How to Use the Meta Ad Library to Find Your Competitors' Winning Ads

The Meta Ad Library is Meta's free, public database of the ads currently running across Facebook, Instagram and Meta's other apps. To find a competitor's winning ads, search for their Page, filter to your country and to active ads, then look for two signals: ads that have been running for weeks or months, and ads the brand has duplicated or varied many times. Brands rarely keep paying for ads that lose money, so these are the best public clues to what works.

This guide covers the searches, the filters, how to read what you find, how to build a swipe file, and how to turn other brands' patterns into original ads without copying anything.

What is the Meta Ad Library?

The Meta Ad Library is a transparency tool at facebook.com/ads/library. It's free, you don't need an ad account to use it, and it covers ads from every advertiser on Meta's platforms. For each ad you can usually see:

• The creative: image, video or carousel, with its primary text, headline and call to action.
• The advertiser's Page.
• An active status and a "Started running on" date.
• The platforms it runs on, such as Facebook and Instagram.
• A Library ID you can use to find it again.
• Whether it has multiple versions, or whether several ads use the same creative.

Ads about social issues, elections or politics carry extra detail, such as spend ranges, and stay searchable long after they stop. Ads shown in the European Union also carry extra transparency information. For everyday commercial ads elsewhere, including India, you mostly see what's live right now.

How to find a competitor's ads, step by step

1. Open the Ad Library and choose a country. Pick the market you sell in, or "All" to see a brand's ads worldwide.
2. Set the ad category to "All ads". The other categories cover political and special-category ads.
3. Search by advertiser. Type the brand's name and choose its Page from the dropdown instead of pressing enter. A plain keyword search matches ad text from every advertiser, which is useful for mapping a category (see below) but noisy when you want one brand.
4. Filter the results, then scroll. Each card shows the creative, the start date and the platforms.
5. Open "See ad details" on anything interesting to see every version, the full copy and where the ad links to.

There's also a shortcut: the Page transparency section of a brand's Facebook Page links straight to its ads in the library.

Search by keyword to find competitors you didn't know about

Keyword search is how you map a category. Search the phrases your customers use, such as "hair growth oil", "protein bars" or "linen shirts", and note the advertisers that keep appearing. Use quotation marks or the exact-phrase option, where available, to cut the noise. Some of them will be brands you never thought of as competitors but who are bidding for the same customers.

Which filters matter for competitor research?

How to tell which competitor ads are working

The library doesn't show performance, so you read it from behaviour. Three signals matter.

1. How long an ad has been running

Every card shows "Started running on". Days running = today's date − the start date. A brand that keeps an ad live for weeks or months is usually getting enough return to justify it. As a working rule, treat 30 days or more as interesting and 60 to 90 days or more as a strong signal, especially when the brand is launching and retiring other ads around it.

It's a signal, not proof. Some brands leave ads running on tiny budgets, some forget to switch them off, and some keep ads live for brand-building reasons. Duplicating an ad also gives it a fresh start date, so a "new" ad may be an old winner relaunched.

2. Variations and duplicates

Cards may show that an ad has multiple versions, or that several ads use the same creative and text. Multiple versions usually mean the brand is testing copy or formats around one idea. Many ads sharing one creative often mean it has been duplicated into more ad sets, campaigns or countries, which is what brands do with winners.

3. What survives a test batch

Brands often launch several new ads in the same week. Note them, check back two to four weeks later, and see which are still active. The survivors show which angle or hook won the brand's own test, which is as close to their results as the library gets.

How to build a swipe file and decode ads into recipes

A swipe file only helps if it's organised enough to act on. A spreadsheet works well, with screenshots in a shared folder.

What to record

Re-check the file monthly: mark what's still running and add new long-runners. Over a few months you'll see which ideas last and which fade, and a fading idea is often a sign of ad fatigue rather than a bad concept.

Decode each ad into its parts

For every ad worth keeping, answer four questions:

• Hook: what stops the scroll? A question, a bold claim, a problem call-out, a surprising visual or a customer quote.
• Layout: how is the frame built? A product hero, a split-screen comparison, a selfie-style UGC video, a text-led static, the product in use or an unboxing.
• Offer: why buy now? A bundle, free shipping, a gift with purchase, a first-order discount or cash on delivery.
• Proof: why believe it? Reviews, certifications, demonstrations, founder expertise, or results shown within Meta's policies.

Then open the landing page from "See ad details". Price anchoring, bundles and guarantees often live on the page rather than in the ad, and they're part of why the ad works.

Write the recipe

A recipe is the structure with every brand-specific element removed. For example: a problem call-out hook, then a split-screen of the routine, then a review quote as proof, then a bundle offer. That recipe belongs to nobody. Rebuild it with your own product photos, your own customers' words and your own offer, and it becomes an original ad.

Worked example: from a competitor's ad to your own

An illustration using a hypothetical haircare brand and a hypothetical coffee brand:

Nothing in your version comes from the original except the structure, and the structure is the part worth learning from.

Test recipes like any other angle, using the 48-hour creative testing framework. For copy structures to pair with them, see Facebook ad copy formulas, and run the finished text through the ad copy checker to make sure the hook isn't cut off on mobile.

What the Ad Library doesn't show you

Knowing the blind spots stops you over-reading what you find:

• Spend and impressions for most commercial ads. You can't tell whether an ad runs on ₹500 a day or ₹5 lakh.
• Results. No CTR, CPA, ROAS or sales figures.
• Targeting and audiences, beyond the extra transparency shown for some regions and ad types.
• Full history. For most commercial ads outside the EU, inactive ads drop out of view, so you can't see everything a brand has tested.
• The rest of the funnel. Email, influencers, marketplaces, retargeting and offline activity may be doing the heavy lifting.
• Their economics. A competitor with a higher margin, a lower cost base or strong repeat purchases can profitably run ads that would lose you money.

So treat everything you find as a hypothesis for your own tests, not a guaranteed winner. Your own first hook rate and cost per purchase will tell you more than any competitor's start date.

Ethics and IP: take the recipe, never the creative

Studying competitors' public ads is normal practice. Copying them is not. Never reuse a competitor's images, video, copy, logos, brand name, music, product shots or customer reviews, and never make a claim that's true of their product but not of yours.

Copying causes three problems. It can infringe copyright or trademarks. It can breach Meta's Advertising Standards on intellectual property, which can get ads rejected or an account restricted. And it's a weak strategy: customers who've seen the original may scroll past a lookalike, or remember the competitor instead of you.

Checklist: a monthly competitor scan

• List five to ten direct competitors and three category keywords.
• Search each Page with your country and active ads selected.
• Log every ad running 30 days or more, with its Library ID and start date.
• Note ads with multiple versions or duplicates, and any new test batches.
• Decode the hook, layout, offer and proof of each long-runner.
• Write two or three recipes with every brand-specific element removed.
• Rebuild the best recipe with your own product, proof and voice, then test it.
• Re-check last month's entries to see which are still running.

How Fly ROAS automates competitor research

Fly ROAS's competitor radar runs this scan for you. It finds competitors' long-running ads in the Meta Ad Library and decodes the hooks, layouts and angles that keep working into recipes, without ever copying their images, copy, logos or brand names. It then turns those recipes into on-brand image and video ads, in 95+ languages, built from your own product photos, with a pre-flight check that flags any ad where the label, text or colours drift from your photo. One click builds the campaign in your own ad account, paused by default until you choose to go live. See how it works.

The bottom line

The Meta Ad Library won't tell you what a competitor earns, but it does show what they keep paying for, and that's the next best thing. Read start dates and variations as signals, decode the patterns into recipes, and rebuild them honestly with your own product and proof. Then let your own tests decide what works for your brand.