100K, 100K to 500K, 500K to 1M, and 1M+. You can filter and sort by these ranges. This data was previously restricted to political and issue ads; now it covers everything. Ads with fewer than 100 impressions carry a "Low Impression Count" badge, which PPC Land covered when it rolled out, so you can spot a competitor's throwaway tests at a glance.
2. Six platform filters, including Threads and WhatsApp. You can now isolate ads by any single surface in Meta's family: Facebook, Instagram, Messenger, Audience Network, Threads, and WhatsApp. That tells you where a competitor is allocating budget, and whether they are early on new inventory like Threads.
3. EU transparency data. Under the Digital Services Act, any ad that delivered impressions in the EU shows an extra transparency section: who paid, the targeting parameters chosen (age, gender, location), and estimated reach broken down by country. Ads are archived for 1 year after their last EU impression. Meta documents all of this in its Transparency Centre, and covered the EU-specific changes when the DSA requirements landed. Most US-focused marketers have never clicked into it.
How do you use the Meta Ad Library for competitor research, step by step?
This is the exact sequence. Budget about 90 minutes for a full category pass the first time. It gets faster.
Step 1: Build a recon list of 6 to 10 pages
Before you open the library, write down who you are researching. The list should have three tiers: 3 or 4 direct competitors at your size, 2 or 3 category leaders one weight class up, and 1 or 2 brands outside your category selling to the same customer. Direct competitors show you which angles work at your budget level, and the bigger brands show you which formats survive real spend. The adjacent brands are where you find hooks your category hasn't stolen yet.
Record: brand name, exact Facebook page name (brands often run ads from multiple pages, so note them all).
Step 2: Pull each page and set your baseline filters
Go to facebook.com/ads/library. Set the country to your target market, set ad category to "All ads", and search the brand name. Click through to the verified page, not a lookalike. Then filter status to Active.
Record: the total active ad count. This number alone is diagnostic. A DTC brand running 15 active ads is coasting or struggling. A brand running 200+ is feeding the algorithm the creative volume that Meta's Andromeda-era delivery system rewards. While you are here, click the platform filter and note the split: a competitor suddenly running Threads or WhatsApp placements is telling you where their media buyer thinks the cheap reach is.
Step 3: Sort by impressions and flag the top bucket
Apply the impression filter and sort so the highest ranges surface first. Everything in the top one or two buckets is an ad the competitor's own performance data told them to scale. Ignore the long tail wearing the Low Impression Count badge; those are tests that haven't earned budget, or ads being kept alive for retargeting pennies.
Record: the impression bucket for each flagged ad, plus a link to the ad (each one has a shareable library URL). Flag no more than 5 to 8 ads per brand. You are hunting concentration, not volume.
Step 4: Read the start dates
Every ad shows when it started running. This is the oldest trick in ad library research and it still carries the most weight: nobody pays to run a losing ad for 3 months. An ad live for 60 or 90+ days is almost certainly profitable. An ad that appeared and vanished within 2 or 3 weeks probably failed its test. Shopify's own Ad Library guide leans on the same longevity logic, but the impression buckets now let you sharpen it. Cross the two signals:
- Long run + high bucket: a core winner. This is the ad to tear down.
- Long run + low bucket: an always-on retargeting ad. Mine it for objection-handling copy, not hooks.
- New + already high bucket: a fresh winner being scaled aggressively. Study it now, because the whole category will be copying it in 6 weeks.
Record: start date and days live for every flagged ad.
Step 5: Expand the variant stacks
The library groups ads that use multiple versions; click into them and count. Also scan for near-identical ads with different opening lines. When a competitor runs 5+ hook variants on one concept, that concept has already passed testing and they are mining it. That is a validated angle for your category, and it maps directly onto a motivation you can test yourself with the 7 hook angles framework.
Record: the concept, the number of live variants, and the first line of each hook, transcribed word for word. The hooks are the payload of this entire exercise.
Step 6: Flip your country to an EU market for audience data
This is the step almost nobody does. If your competitor advertises in Europe, change the library's country filter to an EU market (Ireland and Germany are good defaults) and open the ad's detail page. The EU transparency section shows the targeting parameters the advertiser chose and estimated reach split by age, gender and country. You cannot get this anywhere else without a paid spy tool, and the spy tools are scraping the same source. UK-delivered ads carry similar disclosure. For political and issue ads you also get spend ranges, archived for 7 years.
Record: age and gender skew of their reach. If a competitor's scaled ads reach 65% women aged 25 to 44 and your media plan assumes an older audience, one of you is wrong, and they have the delivery data.
Step 7: Log everything in a recon sheet
One row per flagged ad, across every brand on your list. Columns: brand, ad link, launch date, days live, impression bucket, platforms, format (creator UGC, static, founder-led, demo), hook transcript, angle, offer, CTA. After 25 or 30 rows the category's playbook stops being a vibe and becomes a dataset: which formats dominate the top buckets, which awareness stage the winners target, which offers are being pushed hardest.
Record: at the bottom of the sheet, write the three patterns that show up most. Those three lines are your next test batch.
Which signals matter, and what should you do about each one?
The library gives you signals, not answers. Here is the translation table.
| Signal | What it means | Action |
|---|---|---|
| Active 90+ days, high impression bucket | Proven profitable concept | Tear it down: hook, angle, format, offer. Brief your own version with a sharper differentiator |
| Launched recently, already at 500K+ | Fresh winner being scaled fast | Study it this week. Test a counter-angle before the category saturates on it |
| One concept, 5+ live hook variants | Angle passed testing; they are mining it | Treat the angle as validated for the category and test your version of it |
| Most actives wear the Low Impression Count badge | Heavy testing phase or small budget | Recheck in 30 days and see which tests survived |
| Ad vanished after 2 to 3 weeks | Probably a failed test | Note the angle as a caution flag, not a ban. Their execution may have failed, not the idea |
| Long run, low impression bucket | Always-on retargeting or evergreen | Mine it for objection handling and proof points, not prospecting hooks |
Key takeaway
A winner in the Ad Library is the intersection of two signals: a high impression bucket (they are scaling it) and a long run time (it keeps earning the budget). Either signal alone can mislead you. Together they are the closest thing to seeing a competitor's dashboard.
A worked teardown: from library finding to test brief
Take a crowded public category: electrolyte drink mixes. Run the seven steps across the top five or six brands in the space and the same profile keeps floating to the top buckets. It is rarely the polished brand film. It is a creator in their kitchen, phone-shot, and the composite winner looks like this: live 90+ days, top impression bucket, running on Facebook and Instagram, hook in the first line ("I drank one of these every morning for 30 days"), body built on a habit-swap narrative (replacing a morning coffee or a sugary sports drink), an ingredient comparison as the proof element, and a bundle offer as the CTA.
Here is how that finding becomes a brief, without cloning the ad:
- Validated angle: habit swap. The category has proven that "replace your current morning drink" beats "hydrate better" as a frame. Keep the angle.
- Format: handheld creator UGC in a domestic setting. The top buckets have spoken; do not brief a studio shoot for the prospecting layer. (If you are torn on format economics, the static vs video breakdown covers when each earns its keep.)
- Hook variants, 3 minimum: a 30-day results frame, a direct callout ("if you drink 3 coffees before noon, watch this"), and a contrarian take on the category leader. Never test a validated angle with a single hook.
- Differentiator: the one claim the incumbent winner cannot make: your ingredient, your price per serving, your taste guarantee. This is the line that stops your version being a photocopy.
- Kill criteria: decide before launch how much spend each variant gets before you cut it. (We wrote up how much to spend per creative test separately.)
At Spark we run this exact recon pass before every client creative batch, and the sheet from step 7 becomes the first tab of the batch planning doc. The consistent surprise is not the hooks. It is the formats. Brand-side teams routinely brief polished content in categories where every scaled ad is deliberately rough, and 90 minutes in the library kills that mistake before it costs a production budget. The recon has changed what we produce for clients more often than it has changed what we write.
What the Ad Library still won't tell you
The tool has real limits, and they change how much weight the signals can carry. You cannot see spend on commercial ads, only on political and issue ads. You cannot see CTR, CPA, ROAS or any conversion data, so a high impression bucket proves scale, not profit; a brand can scale a loser for a quarter before finance notices, which is exactly why run time is the confirming signal. And impressions do not separate prospecting from retargeting, so check the EU targeting breadth where you can: a broad-targeted ad at 1M+ impressions is prospecting, a narrow one is probably remarketing.
The deeper limit is strategic. The library shows you the creative, never the economics behind it. A competitor's winning ad might only work because of their landing page, their AOV, or a subscription margin you do not have. Copy the angle, test it against your own numbers, and let your data decide.
FAQ
Can you see how much a competitor is spending in the Meta Ad Library?
No, not for commercial ads. Spend figures are only published for ads about social issues, elections or politics, which are archived for 7 years. For everything else you infer investment from the impression range bucket, how long the ad has been active, and how many variants the advertiser is running.
Does the Meta Ad Library show ad performance data?
Partially, as of 2026. Every ad now displays an impression range bucket, from under 1K up to 1M+, and ads with fewer than 100 impressions carry a Low Impression Count badge. You still cannot see CTR, ROAS, CPA or conversion data, so the working proxy for a winner is a long run time combined with a high impression bucket.
Is it legal to use the Meta Ad Library to research competitor ads?
Yes. The Ad Library is a public transparency tool Meta built and maintains for exactly this kind of scrutiny, partly driven by regulation like the EU Digital Services Act. Browsing it requires no login and breaks no terms of service. What you should not do is clone a competitor's ad shot for shot, which risks IP claims and teaches you nothing.
Why do some ads show audience and reach data and others do not?
Audience and reach breakdowns come from EU transparency rules under the Digital Services Act, so they only appear on ads that delivered impressions in the EU or associated territories. Those ads show targeting parameters and estimated reach by country, age and gender, and stay archived for 1 year after the last impression. Ads that only ran in markets like the US show no audience data.
Finding their winners is the easy half
The Ad Library gives every operator the same free intelligence, which means the recon itself is not an edge. The edge is what happens next. Three validated angles need to become 15 or 20 briefed creatives inside a fortnight, and the survivors need iterating while your competitor is still storyboarding. That production gap is where most brands lose, and it is the gap Spark exists to close: done-for-you UGC and AI-assisted creative, from recon to hooks to finished ads, on a flat monthly plan. There is more of this in our resources library, or you can see how the studio works.
If you have a recon sheet full of competitor winners and no capacity to out-produce them, that is the conversation to have with us.