Every product you are considering has already been tested by someone else — with real money, on the open market, in public. The Meta ad library, browsable by anyone at facebook.com/ads/library, lists every ad currently running across Facebook and Instagram: who placed it, what it says, and when it started running. No account, no cost.
Most sellers treat it as a mood board. They scroll competitor creatives for inspiration, feel vaguely reassured, and close the tab. That is a waste of the most honest research surface available, because the library does not show you what people say about a product — it shows you what businesses are paying, day after day, to sell it.
What follows is a field method for reading it: how to set up the search, the three things to record for every relevant ad, and how to turn a wall of competitor ads into a judgment about the market itself.
Why the ad library is the highest-signal free research surface
Almost every public signal in e-commerce research is cheap to manufacture. Views can be bought. Likes and comments can be botted. Follower counts are decoration. Even a viral clip only tells you that a video performed, not that a product sells.
A running ad is different in kind. It is a recurring bill that a business chooses to keep paying. Every day an ad stays live, someone with access to their own sales numbers has looked at the cost and decided it is worth paying again.
An ad is a bet. An old ad is a bet that keeps being re-placed by someone who can see the results.
That makes longevity a survival signal. Nobody funds a losing ad through two monthly budget reviews out of sentiment; ads that fail to return their spend get switched off within days. The library will never show you a competitor's revenue, but it shows you something nearly as good — which bets they keep re-placing.
Setting up the search properly
Set the country first, before you type anything. An angle that converts in the United States may lean on pains, price points, or shipping expectations that do not transfer to your market — and an advertiser who cannot ship to your customers is not your competitor, only your teacher. Research the country you will actually sell into; browse other markets afterwards for angle ideas.
Set the ad category to All ads. The library's defaults were built for political transparency, and the political filters will hide everything you care about.
Then choose your mode deliberately. Keyword mode is for angle discovery: search the pain or the product noun — "posture", "knee pain", "portable blender" — and the library returns every active ad whose text contains it. This is how you map who is spending in a market you are only beginning to understand.
Advertiser mode is for teardowns. Once a store keeps surfacing in your keyword results, click through to its page and view everything it is currently running. The question flips from "who is in this market" to "how seriously is this one company committed to it".
One habit worth keeping: search the problem before the product. Customers describe pains — "dog hair everywhere", "standing desk fatigue" — and the best advertisers write copy in the customer's words. Product-noun searches find the obvious competitors; pain searches find the smart ones.
Three things to record for every relevant ad
Scrolling informs; recording decides. For every ad genuinely competing in your market, write down three things.
Ad age
Every ad in the library carries a public start date. This one field does more work than everything else combined.
Field judgment, not laboratory science: an ad that has run 60 days or more is very likely paying for itself, because nobody carries a money-losing line item through two budget cycles. Thirty days is a respectable sign. Under two weeks means nothing yet — that is a test, not a business.
This is also why a wall of very young ads should cool your enthusiasm rather than heat it. Twenty advertisers who all started in the past ten days is not proof of a market; it is proof that a video went viral and everyone downloaded the same supplier catalogue.
Variation clusters
Count how many active creatives an advertiser runs for a single product. The number roughly tracks conviction.
One lonely creative is a probe — someone spending the minimum to see if anything twitches. Eight, fifteen, thirty variations on the same product is a scaling operation: a team producing fresh hooks weekly because the product has already earned a real budget. When you find an old cluster — many creatives, oldest start date months back — you have found a product that is very probably profitable at scale.
Angle and offer
Record, verbatim, the pain each ad sells and the price anchor it shows. Not your paraphrase — the actual words. "Wake up without lower back pain" is a different market from "fix your posture at your desk", even when both sell the same brace.
This is the market educating you for free. The advertiser has already paid to test dozens of phrasings; the copy still running is the distilled survivor. The hooks, the mechanism claims, the 39.95-marked-down-from-79 anchors — all of it is conversion data purchased by someone else and published where you can read it.
Reading ad density across a whole market
Individual ads tell you about competitors. The pattern across all of them tells you about the market.
A few advertisers with old ads is a stable, proven market. Demand is durable — incumbents have been paying to serve it for months or years — but you will need a genuinely different angle or audience to displace them, because they hold a data head start you cannot buy.
Many advertisers with mostly young ads is a gold rush mid-burn. The auction gets more expensive weekly, the dominant angle is being copied to exhaustion, and by the time your store ships the wave may already be receding. This pattern is behind most saturation, and it is why counting competitors without checking ad ages misleads.
Zero advertisers is the ambiguous one: the market is either dead or undiscovered, and the library alone cannot tell you which. The tiebreaker is cross-checking the other signals. If search interest is flat or falling and nothing is selling on Amazon, the silence means dead. If searches are climbing and retail proof exists but nobody is advertising yet, you may be genuinely early — rare, but exactly the situation this method exists to catch.
What the library will not show you
Honesty about the tool's limits: the library shows what is running, not how well. There are no engagement counts, no spend figures, no click or conversion numbers. Ads shown in the EU carry rough reach ranges under transparency rules, but reach measures audience size, not profit.
This gap creates a temptation — dashboards promising "engagement metrics" for competitor ads, harvested from sources of unknown provenance and unknown freshness. Treat those numbers with suspicion. You cannot audit how they were collected, when, or from which country's feed.
The honest compensation is already built into this method: longevity and cluster size are your performance proxies. A 90-day-old ad backed by twelve active variations tells you more about profitability than any like count, because likes measure applause and longevity measures the willingness to keep paying. Applause is free. Ads are not.
A 15-minute worksheet
The whole method compresses to a quarter of an hour per market. The sequence:
- Open the ad library. Set the country to your target market and the category to All ads.
- Search the pain phrase first ("knee pain", "posture"), then the product noun ("knee sleeve", "posture corrector"). Skim both result sets and note every advertiser that appears more than once.
- For the three to five most serious advertisers, switch to their advertiser page and load their full active ad list.
- Fill in one row per advertiser: advertiser, oldest ad start date, cluster size, angle (verbatim), price anchor.
- Read the pattern. Old ads and big clusters mean proven demand with entrenched competition. Uniformly young ads mean a rush. An empty table means cross-check search and retail before concluding anything.
Folding this into a full read
Ad evidence is one signal out of four — it needs corroboration from search demand, retail proof, and what real customers say in communities before it becomes a verdict. That cross-checking is the part Kestrel automates: it runs this same read against the Meta ad library alongside Google search demand, Amazon retail proof, and community chatter, then scores the market 0–100 — Hot, Promising, or Weak — with each piece of evidence itemized. The specimen report shows how ad longevity and density surface in a real breakdown, and the free tier covers 20 market scans with no card. The worksheet above gives you the same ad-side judgment by hand; the tool does it in seconds and adds the other three legs.
Reading spend, not applause
The Meta ad library is the closest thing e-commerce offers to reading a competitor's books. It will not hand you their revenue, but it shows you which bets they keep funding, in which words, at which prices — and that is most of what you need to know before risking your own money.
The discipline matters more than the tool. Same fields, every market, written down before you commit: ad age, cluster size, angle, anchor. Ten minutes of recording beats an hour of scrolling, and a market you can defend on paper is a market you can enter with your eyes open.
