Guide
Most swipe files are graveyards — a bookmarks folder of screenshots nobody opens again. A useful one is a working library: organized by what each ad does, so it feeds your next batch of creative instead of collecting dust.
The goal isn't to collect pretty ads. It's to build a memory of patterns that work — the specific hook that stopped you, the format that made a boring product interesting, the structure that kept you watching to the CTA. When you sit down to make new creative, you shouldn't be starting from a blank page; you should be pulling from a shelf of proven moves and adapting them. A swipe file that's just a pile of links can't do that. One that's tagged and searchable can.
Start with your own feed — when an ad genuinely stops your scroll, that's a live reaction worth capturing before you forget it. Beyond that, the Meta Ad Library and TikTok's public ad tools let you browse what brands are actively running. The single best filter is longevity: if an ad has been live for weeks or months, someone is spending money to keep it up, which means it's probably working. Save those over the clever one-off that ran for three days. And don't only save winners in your category — a hook mechanic from a totally different vertical often adapts better than a tired trope from your own.
The tag that matters is function. For each save, note the hook type (question, bold claim, pattern interrupt, problem callout), the format (UGC talking-head, demo, before-and-after, listicle), and the structure (how it opens, where it turns, how it asks). Now your file answers real questions: "show me every problem-callout hook," or "how do the ads I admire handle their CTA." That's a swipe file you actually reach for. It pairs naturally with a disciplined naming and versioning system once those references become your own test variants.
A reference is a starting point, not a template to trace. The hook that worked for someone else's product carried their context, their audience, their timing. When you adapt the pattern to your own creative, you've introduced a dozen new variables — which means "it worked for them" guarantees nothing about your version. This is the step most swipe-file advice skips. Before you spend, screen your adaptation: PreTestAds predicts attention and engagement from the finished cut and scores it against a benchmark of top-performing ads, so you can see whether your take on the pattern actually holds a viewer. Turn the reference into a real ad, screen it, and only launch the ones that hold up — the same discipline behind testing on a small budget and building a bank of variants from one shoot. A prediction screens the idea; live testing confirms it.
Most swipe files start in an ad spy tool or the Meta Ad Library. The catch is those libraries show you an ad exists, not whether it works — so before you model a competitor's creative, check whether it's actually working.
Adapt a pattern, upload your version, and see if it holds attention — up to 8 free credits, no credit card.
Score Your CreativeA swipe file is an organized collection of ads you've saved as inspiration — hooks, formats, structures, and openings that caught your attention. The point isn't to hoard screenshots; it's to build a reusable library of proven patterns you can adapt into your own creative and test before spending.
Your own feed as you scroll, the Meta Ad Library and TikTok's public ad tools, and any competitor or category leader whose ads keep running. If an ad has been live for weeks, someone is paying to keep it up — that's a signal it's worth studying.
Tag each save by what it does — hook type, format, structure — then adapt the pattern to your product rather than copying it. Produce a version, and screen it before launch: PreTestAds predicts whether your adaptation actually holds attention, so a good reference turns into a good ad instead of a guess.