Why Long-Running Ads Are Your Best Creative Research
Competitors will not show you their ROAS, but they show you which ads they keep paying for. Here is why ad longevity is the most reliable free signal in creative research.
Creative research has a data problem. You can see what your competitors run, but you cannot see whether any of it works. Their dashboards are private. Their ROAS is private. Their failed tests are, mostly, invisible.
There is one exception hiding in plain sight: time. The Meta Ad Library shows when every active ad started running. And a brand that keeps paying for the same ad week after week is telling you something its dashboard would tell you, if you could see it.
The logic: survival is a decision made with money
Think about how a competent Meta advertiser operates. They launch new creatives constantly, often several a week. Each one gets budget. Within a few days, the losers are obvious and get switched off. The middling ones get a little more time. The winners keep spending.
That process repeats every week. So an ad that has been active for 60 days has survived something like eight rounds of a team deciding whether to keep paying for it. Each round was a chance to kill it. Every time, they chose not to.
No single decision proves the ad is profitable. Eight in a row is strong evidence. That is why long-running ads are a better research input than the newest, flashiest ads, the most-liked ads, or the ads a creative director thinks are clever.
Why other signals are weaker
| Signal | Why it misleads |
|---|---|
| Newest ads | They are guesses. Most will be off within two weeks. |
| Most likes or comments | Engagement often rewards entertainment, controversy or giveaways, not sales. |
| Best production quality | Polished ads often lose to rough UGC. Cost and performance are unrelated. |
| What big brands run | Large brands run awareness and brand campaigns that are not judged on direct sales. |
| What an agency recommends | Useful, but based on their portfolio, not your niche's live data. |
| Run time (active, 60+ days) | Reflects repeated spending decisions by someone watching results. |
Longevity is not a perfect proxy, but it is the only free one that is directly tied to money.
Where the signal breaks down
Be honest about the edge cases, so you do not overread a single ad:
- Forgotten ads. Some small advertisers set up a low-budget ad and forget it. It runs for months because nobody is looking, not because it works. Watch for brands with only one or two ads, all old; that is often neglect.
- Brand and awareness campaigns. Large brands may run evergreen awareness ads judged on reach, not purchases.
- Retargeting ads. These can run for a long time at low spend because retargeting audiences are small and warm. They are proven for warm traffic, not necessarily for cold.
- Always-on catalog ads. Dynamic product ads pulling from a catalog can stay active indefinitely. They tell you the brand uses catalogs, not that a particular message works.
- Seasonal businesses. A swimwear ad that ran from March to August might be a winner that was paused for winter.
The fix is simple: do not rely on one ad. Look for the same message surviving across several ads, several formats and ideally several brands.
Strengthening the signal
These cues make a long-running ad much more convincing:
- The advertiser is an active tester. A brand with 40 active ads, most of them new, and five that are four months old has clearly been culling. The five survivors are real.
- Multiple versions. Meta groups variations, and the library shows when an ad has several versions. Brands make variations of things that work.
- Iterations on the same concept. The same hook appearing with different creators, colors or offers suggests the brand is trying to scale a winning idea.
- Cross-brand convergence. Three unrelated brands all leading with the same promise for months is market-level evidence.
- A dedicated landing page. Effort spent on a matching page indicates the ad drives meaningful traffic.
What to extract from a long-running ad
The goal is not to copy the ad. It is to extract the reason it works, which almost always sits in the message rather than the visual. For each long runner, break it into three parts:
- Promise. The outcome the ad commits to. "Dinner on the table in 15 minutes." "Skin that doesn't feel tight after washing."
- Proof. What makes the promise credible. A demo, a review, a number the brand can substantiate, a guarantee, a founder story.
- Offer. What happens when you click. A bundle, a quiz, free returns, a trial size.
You can read more about this structure in the promise, proof, offer framework. The winning ad analyzer does the breakdown for you if you paste in the ad text.
A worked example
Say you sell a $45 insulated lunch bag aimed at people who bring lunch to work. You search the Meta Ad Library for "meal prep," "lunch for work" and "packed lunch," filter to active ads in the US, and sort through what has been running since spring.
Hypothetically, you find:
- A lunchbox brand's UGC video, running 110 days, opening with "My salad was still cold at 3pm."
- A meal-prep container brand's static ad, running 75 days, headline about food staying fresh all day, offering a bundle of sizes.
- A thermos brand's carousel, running 90 days, showing temperature readings at intervals through the day.
The common promise: food stays at the right temperature until you eat it, even late. The common proof: a time-stamped demonstration. The common offer: bundles.
That is a validated message, and it is yours to express in your own way. You would not copy the 3pm line. You might brief a creator to film their own lunch at the moment they actually eat it, with a visible clock, and test a two-bag bundle.
Using long-running messages outside Meta
A promise that keeps selling on Facebook is a promise buyers care about. That makes long-running Meta ads a research source for every channel you run.
ChatGPT ads are a particularly good fit. People ask ChatGPT things like "what lunch bag keeps food cold all day for work?" The ad appears below the answer and should answer the question itself. The validated message from Meta translates almost directly:
- Headline: "Keeps Lunch Cold Until You Eat It" (33 characters)
- Body: "Insulated bag that holds temperature through a full workday. Fits standard meal-prep boxes." (91 characters)
Notice that the copy describes the product, not the test results, and avoids unprovable superlatives. That keeps it within OpenAI's ad policies. For the full translation process, see how to turn Meta winners into ChatGPT ads.
How often to revisit
Long-running ads change slowly, which is part of their value. A quarterly deep pass is enough to refresh your view of the market, with a lighter monthly check on your top competitors. Look for two things each time: new ads that have just crossed 30 days (emerging winners) and old winners that have disappeared (possible fatigue or a shift in the market).
How SecondWin uses ad longevity
SecondWin is built on this signal. When you enter your URL, it identifies your product, buyer and niche, then mines the Meta Ad Library for ads in that niche that have survived 60+ days. It extracts the promise, proof and offer from those long runners, maps them to the questions buyers ask ChatGPT, and writes original ChatGPT ads around them. Ads are policy-checked before they go live and run in your own OpenAI ad account.
Get a free analysis of the proven messages in your niche, or see plans and pricing for managed campaigns.
FAQ
How long does a Facebook ad need to run to be considered a winner?
There is no official threshold, but many media buyers treat ads still active after 30 days as promising and after 60 days as strong evidence of profitability. By 60 days, an ad has survived many weekly decisions to keep funding it. Look for supporting signals too, such as multiple versions, iterations on the same concept, and the same message appearing across several brands.
Can a long-running ad still be unprofitable?
Yes. Small advertisers sometimes forget low-budget ads, large brands run awareness campaigns not judged on sales, and retargeting or catalog ads can stay on indefinitely at low spend. That is why you should look for patterns rather than single ads. A message that survives across multiple ads, formats and brands is far more reliable than one old ad from a page with little activity.
Where do I see how long an ad has been running?
In the Meta Ad Library at facebook.com/ads/library. Each active ad shows a "Started running on" date. Filter to active ads, choose your country, and search by keyword or advertiser. Subtract the start date from today to get the ad's age. You can also use the date filter to focus on ads that began running in a particular window.
Should I copy a competitor's long-running ad?
No. Copying wording, images, video or brand elements can infringe copyright and trademarks and makes your brand look derivative. Extract the underlying message instead: the promise, the type of proof and the offer structure. Then build your own ad with your own product, your own creators and claims you can substantiate. The message is the transferable part.