Every ad your competitors are running right now is public. All of it — every image, every headline, every video, in every country they run in — sitting in a free, searchable library that has been open since 2019.
Almost nobody uses it well.
Not because it’s hidden, and not because it’s hard to open, but because a grid of four hundred competitor ads is not an insight. It’s homework. The brands that get value out of ad libraries have a method for deciding which ads to read, a checklist for what to read them for, and a habit of turning what they find into something they can actually test. This is that method.
Why competitor ads are the best free research you have
It is the only place performance leaks
Nobody publishes their cost per purchase. But every brand publishes, involuntarily and continuously, the creative decisions they are making with their money — and those decisions are the output of tests they paid for.
A competitor who has run the same ad structure for five months has told you something they’d never say out loud: it works. A competitor who has quietly replaced their whole creative library twice this quarter has told you something too. You cannot buy that signal, and none of the paid research tools has better access to it than you do — they read the same public library you can, and charge you for the interface.
The inference is imperfect and worth stating plainly: duration is a proxy, not a metric. A well-funded brand can run an unprofitable ad for a while out of inertia or brand budget. But across twenty ads and five competitors, the noise cancels and the pattern is real.
It beats surveys, and it beats instinct
Ask customers what makes them buy and they will tell you a story about quality and value. Watch what the market is actually spending money to say and you get a much less flattering, much more useful picture: which objection everyone is fighting, which proof point everyone reaches for, and which price framing the category has settled on.
Your own instinct is the weakest of the three inputs, because you have seen your product every day for two years and your buyer has seen it for two seconds. You are, structurally, the person least able to judge what is obvious about your own product.
There is a second-order benefit too. Reading a category’s ads regularly recalibrates your sense of what “normal” looks like — how aggressive the discounting is, how direct the copy is, how much explanation the product needs. Most brands who feel their ads underperform are simply running creative pitched for a different category’s conventions.
Where to actually look
The Meta Ad Library
The primary source. Go to facebook.com/ads/library, set the country, set the category to “All ads”, and search by advertiser name. You get every ad currently active for that page across Facebook and Instagram, with the date each one started running.
No account is needed. Two things are worth knowing up front: spend and impression data is only published for political and social-issue ads, so for ecommerce you’re inferring from duration rather than reading performance directly, and the library shows active ads, so anything a brand switched off last week is already gone. That last point is the argument for checking weekly rather than quarterly.
A few practical notes. Search by page name rather than by product keyword — keyword search in the library is unreliable and misses most of what you want. Check more than one country if your competitor sells internationally, because the creative frequently differs and the differences are informative. And use the library’s own detail view rather than screenshotting the grid: it shows every active variant of an ad under one entry, which is exactly the repetition signal covered below.
The TikTok Creative Center
TikTok’s equivalent is more generous in one respect: it surfaces top-performing ads with engagement signal attached, filterable by region, industry and time window. It is less useful for tracking a specific competitor and more useful for reading a category — what formats, hooks and pacing are working right now in your vertical, across brands you may not have heard of.
If your category skews younger or is heavily creator-driven, read TikTok first and Meta second. The ideas usually appear there before they appear in a Meta feed, often by a month or two.
The filters worth using are industry and time window. Set the window short — the last 7 or 30 days — because the point is to catch what is working now, not to admire an ad that won last spring. And read the top-performing list for structure rather than trying to reproduce a specific execution; TikTok creative ages fast, but the underlying formats age slowly.
The brand’s own organic content
Underrated, and free. Many brands test hooks organically before promoting the winners, so a brand’s recent TikTok or Reels posts often show you the raw version of an idea before the marketing team sanded the edges off it.
Cross-reference: when you find a hook in someone’s organic feed that later shows up as a paid ad, you have caught a winner in the act of being promoted, and you know the shape of the idea before and after it was polished.
Comments are worth two minutes as well. The objections people raise under a competitor’s organic post are the objections your ad will need to answer, phrased in the words your buyers actually use. That is free copy research, and it is more honest than any survey you could run.
How to find the ads worth reading
You are looking for a shortlist of about twenty ads, not a comprehensive archive. Three filters get you there.
Sort by how long it has been running
This is the single most useful signal available and it costs nothing. Ad accounts are ruthless: creative that loses money gets switched off in days. An ad still running after three or four months in a competitive category is almost certainly profitable.
Scan the start dates, and put anything older than about sixty days at the top of your list. You have just filtered several hundred ads down to a couple of dozen that have survived contact with a real audience.
Calibrate the threshold to your category’s tempo. In fast-moving categories with heavy creative turnover — supplements, fashion, anything creator-led — thirty days is already a meaningful survival. In slower, higher-consideration categories, sixty to ninety is the right bar. If everything in a competitor’s library started last week, that itself is the finding.
Look for structures that repeat
The second signal is repetition. When a brand runs six variations of the same layout — same hook shape, same proof placement, different product or different colorway — they are not being lazy. They are scaling a winner.
A repeated structure is a stronger signal than a single long-running ad, because it means the brand has enough confidence in the pattern to invest production time in extending it. Note the pattern, not the specific asset.
The most valuable version of this is a structure repeated across different brands. When three independent competitors have all converged on the same layout and the same proof placement, you are looking at something the category has collectively discovered, and it is very unlikely all three are wrong.
Watch for the sudden library refresh
The third signal is a change in tempo. If a competitor’s library was stable for months and has suddenly turned over completely, something changed — a new offer, a new creative team, a new product line, or performance falling off a cliff.
Either way it is worth ten minutes. New creative from an established brand is either their next winning direction or their panic, and both are informative.
Keep a one-line note per competitor per week — how many active ads, roughly what’s new. It takes ninety seconds and it turns a static snapshot into a trend line. Six weeks of those notes will tell you more about your category’s direction than any industry report.
What to read each ad for
Not “is this good”. Six specific things, and it takes about ninety seconds an ad once you’re practiced.
| What to note | Why it matters |
|---|---|
| Hook, word for word | The phrasing is most of the value; paraphrasing loses it |
| Awareness stage | Reveals which part of the funnel the category is fighting over |
| Objection answered | Tells you what the market believes the friction is |
| Proof type | Categories converge on one, and the wrong one reads as amateur |
| Offer and its timing | Discount-first and discount-last are different margin strategies |
| Format and length | Tells you what buyers need to see before they’ll believe you |
Keep the notes in one place, in your own words. The artifact you want at the end of a month is not a folder of screenshots — it’s a page of patterns.
The hook and the awareness stage
What does the first line or the first frame do to interrupt a scroll? Write it down verbatim — the exact words, not a paraphrase, because the phrasing is most of the value.
Then ask who it’s talking to. Someone who doesn’t yet know the problem exists needs a completely different opening from someone already comparing two brands. Most ecommerce ad accounts over-serve one stage badly — usually the most-aware end, because it’s the easiest to write — and the fastest way to find a gap in your own account is to notice which stage your competitors are covering that you aren’t.
A rough test: read your own five best-performing ads and ask what the reader must already believe for the ad to make sense. If all five assume the reader already wants the product, your account has no top of funnel, and no amount of creative polish will fix that.
The objection and the proof
Every ad is answering a reason not to buy: too expensive, won’t work for me, don’t trust the brand, tried something like it before. Identify which one, and you have identified what the category believes its main friction is.
Then look at what carries the claim. A number, a review screenshot, a demo, a before-and-after, a comparison table, a founder to camera? Proof types are surprisingly category-specific and surprisingly durable, and using the wrong one reads as amateurish even when the claim is true. Skincare runs on before-and-afters; tools run on demos; anything expensive runs on reviews and returns policies.
The offer and the format
What is actually being asked for, and when in the ad does the ask arrive? A discount in the first frame and a discount in the last frame are different strategies with different downstream margins.
Note the format too — static, carousel, or video, and if video, its length. Categories converge on formats for reasons, and a category where every winning ad is a 20-second demo is telling you something specific about what buyers need to see before they’ll believe you.
If you find the split is roughly even between statics and video, that’s good news: it means you can test messages cheaply as statics and only invest in video once you know which message is worth filming.
Turning a folder of screenshots into angles
This is where most competitor research dies. You have twenty ads read and nothing to do on Monday.
Group by claim, not by brand
Sort your notes by what the ads are arguing, ignoring who ran them. You’ll typically find three to six recurring arguments across the category — “faster than the alternative”, “the one dermatologists use”, “finally, one that fits”, and so on.
Recurring arguments across multiple independent advertisers are the highest-value output of the whole exercise. Each is a hypothesis several competitors are funding simultaneously, which is about as close to validated as free research gets.
Equally interesting is the argument nobody in your category is making. Sometimes that’s because it doesn’t work and everyone has already found out. Sometimes it’s a genuine gap, and it is the cheapest differentiated position available to you. You’ll only know by testing it, but you’d never have thought to ask without the grouping exercise.
Write each one as a testable angle
Convert each argument into something you could brief: a hook, a message, the awareness stage it targets, the product it applies to, and the proof you would need. If you can’t name the proof you’d use, the angle isn’t ready — that’s the gap to fix before you spend anything.
Aim for six. Enough to cover the awareness spectrum, few enough that a small ad account can give each one real budget rather than a rounding error. Twenty angles at $20 a day teaches you nothing except that small samples are small.
Take the structure, never the claim
This is the line, and it matters both legally and commercially. Their layout, their objection-handling, their pacing, their offer shape: all fair game, all essentially the accumulated learning of your category.
Their photography, their copy and their claims: not yours. A competitor asserting their serum is clinically proven is a fact about their marketing, not about your product. Repeat it and you’ve inherited a claim you may have to defend to a platform reviewer, a regulator, or an unhappy customer — none of whom will be interested in where you found it.
The practical version of this rule: write the angle down as a sentence about what the ad is doing, not as a copy-paste of what it says. “Answers the ‘won’t work on my skin type’ objection with a four-week before-and-after” is a brief. The competitor’s headline is not.
The mistakes that make this a waste of time
Reading too many ads
Twenty well-read ads beat four hundred skimmed ones by a wide margin. The instinct to be comprehensive is the enemy here; you are looking for repeated patterns, and patterns show up in a good sample long before they show up in a complete one.
There’s a related trap: reading every ad from one competitor rather than a few ads from several. One brand’s library tells you what one brand believes. Five brands’ libraries tell you what the category has learned, and only the second thing is worth planning around.
Copying the biggest brand in the category
The market leader’s ads are optimized for a brand-aware audience you don’t have. Their “just the product, no explanation” hero shot works because everyone already knows what it is. Run the same ad as a challenger and you have bought an expensive lesson in the difference between recognition and interest.
Read the leader for category conventions — what the category takes for granted, what it never bothers to explain. Read brands your own size, or one rung above, for what actually converts cold traffic. Those two reading lists have almost nothing in common, and confusing them is the most common way this exercise goes wrong.
Doing it once. A quarterly deep dive surfaces trends about six weeks after they were useful. Because the Meta Ad Library only shows active ads, a competitor’s most interesting test — the one they ran for three weeks and killed — is invisible to you unless you were looking that month.
Little and often beats thorough and rare. Forty-five minutes weekly is the right shape.
Doing this automatically
The method above works. Its only flaw is that it costs you most of a morning every week, forever, and it is always the thing that gets dropped when something urgent happens.
What an agent does differently
Wisry’s AdClone competitor research runs the same method with an agent instead of your Monday. It collects the creatives your competitors are running on Meta and TikTok, reads each one for hook, awareness stage, objection, proof, offer and format, groups the recurring arguments, and returns six campaign angles with the source ads attached to every one of them.
The sourcing is the part that matters. When an angle doesn’t work, you can open the three competitor ads that justified it and decide whether the reading was wrong, the pattern doesn’t transfer to your price point, or your execution missed. Three diagnoses, three different next moves — versus one unhelpful conclusion that “AI ads don’t work for us”.
The other thing an agent does better than a person is coverage. Reading five competitors properly is a morning; reading twenty is not something anyone sustains. The long tail of smaller brands in your category is frequently where the newest ideas are, precisely because they can’t afford to keep running last year’s winner.
From angle to live campaign
The other difference is what happens after the research. Angles feed straight into creative: static ad generation for testing messages cheaply, or scene-by-scene video ads once you know which message is landing. Both are produced against your own brand and product facts, so a competitor’s claim never quietly becomes yours, and both publish into Meta ad sets grouped by the angle they came from — which is what makes the results interpretable a month later.
A 45-minute weekly routine
If you take nothing else from this, take the routine. Fifteen minutes in the Meta Ad Library on your five closest competitors, sorted by duration, noting anything new that’s over sixty days old. Fifteen minutes reading five of those ads properly against the six-point checklist. Fifteen minutes updating your angle list — adding what’s newly recurring, retiring what nobody runs anymore.
The routine survives busy weeks because it is short and because it produces a running artifact — the angle list — rather than a pile of screenshots that nobody opens again.
Do that for six weeks and you will know your category’s creative landscape better than the people running most of the ads in it. Do it for six months and you’ll be able to predict what your competitors are about to launch, which is a genuinely strange and useful place to be.
And if forty-five minutes a week is forty-five minutes you don’t have — which, for most founders running a store, it is — that is the honest argument for handing the reading to an agent and keeping the judgment for yourself.
