Swipe file & analysis · Facebook Ad Examples

Facebook ad examples: browse live Facebook advertising examples and sample Facebook ads

Facebook ad examples are easiest to find at the source. Meta's own Ad Library at facebook.com/ads/library holds every ad currently running on Facebook and Instagram, it is free and public, and you can search it by brand name or by keyword. That is also the difference between a live example and the ones in a listicle: ordinary commercial ads are only listed while they are still running, so a curated roundup of screenshots from three years ago is showing you campaigns that were switched off long ago.

The problem with browsing the library raw is volume and noise. A search for one furniture brand can return hundreds of near-identical variants, and nothing on the page tells you which of them is carrying the budget. What you actually want out of an example is the pattern behind it: the hook in the first line, the angle it argues, the format it runs in, and some evidence the advertiser is still paying for it. AdEye does the sortable, retained version of this job for Google advertising rather than Meta: it reads the Google Ads Transparency Center, tags each ad by hook, angle and format, and shows run length, so the creatives a company keeps funding rise above the ones it tested for a week and killed. For the Facebook and Instagram examples themselves, Meta's library is the source and this page explains how to get the most out of it.

Use examples the way a media buyer does, not the way a moodboard does. Search your whole category rather than the three competitors you can name, filter to the format you are about to brief, and read twenty ads before you write one. Save the ones worth borrowing structure from into a swipe file, because the reference disappears when the ad comes down. All of it is public transparency data the platforms publish deliberately, so there is nothing underhanded about studying it.

Ad Intelligence Feed

Live Google data

Enter a brand or a domain

Searching the Google Ads Transparency Center...

Try:

Public Google Ads Transparency data. Nothing behind a login.

Open your swipe file

Google Ads Transparency has no ads on record for that search. Google indexes ads by advertiser, so the brand's exact domain works best — for example nike.com.

Google Ads Transparency has no ads on record for . That is an answer about the brand, not a problem with the search: they are not running Google ads that Google publishes, or they advertise under a different registered name.

We could not reach Google Ads Transparency just now, so this is not a result about that brand, we simply could not look. Please try again in a moment. This search has not been counted against your allowance.

Open your swipe file

free left ·

See it run

Live Google Ads Transparency data · Run length · Public data only

GOOGLE ADS TRANSPARENCY

Run length on every ad

AI-tagged hooks & angles

Why it works

How to find Facebook ad examples that are actually still running

A live example is worth ten from a listicle

Meta lists ordinary commercial ads only while they are actually delivering. The moment an advertiser turns a campaign off, the ad drops out of the library. That rule is inconvenient for archiving and excellent for judging quality, because anything you can still see today is something a company is still choosing to pay for right now. Most published collections of Facebook ad examples were assembled once and never revisited, so a good share of what they show you has not run in years. Political and social-issue ads are the exception and stay listed for seven years, and ads delivered in the EU and UK are kept for one year after their last impression.

Search the category, not just the competitor

Typing in the three brands you already watch gives you the examples you half expect. The keyword search runs across ad text from every advertiser, so searching a category term, an offer type or a common objection surfaces companies you did not know were bidding for the same customer, plus the smaller advertisers who often test harder than the market leader. That is where genuinely new angles come from. Note that Meta's keyword search is a feature of Meta's library, not of AdEye: AdEye searches Google advertisers by brand name or domain, because Google indexes its ad archive by advertiser rather than by ad text.

Read the format before you copy the creative

Meta names a specific set of formats and each one asks a different thing of your copy. A carousel holds up to 10 images or videos, and every card gets its own link, so it suits multi-product and step-by-step arguments. Collection ads open into a fast-loading browsing experience built for catalogs. Video and image are the workhorses in feed, slideshow builds a light video out of stills, and playable gives app buyers a try-before-download preview. Copying a carousel hook into a single-image ad usually fails because the structure the hook depended on is gone.

What it surfaces

Every ad card carries the hook, angle and a working signal

You type a brand or a domain. AdEye pulls that advertiser's live public Google ads into one feed and AI-tags each creative on the paid plans, so you read the strategy and spot the long-running winners before you save a single one.

Read live Facebook ad examples straight from Meta's library
Search Google advertisers by brand name or domain
Filter Google ads down to the format you are about to brief
See run length on every Google ad
Compare how one advertiser words the same offer across ads
Read hook and angle tags instead of guessing the pattern
Save keepers to a swipe file before the ad comes down
See how long each competitor ad has been running

GOOGLE Top performer

image

META PLATFORMS, INC. · RUNNING 1,765D

Why AdEye

One feed, the AI read, the working signal in view

Not a stale doc full of screenshots. One query for a brand, the run length on every creative, and a swipe file worth reopening. Public ad-library data only.

Every platform at once

The Google Ads Transparency Center searched by brand or domain, so competitor ad research is one query instead of clicking through Google's own viewer.

The AI reads the why

Every creative is tagged with its hook, angle and format, and a working signal flags the long-runners and scalers, so you see why an ad works, not just what it ran.

A swipe file that stays current

Save winners to boards, each keeping its run length and a link back to the original creative on Google, so the file is still useful months later.

Good questions

Questions about facebook ad examples

The Meta Ad Library at facebook.com/ads/library is the primary source. It is free, needs no advertising account, and lists every ad currently running on Facebook and Instagram. Search a brand name to see one advertiser, or use keyword search to pull ads from across a whole category. AdEye does not index Meta's library, so for Facebook examples use Meta directly; AdEye adds search, tagging and saving for Google ads.

The useful ones are not the prettiest, they are the ones still running after months. That means the advertiser tested the concept and kept paying for it. A single beautiful ad that launched last week proves nothing. Sort by run length before you judge on design.

Skip brand names and search the language your market uses: the category term, the offer, the objection, or a phrase your customers say. Keyword search reads the ad text itself, so it returns advertisers you have never heard of alongside the obvious ones. Then narrow by format and by how long each ad has been live.

Yes, and it is fully sanctioned. Meta publishes every active ad in the Ad Library on purpose, so searching a competitor returns their live Facebook and Instagram creatives with the copy, the format and the date each ad started. The advertiser is not notified and nothing about looking breaks a rule.

Browsing the Meta Ad Library costs nothing. What it does not give you is search across platforms, tagging, filtering by how long an ad has run, exports or saved boards, so the cost shows up as your time. Tools like AdEye charge for that layer, starting at $39 per month.

Ordinary commercial ads are listed only while they are running and vanish once the advertiser stops them. Ads about social issues, elections or politics are archived for seven years whether active or not. Ads delivered in the EU and UK are kept for one year after their last impression. So competitive history has to be captured while it is live.

Meta names image, video, carousel, collection, slideshow and playable, plus the Instant Experience that a tap can open into. Carousel is available on Facebook, Instagram, Messenger and Audience Network. Each format changes how the copy has to be structured, which is why format is worth filtering on before you collect examples.

A carousel holds up to 10 images or videos in a single ad, and each card carries its own link. That makes it the format of choice for showing several products, walking through a process step by step, or stacking a sequence of proof points that would not fit in one image.

A specific hook in the first line, one clear angle rather than four, a format that matches the argument, and a promise the landing page actually keeps. Beyond that, the honest answer from the data side is longevity: an ad a company has funded for three months has passed a test no critique of the design can replicate.

Nobody outside the advertiser can see conversion rates, so treat any page claiming to show you conversion numbers with suspicion. What is publicly observable is run length. Ads that have run a long time are the closest public proxy for converting ads that exists.

Strong copy usually opens on the reader rather than the product: a symptom they recognize, a number that stops the scroll, or a direct question. The body earns the claim with one specific proof point, and the call to action names what happens next. Pull ten examples from your own category and the shared pattern will be obvious.

Copy the structure, never the asset. The hook shape, the angle, the format choice and the offer framing are all fair game and are how the industry has always worked. Lifting their images, video, logo or copy word for word risks a copyright or trademark claim and gets you an ad that sells their brand, not yours.

No. The Ad Library shows no spend, budget, click-through rate or conversion figure for commercial ads. Political and social-issue ads carry spend ranges, and EU-delivered ads carry impression ranges, but for an ordinary ecommerce or SaaS advertiser there is no public number. Anyone quoting a precise competitor spend is modelling an estimate.

Yes. Meta publishes Facebook and Instagram ads through the same library, so an advertiser search returns both, with the placement shown per ad. AdEye does not read that library. It covers the same brand's Google advertising instead, which is useful as a second angle on the one core idea most brands adapt per platform.

Search the library by market rather than by brokerage, because the ads that matter are the local ones. Filter to your metro area and you will see the pattern quickly: listing carousels, single-image ads for open houses, and lead forms offering a valuation or a buyer guide. Local advertisers turn campaigns over fast, so what is live this month is the only useful reference.

Lead ads open a form inside Facebook instead of sending the click to a landing page, so the copy has to do all the qualifying up front. The examples that keep running tend to name the deliverable plainly, state what happens after the form, and set an expectation about follow-up. You can spot them in the library by the form-style call to action on the creative.

They make the point before the sound is on and before the viewer decides to scroll. Meta advises using movement and sound to show a product quickly, and the ads that survive in the library follow it: the hook lands in the first two seconds, captions carry the message on mute, and the offer appears well before the end.

Twenty from your own category beats two hundred from a general roundup. At around twenty you start seeing the same three or four angles repeat, which tells you what the market has already validated and, more usefully, which angle nobody is running. That gap is normally where the next test should go.

Read every question we get

Explore more