Steal what works: building ads from competitors' longest-running ads

You can't see a competitor's results. You can see how long they've kept an ad running, and that's nearly as good.

Advertisers turn off ads that lose money. Ad platforms charge for every day an ad runs. So an ad still running after two or three months is, in most cases, paying for itself. Its hook, angle and format have been tested on real buyers with real money, and they passed.

This doesn't mean the ad is perfect, or that it will work for you unchanged. Big brands sometimes leave ads running out of neglect, and your audience, price and proof differ. But as a starting point it beats guessing, and it beats a brainstorm.

Two rules before you start:

• Study patterns, not assets. You're learning what angle and format work in your market. Copying someone's video, images, script or brand is unethical, often a copyright problem, and against ad platform rules.
• Use several competitors. One brand's long-running ad might be a fluke. The same pattern across three brands is a market signal.

Meta Ad Library (facebook.com/ads/library)

1. Choose your country and "All ads".
2. Search the competitor's brand name or Facebook page.
3. Filter to active ads.
4. Each ad shows "Started running on" with a date. Note the oldest active ones.
5. Check whether several versions of the same ad are running. Many variants of one idea is another sign it works.

Google Ads Transparency Center (adstransparency.google.com)

1. Search the advertiser's name or domain.
2. Filter by region and format (text, image, video).
3. Look for ads shown over a long period and the text ads they keep coming back to.

TikTok and LinkedIn

• TikTok's Creative Center shows top-performing ads by industry and region.
• LinkedIn shows a company's ads on its page under "Ads", and through its ad library.

Collect 10–20 ads across 3–5 competitors in a sheet: link, start date, format, the first line or first 3 seconds, the offer, and the call to action.

For every ad in your sheet, write down four things:

Then look for patterns across the sheet:

• Repeated angles. If three competitors all lead with "time back", that's the market's main pain.
• Repeated formats. If the long-running ads are all talking-head UGC ads, studio shots probably lose in your category.
• Gaps. An angle nobody uses, like price or a specific integration, might be your opening, especially if it's your real strength.

You should end with 2–4 patterns worth testing, each described in one line: "Talking-head UGC, hook on the spreadsheet pain, offer a free month."

Now build 3–5 variants per pattern with your own material. What changes and what stays:

Keep from the pattern

• the structure of the hook (a question, a surprising number, a before-and-after),
• the angle,
• the format and length,
• where the offer sits.

Make your own

• every word of the script and copy,
• your product, your screens, your customers' words (with permission),
• your proof: real numbers, reviews or results you can back up,
• your brand: colors, fonts, voice.

Vary one thing per variant so you learn something. For a talking-head video: same script, three different hooks. For a static: same image, three different headlines.

Write the claims carefully. Ad platforms reject ads that make unsupported claims, and "cut scheduling time by 80%" needs evidence. "Built for single-location restaurants" doesn't.

A test needs a fixed budget, a clear metric and patience.

1. Create the campaign paused. Check the budget, audience, placements and conversion event before anything spends.
2. Pick one metric. Cost per sign-up, cost per purchase or cost per lead. Not clicks, not likes.
3. Set a budget and a stop rule. For example: $30 a day per ad set for 7 days, then stop anything above your target cost per result.
4. Let it spend enough. Judging after $20 is noise. Many teams wait until each variant has spent at least one to two times their target cost per result.
5. Keep the winner, kill the rest, and write down why it won.

On Meta, automated options like Advantage+ may decide which variant gets shown. That's fine, as long as you compare results on your chosen metric and check the totals against your real revenue (MER).

Every winner fades. As the same people see an ad again and again, click-through drops and cost per result rises. That's ad creative fatigue.

Keep a loop running:

• Re-check competitors' long-running ads every month. New ones show you where the market is moving.
• Make the next variants of your winner before it fades: new hook, same angle.
• Watch frequency and cost per result weekly, and swap creative when they turn.

The teams that win at paid social usually aren't the ones with the best single ad. They're the ones with a steady pipeline of new tests.

This loop is built into Cosmo. During onboarding he finds your competitors and pulls their longest-running ads through Meta's Ad Library and Google's Ads Transparency Center. He breaks them into patterns, makes your versions (statics and UGC videos with AI actors; 100 creatives and 10 videos a month included), and builds the campaign on your own Meta, Google, TikTok or LinkedIn account, created paused.

The card looks like this: "7shifts has run one UGC ad for 94 days. I made our version (2 videos). It's in Meta, paused, $40/day. Want it live?" (from the Shiftly example). You reply 👍, and he tells you what's winning every morning. See Cosmo for ecommerce brands.