Last updated: October 2026 · By The Ecommerce Benchmark Team
Finding winning ads in your niche is a scheduled sweep of a fixed watchlist, not an afternoon of browsing when the creative pipeline runs dry. The difference matters because the useful signal, how long an ad has been running, only exists if you looked before. Omniconvert has measured creative and conversion performance across 70,000+ experiments and 2,500+ Shopify stores over 13 years in eCommerce, and the teams with a genuine creative advantage are rarely the ones with better tools. They are the ones who looked at the same twelve advertisers every week for a year and wrote down what they saw. This guide sets out the Four-Pass Sweep, the three signals that identify a winner from outside, who belongs on the watchlist, and what to record so the log is still useful in six months.
A winning ad is defined as a competitor creative that has run long enough, broadly enough and through enough iterations that it is almost certainly profitable for the advertiser. You find them with a repeatable weekly sweep rather than a browse: build a watchlist of ten to fifteen advertisers, check each one in the public ad transparency libraries, read longevity as the primary signal, capture the variants, and log the first-seen date so the next sweep can compute how long each creative has survived. The process is the asset, because longevity cannot be measured without a prior observation.
- Longevity is the strongest signal available from outside. Nobody funds a losing creative for months.
- You cannot measure longevity retrospectively, which is why the schedule matters more than the tooling.
- Keep the watchlist to ten to fifteen advertisers, or the sweep stops happening.
- Include adjacent brands and overseas category leaders, not just direct competitors. That is where new angles come from.
- Always save a copy. Ads are pulled without notice and a dead link teaches nothing.
Why finding winning ads is a process problem
This is the point most competitor research advice skips. It recommends the tools, which are genuinely good and mostly free, and then leaves the reader staring at a wall of a competitor's current creatives with no basis for ranking them.
Ad transparency libraries do publish a start date for each creative, which gets you part of the way. But the richer signal is the change between sweeps: which ads disappeared, which survived, which spawned variants, which suddenly appeared across more placements. All of that is a difference measurement, and a difference needs two observations.
So the first sweep is mostly setup and returns little. The fourth is where the method starts paying, and the twentieth is a genuine competitive asset that a rival starting today cannot replicate for five months. That asymmetry is the whole reason to treat this as a standing process with a calendar slot and an owner.
The Four-Pass Sweep
The whole sweep should take under an hour for a watchlist of twelve advertisers, once it is established. If it takes longer, the watchlist is too big or the capture step is too thorough.
- Refresh the watchlist. Two minutes. Add any advertiser you noticed during the week, remove any that has stopped advertising entirely. Resist growing it: a list of thirty gets swept twice and then abandoned.
- Read longevity. For each advertiser, compare the current creative set against your last log. Note which ads survived, which are gone, and the first-seen date for anything new. This pass is the one that produces the ranking.
- Capture the survivors. Only for creatives that have cleared your longevity threshold or appeared as a new variant of a survivor. Save the asset itself, not a link.
- Log the pattern. One sentence per captured ad on what you think is making it work: the hook, the claim, the offer, the objection it answers. This is the field that makes the log searchable later.
The discipline is in pass three. The instinct is to save everything, which produces a folder of several hundred ads nobody will ever revisit. Saving only what cleared the longevity bar keeps the archive small enough to actually use.
The three signals that identify a winner from outside
None of these gives you the advertiser's actual numbers, and no public source does. What they give you is an inference about the advertiser's own behaviour, which is a more reliable guide than judging the creative on whether you personally like it.
The data is public and free. Meta publishes its Ad Library covering active creatives and their start dates, Google runs an equivalent Ads Transparency Center, and TikTok maintains its own commercial content library. Between them you can see most of what a consumer brand is running without paying for a tool.
Two cautions on reading the signals. A long-running ad may be a brand campaign with goals other than direct response, which is worth noticing before you copy its structure into a performance test. And a brand with a large budget can sustain a mediocre creative longer than a small one can, so compare longevity within a peer group of similar size rather than across the whole market.
Who belongs on the watchlist
Direct competitors are the obvious group and the least productive one, because you are mostly watching people who are watching you. The angles converge, and after six months the sweep tells you that everyone in your category is saying the same three things.
Adjacent brands break that. A company selling a different product to the same customer has had to solve the same objection with a different proposition, and the hook they landed on often transfers. A skincare brand and a supplement brand selling to the same buyer are running genuinely different creative against the same psychology.
Overseas category leaders are the highest-yield group. Formats, offers and creative conventions diffuse between markets with a lag, and a brand watching a more mature market has a reliable preview. This is also the safest group to borrow structure from, since you are not competing with them for the same impressions.
On list size: the constraint is not information, it is the hour. A watchlist you sweep every week for a year beats a comprehensive list you sweep twice. Twelve is a good target.
What to record, and why the log beats the folder
The table sets out the fields worth keeping and what each one is for, which is the difference between a log that informs a creative brief and an archive nobody opens.
| Field | Why it is there | What it costs you to omit |
|---|---|---|
| Advertiser and first-seen date | Makes longevity computable on the next sweep | The primary signal becomes unavailable permanently |
| Saved copy of the asset | Ads are pulled without notice; links rot | Your best examples vanish exactly when they are proven |
| Hook in the first three seconds | The most transferable element between brands | You keep the ad and lose the reason it worked |
| Core claim and the objection it answers | Connects creative to the customer belief it targets | Briefs end up describing a format rather than an argument |
| Offer and call to action | Separates a creative win from a pricing win | You copy a hook that was actually carried by a discount |
| Format and placement | Shows where the concept was given budget | Breadth, the second signal, cannot be read |
The fifth row is the one that saves the most wasted work. A great many apparent creative winners are offer winners, and a team that copies the hook while keeping its own pricing will conclude that the concept does not work.
What a performance marketer should do this week
- Write the watchlist: six direct competitors, four adjacent brands, two overseas category leaders.
- Run pass one and two for each. Record first-seen dates even where you capture nothing else.
- Set the longevity threshold you will use. Four weeks is a reasonable starting point in most consumer categories.
- Create the log with the six fields above before you save a single screenshot.
- Book the same hour next week, and the week after. Two entries in a calendar is the whole implementation.
- After four sweeps, review what survived and brief your first test from the pattern, not from a single ad.
The sweep tells you what competitors are running; it does not tell you how your own store compares on the dimensions that convert that traffic once it lands. The free Ecommerce Benchmark leaderboard score rates a store across six dimensions, Creative and Ads, Reviews and UGC, AI Visibility, Agentic Commerce, Competitor Synthesis and CRO, benchmarked against real competitors in your category and country, and the paid audit report returns the full prioritised detail. Pair it with the CRO side of the problem, because a strong ad pointed at a weak landing step is an expensive way to find a conversion leak: the CRO audit checklist covers that half. Once you know where you rank, Nexus by Omniconvert is the AI for eCommerce growth engine that unifies commerce data, prioritises experiments by True Profit, and generates the campaigns and creative you approve before they go live.
Frequently asked questions
How do you find winning ads in your niche?
Build a fixed watchlist of competitor advertisers, then sweep it on a schedule rather than browsing when you feel stuck. Public ad transparency libraries let you see which creatives an advertiser is currently running and when each one started, and the single most reliable signal available from outside is longevity: an ad still running after many weeks is almost certainly paying for itself. A repeatable weekly pass over the same watchlist is what turns that signal into a usable pipeline of creative ideas.
What makes an ad a winner from the outside?
Longevity first, then breadth, then iteration. An ad that has been running for weeks is being funded, and nobody funds a losing creative for long. Breadth, the same creative appearing across several placements or countries, suggests it survived a scaling decision. Iteration, where you see several close variants of one concept, shows the advertiser is investing in refining that angle. No single signal is proof, but an ad showing all three is as close to a confirmed winner as outside observation gets.
How often should you run a competitor ad sweep?
Weekly for a watchlist of ten to fifteen advertisers, taking under an hour once the list is built. Weekly is frequent enough to catch a new creative while it is still being tested, which is when the information is most useful, and infrequent enough that the task survives a busy quarter. Monthly sweeps miss the test phase entirely and only ever show you concepts that have already scaled, by which point the advantage has gone.
Who should be on your competitor ad watchlist?
Three groups, and the third is the one most people omit. Direct competitors selling the same thing to the same buyer. Adjacent brands selling something different to the same buyer, which is where fresh angles usually come from. And category leaders from outside your market or country, who are often running the concepts your market will see in a year. Keep the list to ten to fifteen names, because a list longer than that stops being swept.
What should you record about each ad you find?
Record the advertiser, the first-seen date, the format, the hook in the first three seconds, the core claim, the offer and the call to action, plus a screenshot or saved copy because ads disappear without notice. The first-seen date is what lets you compute longevity on the next sweep, and the saved copy is what lets you study a creative that has since been pulled. Without those two fields the log becomes a folder of images nobody can interpret.
Once the sweep surfaces a survivor, the next job is to Reverse-Engineer why an ad is winning rather than copying what it looks like.
The bottom line
Finding winning ads is a measurement problem disguised as a research problem. The signal that matters most, how long a creative has survived, cannot be read from a single visit to an ad library, which is why the schedule beats the tooling every time. Keep the watchlist to about twelve advertisers and make sure a third of them sit outside your direct competitive set, because that is where unfamiliar angles come from. Run the four passes in order, capture only what clears the longevity bar, and record the first-seen date and a saved copy without exception. The first sweep returns almost nothing. By the fourth it is producing briefs, and by the twentieth it is an asset a competitor starting today cannot match for months.