Blog - Growth, Engineering & AI

Here's a fact most app teams underuse: every single ad your competitors are running right now is publicly viewable, for free, on every major ad platform. After the 2018 transparency push, Meta, TikTok, and Google were all forced to publish their ad inventories—and the result is the greatest competitive research gift performance marketers have ever received. Yet in the accounts we audit, fewer than one team in ten has an actual process around these libraries. Most check once, screenshot three ads, and never return.

That's a mistake, because competitor ads are not inspiration—they're market data. Every long-running ad you find represents thousands of dollars of someone else's testing budget converging on a message that works. This guide is the full process we use: where to look on each platform, what data each library actually exposes, how to tell winning ads from noise, and the decomposition framework that turns what you find into your own winning creatives instead of pale copies.

Meta Ad Library: the workhorse

Start at the Meta Ad Library. Set the country, select "All ads", and search your competitor's brand name or Facebook page. You'll see every active ad they're running across Facebook, Instagram, Messenger, and the Audience Network—video, image, carousel, everything.

The single most valuable data point here is the launch date, shown on every ad. Ad platforms are ruthless auctions: nobody keeps a losing ad alive for months. So the heuristic is simple:

An ad running for 7–14 days is a test. Note it, but don't conclude anything.
An ad running for 30+ days is probably working—it has survived at least two optimisation reviews.
An ad running for 60–90+ days is almost certainly a profitable winner. These are the ads to study frame by frame.

Beyond longevity, look at variant clusters. When you see six versions of the same concept with different opening shots, you're watching a hook test in real time; when five of the six disappear two weeks later, the survivor just told you which hook won—research your competitor paid for. Also check the country filter: an app running distinct creative per market has learned that angles don't translate, and their market-specific messaging is a free localisation brief.

What Meta's library doesn't show you: spend, impressions (except for EU-targeted ads, where reach data is public thanks to the DSA—use an EU country filter to unlock it), or targeting. You're inferring performance from survival, which is exactly why the launch-date heuristic matters.

TikTok Creative Center: the only library with real metrics

TikTok goes much further than Meta. The TikTok Creative Center has a Top Ads dashboard that ranks the best-performing ads by region, industry, objective, and time window—and unlike Meta, it attaches actual engagement data: CTR percentile, watch-through curves, and the exact second viewers drop off.

The workflow we run: filter by your app's vertical and target region, sort by CTR for the last 30 days, and study the top 20. For each, open the analytics panel and look at the retention curve—the moment the curve cliffs is the moment the creative stops earning its place. Then watch second 0 to 2 of every top ad in a row. After twenty in a row you will physically feel the pattern: motion in frame one, a spoken or on-screen claim inside the first second, and the product visible by second three.

Two more Creative Center tools worth bookmarking: Keyword Insights shows which spoken and on-screen phrases are trending in ads (literally the words that convert right now), and Trend Discovery surfaces sounds and hashtags before they saturate. Both feed directly into creative briefs.

Google Ads Transparency Center: search and YouTube angles

At the Google Ads Transparency Center you can search any advertiser and see their ads across Search, YouTube, Display, and Shopping. It's less rich than Meta or TikTok—no dates on many formats, no metrics—but it answers two questions nothing else does: what YouTube pre-roll angles are competitors betting on (these are usually their most polished, highest-production message), and which search terms they're defending with paid placements. If a competitor suddenly starts running search ads on "your-brand alternative", you want to know that week.

The App Store blind spot, and how to work around it

Apple Search Ads has no public library. The workaround: search your own brand name and your top category keywords in the App Store across a few markets and note who appears in the ad slot. Do this monthly. If a competitor is bidding on your brand, you'll see it—and you can decide whether to defend. For broader network intelligence (which DSPs and ad networks a competitor spends on, and their top creatives across networks), paid tools like Sensor Tower, AppMagic, or Apptica fill the gap; AppMagic's creative gallery is the best value entry point at a few hundred dollars a month.

Decode, don't copy: the decomposition framework

Now the part everyone skips. Cloning a competitor's ad gets you a worse version of something the audience has already seen—you inherit the fatigue without the novelty. What actually works is decomposition. For every long-running ad you find, log six fields:

Hook type — question, shock stat, bold claim, demo-first, story open, or negative hook ("stop doing X").
Emotional driver — fear of missing out, relief, aspiration, curiosity, belonging, vanity.
Format — UGC selfie, screencast, motion graphics, street interview, split screen, AI avatar.
Proof mechanism — before/after, testimonial, on-screen numbers, live demo, authority figure.
CTA structure — direct ("download now"), soft ("see how it works"), urgency, or social ("join 2M users").
Length and pacing — total seconds, cuts per ten seconds, and when the product first appears.

After you've logged twenty ads in your category, patterns become unmistakable: maybe 70% of winners open with a question hook, UGC beats studio 3-to-1, every survivor shows the product by second four. That pattern map—not any individual ad—is the asset. You then rebuild the winning patterns around your product's strongest moments, which produces ads that feel native to the market's meta without repeating anything the audience has seen.

Make it a system: the weekly 30-minute ritual

Intelligence decays. An ad library snapshot from March is worthless in July, so this only compounds if it's a routine. Ours takes thirty minutes every Monday:

Check five competitors across Meta Ad Library, TikTok Creative Center, and Google Transparency (10 min).
Log new ads and mark dead ones in the tracking sheet (10 min).
Flag any ad that just crossed 30 or 60 days of longevity, and decompose it into the pattern library (10 min).

Within eight weeks you'll know your market's creative meta better than most agencies serving it. Every brief you write afterwards starts from evidence instead of a blank page, and your hit rate on new creatives goes up because you're recombining proven patterns rather than guessing. Pair this ritual with rank and download monitoring—our guide to tracking competitor app rankings and revenue covers that half of the intelligence loop.

One honest caveat: everything here is legal and built from public data—that's the entire point of the transparency rules. The line you don't cross is lifting someone's actual footage, claims, or trademarked assets. Patterns aren't property; pixels are.