Guide
Everyone's ad folder eventually fills with final_v2_FINAL_real.mp4. The problem isn't tidiness — it's that you lose the ability to answer the only question that matters: what actually changed between the ad that worked and the one that didn't.
Testing creative only compounds if you can trace results back to decisions. If your winner is named tiktok_edit_3.mp4, you know it won but not why — was it the hook, the format, the length, the angle? Next month you rebuild from scratch and accidentally re-test ideas you already killed. A naming convention turns every file into a small experiment record. When you repurpose one video into multiple cuts, the names are what let you compare them fairly instead of eyeballing a folder of near-identical thumbnails. The same discipline applies upstream, to the references you collect: a well-organized ad creative swipe file feeds named, testable variants instead of a pile of screenshots.
Encode the variables you actually change between tests, and nothing else. A practical set:
A finished name looks like CONCEPT-founderstory_HOOK-question_9x16_15s_v3. Ugly, but every underscore is a decision you can filter and compare on later.
Three rules do most of the work. First, never reuse a version number — v2 always means the same cut, forever, even after it's archived. Second, increment the version only for a real creative change, not for a re-export or a caption tweak; otherwise the number stops meaning anything. Third, one concept, one hook change at a time — if you swap both the hook and the length between v2 and v3, you can't attribute the difference to either. That discipline is the same reason creative testing isolates one variable per comparison.
A naming system earns its keep when each variant carries a result. Before you launch, run each cut through pre-testing and log the predicted attention score next to the name. PreTestAds scores your creative as a percentile against a benchmark of top-performing short-form ads and flags Hook Strength, attention drop, and the CTA window — so before a single dollar moves, your spreadsheet already ranks the variants. Add the live results after launch, and over a few months your naming log becomes a private swipe file: a record of which hooks and formats consistently earn attention for your product, not a generic best-practices list.
Don't build a taxonomy so elaborate that nobody follows it. Five fields and a shared doc beats a 20-field schema that gets abandoned by the second week. The test of a good system is simple: six months from now, can you look at a filename and know what you tested, what it scored, and whether to build on it? If yes, it's doing its job — and it pairs naturally with a repeatable creative testing framework.
Upload your cuts, get predicted attention for each, and start a log worth keeping — up to 8 free credits.
Score Your VariantsOnce you run more than a handful of variants, files like final_v2_REAL.mp4 make it impossible to know what changed between winners and losers. A consistent naming convention lets you trace which hook, format, or angle actually drove results — so you scale what worked instead of guessing.
A useful name encodes the variables you test: concept or angle, hook variant, format or aspect ratio, length, and a version number. For example CONCEPT-founderstory_HOOK-question_9x16_15s_v3. The point is that every field maps to a decision you can compare later.
A naming system is only useful if each variant carries a result. Attach the pre-test score and later the live metrics to the same name, and your file names become a searchable record of which hooks and formats earn attention — a swipe file of your own proven work.