The best ad material in your archive is the footage nobody ever used: raw rushes, alternate takes, and the b-roll that got cut for time. The worst is your finished brand film, because it is graded, married to music, locked to a rhythm, and every good shot in it has already been used at its best angle. Rank your archive by how much unexposed, unlocked coverage it contains and you will get a list that looks nothing like the order the files were important in.
Every source, ranked by how well it converts into ads
Yields below are what we typically see, offered as experience rather than as a study. A usable clip means roughly 1.5 to 4 seconds of clean action with no burnt-in text, no hard grade, and no music welded to it.
| Source | Ad yield | Typical usable clips | Why it ranks there |
|---|---|---|---|
| Unused takes and full rushes from past shoots | Highest | 40 to 80 per 30 minutes of rushes | Unseen by your audience, shot with advertising intent, and the coverage is already there. Nobody has to like it for it to work |
| Studio product rolls and photography-shoot video | Very high | 20 to 40 per shoot | Clean, well lit, no audio to fight, and usually far more angles than the final selects used |
| Behind the scenes and production footage | High | 15 to 30 per shoot day | Reads as real in feed, which is exactly the register short-form rewards, and it has never been published |
| Founder and team interviews, podcast video | High for script, medium for visuals | 10 to 20 visual clips, plus dozens of usable lines | The best source of language you did not invent. Real phrasing beats written copy for hooks |
| Customer testimonials and UGC | High, limited supply | 5 to 15 per submission | Strongest proof you can put on screen, but rights, audio quality and quantity all constrain it |
| Long-form tutorials, demos and unboxings | Medium to high | 10 to 20 per 10 minutes | Demonstration shots are hard to fake and hard to shoot. The pacing needs work, the content does not |
| Event and trade show footage | Medium | 10 to 25 per event | Great for texture and scale, weak for product detail, and watch for third parties in frame |
| Webinars and conference talks | Low visually, high verbally | 3 to 8 visual clips, 20 or more script lines | Mine the transcript, not the video. A slide deck on screen is not ad footage |
| Animated explainers and motion graphics | Low | 2 to 6 per video | Locked compositions, burnt-in text, and a style that fights any other footage you cut it against |
| Finished brand films | Low | 6 to 12 per film | Graded, scored, and cut to a rhythm that does not survive being broken up. Covered below |
| Ads that already ran hard | Fast but exposed | 6 to 10 per ad | Fastest possible start because the footage passed brand and legal review, but the audience has already seen it. See the caveat below |
Two things drop out of that table that most teams find counterintuitive. Your most expensive video asset is usually your least useful, and the folder your editor called "unused" is the one worth opening first. Our writeup on turning existing footage into ads covers the workflow once you have picked a source.
Why finished brand films are poor ad material
Five specific reasons, all of them structural rather than aesthetic.
It is graded. A heavy colour treatment cannot be undone, so every clip you pull carries a look that will not match anything else in your library. Cut two graded films together and it reads as a mistake.
It is married to music. If the audio bed runs under the whole film, any clip you extract either brings a fragment of music with it or has to be silent. Both options are worse than a clip with clean or usable ambient sound.
It is cut to a rhythm. A 60 second film builds. Shots are chosen for their position in a sequence, which means many of them make no sense alone: a slow push that pays off three shots later is not a hook.
It has burnt-in text and lower thirds. Titles, names, end cards, legal lines. Anything with type on it is unusable in a new context, and the type is often in the most useful part of the frame.
The good shots are already spent. A brand film uses the best angle of the best moment, and the archive it came from is where the alternates live. Go to the rushes instead. You are not looking for what the editor chose, you are looking for what the editor rejected.
Why rejected takes are the valuable ones
Takes get rejected for reasons that have almost nothing to do with whether they perform in feed.
The line was slightly off. Someone glanced at the camera. There was a stumble, a laugh, a pause. The director wanted a different energy. In a 60 second brand film those are defects. In a nine second vertical ad they are the reason someone watches, because short-form rewards footage that does not look produced, and an imperfect take reads as a real person doing a real thing.
There is a harder economic argument too. Rejected takes have zero accumulated exposure, and exposure is the thing that kills creative. Our 2026 fatigue benchmark records CTR dropping 45 percent after a fourth exposure to the same creative, and puts week three at 45 to 70 percent below the launch baseline. Footage nobody has seen starts that clock from the beginning. Footage from an ad that spent heavily for six weeks does not.
Volume favours you here as well. A shoot that produced one 60 second film typically shot eight to twelve times that much material. That ratio is where your next quarter of creative already lives, unedited, on a drive somebody has to go and find.
How to audit your archive in an afternoon
Work through these locations in this order, because it is roughly the order of yield per minute spent.
- The editor's project archive, not the delivered files. Rushes, selects, and the bins nobody cleaned out. If an agency shot it, the contract may or may not give you the rushes, and it is worth asking.
- The photographer's rolls. Photo shoots almost always capture video, and nobody remembers.
- Your own phone and your team's phones. Production days, warehouse, packing, install, repair. Unglamorous and highly usable.
- Published long-form. YouTube, webinars, podcast recordings. The transcripts alone are a script mine.
- Testimonials and support recordings, subject to permissions, which need checking before anything else happens.
- Old ad accounts, for finished ads and their source files.
Then check three things per candidate clip, in this order: is there burnt-in text, is the grade heavy, and is the audio married to music. A clip that fails all three is not worth logging. A clip that passes all three goes into the library even if it looks boring on its own, because boring coverage is what makes a variation possible.
Tag as you go, or have something tag for you. An untagged library is a folder, and a folder does not produce ads. What makes any of this work is having the transcript and the tags in place before you try to build anything, which is the whole function of a clip library.
Getting short ads out of long content
Long-form is the source most teams have most of, and the mistake is treating it as a trimming exercise. Cutting a 12 minute demo down to 15 seconds gives you a bad 15 seconds, because the good moments are scattered and the good sentence is usually four minutes away from the good shot.
The approach that works is to separate audio and visuals. Read the transcript first and pull the three or four sentences where somebody says something specific and unrehearsed, then find visuals for them anywhere in the library. Your hook can come from minute nine and your product shot from minute two, and the ad will be better than any contiguous slice. More on that split in our notes on turning long video into short ads.
Disclosure: Genyad is our product and it is built for exactly this input. You upload the library once, it transcribes and tags every clip, and each variation is a fresh script, shot selection, voiceover, caption set and export drawn from that library rather than a recut of one timeline, in 9:16, 4:5, 1:1 and 16:9 at 1080p with no watermark. Uploading, editing and re-exporting cost no credits. It cannot fix what is not there: no AI avatars or synthetic presenters, no static banners, no product-URL import, no product-feed rendering, no predicted performance scores and no direct publishing to Meta or TikTok. A library of twelve graded clips will produce twelve graded clips' worth of variety, no matter what tool reads it.
Frequently asked questions
What is the best existing video content to turn into ads?
Unused takes and raw rushes from previous shoots, because they were filmed with advertising intent, nobody has seen them, and the coverage already exists. Studio product rolls and behind the scenes footage come next. Finished, graded brand films are the least useful despite usually being the most expensive thing in the archive.
How much usable footage do I need before starting?
Enough for genuine variety rather than a specific minute count, which in practice means roughly 30 to 60 distinct usable clips covering product, people, context and detail. Below that, variations start reusing the same shots and your batch collapses into one idea in several costumes. Adding coverage is usually cheaper than any tool.
Can I use footage from a video an agency produced for us?
The finished film is normally licensed to you, and the rushes often are not, so check the contract before planning around them. Music, stock elements and on-screen talent frequently carry separate term-limited licences that a new ad would fall outside. Worth resolving before you build a library on top of it.
Why does my finished brand film cut badly into short ads?
Because it is graded, scored, and edited to a build where individual shots only make sense in sequence, with titles and lower thirds burnt into the frame. Extracting a clip brings the grade and often a fragment of music with it. Go to the rushes that film came from instead, where the alternates are clean.