How to Find Winning Ads in the Meta Ad Library

Every brand's live Meta ads are public, which means the hard part was never finding ads. It is working out which of the forty creatives a competitor is running right now are the ones actually earning their place, and which are tests that will be switched off next week.

How do you find winning ads? Read the signals the advertiser cannot hide: how long an ad has been running, how many people it has reached where Meta publishes that figure, the two divided into a daily reach rate, how many variants of the same idea are in market, and how concentrated the landing pages are. An ad that has run for months and reaches people quickly is one someone keeps choosing to pay for.

This guide is the method, in the order you should apply it, using only data that is free and public.

Why the question is hard: Meta publishes no commercial spend

For a commercial ad, Meta publishes no spend figure and no impression count anywhere. Those numbers exist in the Ad Library only for ads about social issues, elections or politics. So any guide that promises you a competitor's budget is either modelling it or guessing, and the honest version of this work starts by accepting that the budget is not on the table.

What is on the table is delivered reach. Meta publishes delivered reach for ads shown in the EU and the UK, along with a breakdown of that reach by age, gender and country. Meta's own field reference states the availability plainly: the reach total is available only for ads delivered to the EU, and the age, country and gender breakdown is available for ads delivered to the UK and the EU. The driver is the EU's Digital Services Act, and the UK figures sit in the same transparency panel. Meta's Transparency Center describes the scope of its Ad Library tools in the same terms.

That asymmetry is the whole opportunity. Spend is hidden, reach is published, and reach plus time is a better signal than either on its own.

The signals, strongest first

1. Days running

Run length is the signal every experienced buyer reaches for first, and the logic is simple: Meta's delivery system consolidates spend onto what performs, and an advertiser switches off what does not. An ad that has been live for four months has survived every optimisation decision made in those four months. A creative that launched three days ago has survived nothing.

The threshold worth using is ninety days. Below it you are looking at ads that may still be in test; at ninety days and above you are looking at something the advertiser has decided about. In Wilow's free Meta Ads Library search, an ad live for ninety days or more carries a badge with its day count, and opening any ad shows the date it started running alongside how many days it has been live.

2. Delivered reach, where Meta publishes it

Reach is the number of people an ad has been shown to, and it is Meta's own figure rather than an estimate. It is published for ads delivered in the EU and the UK, so a brand that advertises only in the United States has no reach figures at all, and a brand running the same creative in both places has figures for the European and British delivery only.

Reach on its own is still a blunt instrument, because a large number can mean a strong ad or simply an old one. It becomes useful the moment you put it against time.

3. Daily reach: the two signals together

Daily reach is people reached divided by days running, and it is the single most useful number in this whole exercise. It separates an ad that has quietly accumulated reach over six months from one that is reaching a large audience right now, which is the difference between a creative that works and a creative the advertiser is currently backing.

Wilow's public leaderboards sort brands by exactly this: each row shows one brand's best live ad as a reach-per-day rate, with the total people reached and the number of days live underneath it, and ads past ninety days flagged as proven. If you take one habit from this guide, make it ranking by daily reach rather than by reach.

You can turn reach into a spend estimate, and it is worth being precise about what that is. Estimated spend is reach multiplied by a CPM you choose, divided by a thousand. The CPM is your assumption, not Meta's data, which is why Wilow puts it on a slider you control rather than printing a single authoritative-looking number. Wilow's note on how the spend estimate works sets out the arithmetic and its limits.

4. How many variants of one idea

Advertisers do not iterate on ads that are failing. When you see the same hook in five cuts, the same offer in four thumbnail treatments, or one script running in both square and vertical, you are looking at a concept the brand has decided to invest production time in. One ad is a test. Five variants of one ad is a verdict.

Count variants before you count ads. A brand with sixty live ads that are all versions of three ideas is telling you about three ideas, and that is a more useful read than the number sixty.

5. Where the ads send people

The destination is the part most people skip, and it is where the strategy becomes legible. A brand pointing thirty ads at one product page is scaling a proven offer. A brand pointing them at twelve different pages is still looking for the offer that works. In the signed-in product, Wilow's competitor tracking maps a brand's ads to their destinations with the width of each ribbon set by how many ads point at that page, so the concentration is visible without counting anything by hand.

Landing pages also catch something the creative alone will not tell you: whether an ad you admire is selling the thing you sell. A beautiful ad for a subscription quiz funnel is not a template for a one-off product page.

6. Who the brand is advertising with

Partnership and collaboration ads, the creatives a brand runs jointly with a creator or another brand, carry a different signal from the brand's own output. They cost more to produce and they usually follow a result, so a brand that has moved from in-house creative to a run of creator collaborations has probably found that the collaborations work. Wilow's library detects these and flags them, which matters because a co-branded ad otherwise blends into a brand's feed and gets read as ordinary output.

This is also where you learn who your competitor thinks their audience trusts, which is often a faster route to an audience insight than anything in the creative itself.

Running the check, step by step

The method takes about ten minutes per brand once you have done it twice.

1. Pull the brand's live ads. Search the brand by name and pick the market you care about. Wilow's free library search asks for an advertiser and a country and needs no account; Meta's own Ad Library works too, and a side-by-side is worth doing once to see which figures each surface gives you.
2. Sort by what is old, not what is new. Find the ads flagged at ninety days and above and read those first. They are the brand's answer to the question you are asking.
3. Note the daily reach of the top few. Where reach is published, divide it by days live. You are looking for ads that are both old and fast, which is the rarest and most informative combination on the page.
4. Group the rest into concepts. Ignore ad count. Write down the three to five distinct ideas in market and how many executions each one has.
5. Follow the links. Check how many destinations the ads share and what the page does when a visitor arrives.
6. Write down the mechanism, not the ad. The output of this exercise is a sentence like "their best-performing creative is a founder talking to camera about a specific objection, running for five months, pointed at a single comparison page." That sentence is reusable. A screenshot is not.

When a brand has no reach figures

This is the case no guide covers, and you will hit it constantly if you study American brands. A brand that delivers no ads in the EU or the UK has no published reach, so three of the signals above come back empty. You are not stuck, but you are working with less:

• Run length still works. Start dates are published for every ad in every market, so the ninety-day test is always available.
• Variant counts still work, and they get more important, because they become your main read on what the advertiser believes.
• Destinations still work, and so does the shape of the creative itself.
• Check whether the brand has any European delivery at all before concluding its reach is zero. A brand running a small UK campaign alongside a large American one will publish reach for the UK slice, and that slice is often enough to rank its creatives against each other even though it understates the total audience.

Read testing pace, not just one ad

A single winning ad tells you what worked once. How fast a brand launches new creative tells you how it operates, and that is the more durable intelligence. A competitor shipping fifteen new ads a week is running a volume strategy, and beating them on polish is the wrong plan. A competitor shipping two a month has a small number of ideas you can study closely.

Wilow's in-app library organises a brand's currently-running ads by recency, grouping new launches into trailing seven-day slices with a count on each one, and keeps a separate row of ads live for ninety days or more ordered longest-running first. A per-brand velocity view charts one bar per day for how many ads that brand launched, with quiet days shown as gaps. Both answer pace rather than quality, and pace is what tells you which game you are being asked to play.

Turning the read into a brief

A list of winning ads is not yet useful to whoever has to make the next one. What converts research into production is describing each winner along the same handful of dimensions every time, so that patterns across a competitor's set become visible and a brief can be written from the pattern rather than from one ad.

The dimensions worth capturing are the ones a creative team actually has to decide on: who the ad speaks to, what offer it leads with, what format and style it uses, what objection it answers, and what it asks the viewer to do. Wilow runs this reading automatically on a brand's library through a framework called POSTER, which describes a brand's advertising along dimensions including target audience, offer type and ad style rather than stopping at the visual label. The POSTER framework guide sets out what each dimension covers.

Done by hand, this is a spreadsheet with one row per winning ad and one column per dimension, and it is worth building even if you never automate it. The moment you have fifteen rows, the repeated values in each column are your competitor's strategy written out, and the empty cells are the angles nobody in your category is running.

What this method cannot tell you

Three limits, stated plainly, because a method oversold is a method that eventually embarrasses you in a meeting.

• Reach counts people while CPM prices impressions. A spend estimate built from reach and a CPM is therefore a floor rather than a best guess, since one person may have been shown the ad several times.
• Coverage is EU and UK delivery only. Published reach exists for those markets; everywhere else, the figure does not exist and no tool can produce it.
• One ad's reach is not a budget. It is one creative's audience in one market, and rolling it up into an account-level spend figure stacks assumption on assumption.

None of that undermines the exercise. It means the output is a ranked read on which creatives an advertiser is backing, not a financial statement, and a ranked read is what you needed in order to decide what to make next.

Common questions

How long does an ad have to run to count as a winner?

Ninety days is the practical threshold. It is long enough that the ad has survived the optimisation decisions an advertiser makes while spending, and it is the point at which Wilow flags an ad as proven. Below thirty days, treat an ad as a test rather than a result.

Can I see how much a competitor spends on Meta ads?

Not as a published figure. Meta reports spend for ads about social issues, elections and politics only; for a commercial ad there is no spend anywhere in the Ad Library. What you can do is model it from published reach and a CPM you choose, and label the result as an estimate every time you show it to anyone.

Is there a free way to find winning ads?

Yes. Meta's Ad Library is free and so is Wilow's library search, which needs no account and shows the reach each ad has delivered where Meta publishes it. The paid part of this category is workflow, not access: saving, tagging, scoring and comparing at scale.

Why do some ads show reach and others show nothing?

Because reach is published for ads delivered in the EU and the UK. An ad that ran only in the United States, Canada or Australia has no published reach figure, and a blank is a statement about where the ad was delivered rather than about how well it performed.

What should I do with a winning ad once I have found it?

Extract the mechanism rather than the execution. Name the audience it addresses, the objection it answers, the offer it leads with and the format it uses, then rebuild that structure in your own brand's voice with your own footage. Copying the creative gets you a worse version of someone else's ad; copying the mechanism gets you your own.

If you want the same read on where a brand's ads live before you judge them, the companion guide on how to see other companies' Facebook ads covers the finding half of the job.