Meta Ad Library: How to Use It to Find Angles
The Ad Library shows every ad running on Meta. How to read it for angles and hooks instead of scrolling it, and what it will never tell you.
The Ad Library is the most used and least methodical research tool in media buying. Most people open it, scroll a competitor for ten minutes, feel vaguely informed, and close it with nothing written down.
The tool is good. The way it is usually used is not.
What it is
A public archive of every ad currently running on Facebook and Instagram. Search by advertiser page or by keyword, filter by country, media type and date, and see the creative and the copy exactly as they serve.
Meta publishes it as a transparency measure. That origin explains its shape: it is designed to show you what is running, not to help you compete.
The one signal it gives you
Run time.
An ad that has been live for four months is an ad somebody is happy to keep paying for. An ad live for six days is a test that may already be dead.
That is blunt, and it is essentially all you get. Spend and impression ranges are published only for ads about social issues, elections and politics. For a commercial advertiser you see the creative, the copy and the dates.
Run time also has real exceptions worth naming. Large advertisers keep ads running for reasons unrelated to performance: brand campaigns with fixed flights, contractual commitments with creators, and plain neglect in accounts nobody prunes. Treat a long run as evidence, not proof.
Read by angle, not by competitor
This is the change that turns browsing into research.
Browsing competitor by competitor produces a list of ads and a vague sense that everyone is doing UGC. Browsing by angle produces something you can brief against.
Pick the angle first. Before and after. Us against them. Founder story. Objection handling. Price justification. Social proof. Then search that angle across several advertisers in your category and compare how each one opens.
| What to record | Why it matters |
|---|---|
| The first three seconds | Where the scroll decision happens, and the only part most viewers see |
| The claim being made | The angle stripped of production value |
| The objection being addressed | The reason someone was not buying |
| The format and ratio | Whether it was built for feed or for story |
| How long it has run | The only performance proxy available |
Five columns, filled while you browse. Twenty rows is a brief. Twenty minutes of scrolling without the table is entertainment.
What to look for in the first three seconds
Most of the decision happens before anyone hears your offer. When you catalogue openings, the recurring shapes are worth naming:
- A problem stated as a question the viewer has actually asked themselves
- A visual contradiction that does not resolve immediately
- A number specific enough to sound measured rather than marketed
- An objection said out loud before the viewer thinks it
- A face and a direct address, which still outperforms most production
You are not collecting these to copy them. You are collecting them to see which ones your category has exhausted and which ones nobody in it has tried.
Cross it with your own data
The library tells you what competitors keep running. It cannot tell you what your own audience objects to.
Your ad comments can. They contain the exact words your customers use and the exact reasons they hesitate, which is more specific than anything a competitor’s creative will reveal.
An angle that appears in both places, a competitor running it for months and your own commenters raising it unprompted, is about as strong a brief as this work produces. That intersection is worth more than either source alone.
A workable routine
Once a month, not once a quarter, and not daily.
- Pick three competitors and two adjacent brands outside your category
- Filter to ads running more than sixty days
- Fill the five column table, twenty rows maximum
- Mark which angles your account has never tested
- Brief two of them
The limit of twenty rows is deliberate. The failure mode of this exercise is collecting more than you will ever act on.
What this does not solve
The library shows what was made. It never shows why it worked. An angle lifted without its offer, its price point and its audience reproduces the surface and none of the mechanism, which is how brands end up running a competitor’s format at half their conversion rate.
It also shows only what is running now. An advertiser’s best performing ad from last quarter, the one that actually built the account, may have been retired and left no trace.
And it is a research tool, not a testing plan. Twenty recorded angles do not become results until two of them are produced, launched against a structure that can read them, and given enough budget to leave the learning phase.
Frequently asked questions
What is the Meta Ad Library?
A public archive of the ads currently running across Facebook and Instagram. You can search by advertiser page or by keyword, and filter by country, media type and date. Meta publishes it for transparency reasons, and it covers commercial advertising as well as political.
Can I see how much a competitor spends?
Not for commercial ads. Spend and impression ranges are only published for ads about social issues, elections and politics. For a skincare brand you see the creative, the copy and the dates, and nothing else.
What signal can I actually trust?
Run time. Nobody keeps a losing ad live for four months. It is a blunt signal and it has real exceptions, but it is the only one the library gives you and it separates tests from sustained winners.
How is this different from a paid spy tool?
Paid tools mostly add search, filtering and history on top of the same public data, plus estimates they model themselves. The underlying ads are the same ones the free library shows.
quotidiens →