Read competitor ads and build a testable message

A public ad library gives you creative to inspect: the promise, the visual, and the destination. Use those observations to develop a testable message for your own product. Keep performance conclusions separate; the library does not reveal whether a particular ad is profitable.

Look the ads up yourself. Start with the Meta Ad Library and the Google Ads Transparency Center. An active ad tells you the advertiser was willing to show that creative. It does not tell you the budget, the targeting, the click rate, or whether the ad paid for itself. A start date is not a profit statement. A large set of ads can be a test, a set of regions, or something nobody turned off. The library will not tell you which.

What to write down when you open one

Pick three or four products a buyer might consider next to yours, and one adjacent product that solves the same job a different way. Search the advertiser's exact name. If you are not sure the page is theirs, stop. The wrong company produces a confident, useless sheet.

For each ad you personally open, record only what is on the screen:

• The promise in the first line.
• The image type: product UI, plain type, a person, or a before-and-after.
• The click label, and any offer stated in the ad.
• The destination, and whether that page repeats the ad's promise.
• The date the library displays, if it displays one, left as a date.

Save the library URL and the day you looked. A note that says "seen on this date" stays honest. A note that says "their best ad" has added a conclusion the screen did not provide.

Read down the promise column. The sentence that appears most often is the crowded sentence in the set you saw. It is not the correct sentence for your product. A buyer, a limit, or a kind of proof that never appears is a gap in that set. A gap in a handful of ads is a hypothesis. Those advertisers may say the missing thing in ads you did not see, or they may have tried it and stopped for reasons you cannot see.

**Hypothetical: appointment reminders.** Suppose the three advertisers you open all lead with "reduce no-shows" and a calendar screenshot. None of the ads you saw mention reminders that come from the front desk's own number. If your product sends from that number, you have a candidate message. You do not have evidence that their ads are profitable, or that your angle is empty territory.

When an ad shows a destination, open it. Message match is visible without private data. The ad talks about missed appointments and the page hero says "the AI platform." The visitor has to re-learn the offer. After you notice that pattern elsewhere, check your own page the same way. Landing page best practices gives you language for what the page makes obvious.

What you may carry home

You may borrow a structure: a problem, then a crop of the product, then one next step. You may not borrow their sentences, screenshots, logos, or testimonials. Pasting their line into your ad throws away the difference the teardown was for, and it sticks their claim onto your product.

Ask an agent to sort notes you collected. Do not ask it to scrape a market and declare winners. Publishing and spending stay behind your explicit approval. Running launch ads through an agent is the companion for that boundary.

Prompt

I will paste notes I took from public ad library pages. Each note has the advertiser, the promise line, the visual type, the destination, the date shown, and the day I looked.

Group the promises. Name the sentence that repeats in these notes. Name one buyer, proof, or limit that is absent from these notes only.

Do not call any ad profitable. Do not infer targeting, spend, or conversions. Do not write copy that echoes their sentences. Do not invent ads I did not paste.

Then list three claims my product could test, using only this brief:
[paste brief]

The sheet is done when every ad you opened has a promise and a destination, each competitor you listed has either rows or a line that says none were found on that date, you have named the repeated promise and one gap you will test, and the sheet contains no language about their profit. If a library would not load, write that down. An empty honest sheet beats a borrowed example.

When the gap is in your words, check that your page says it. The free scan scores the public page and names the largest gap it can see from the outside. It will not confirm that a competitor's ads are failing.