How to Repurpose One Video Into Six Ad-Ready Assets

Start from a 25-to-35-second vertical master with a clean two-second hook, a demo beat that freezes well, and a CTA that still makes sense if the last line is cut. If the only interesting frame is a blurry pan, you do not have six assets — you have one video and some ugly stills. Film or generate with extra handles on the hook and the product shot so a still has somewhere to land.

Watch the master once and mark four timestamps: hook peak, proof peak, product-in-hand, and CTA face or packshot. Those four frames are the stills and the carousel. If a mark lands on a half-blink or a busy caption, nudge until the still could run as an ad on its own. Burned-in captions that cover the product will haunt every static you pull, so prefer a caption-free master or a duplicate track you can hide before you start grabbing frames. Store those four timestamps in the project notes so a later editor does not have to re-watch for harvest.

Do not harvest from a 16:9 talking-head that was never meant for feed. Cropping a wide master into vertical and into square for six assets usually means the subject is off-center in every still. If you already have a 9:16 master, that is the source. Keep the project file. Repurposing is an edit job, not a re-export of the same 30-second file at different bitrates. Screenshotting a compressed Ads Manager preview will give you stills that look like screenshots, because they are. If the face sits in the lower third, you will lose it to UI — reframe the master before harvesting stills.

Delete the second agitation line and the extra B-roll, not the demo. The 15-second cut is the same ad with less air: hook, one proof shot, CTA. If you 'speed up' the whole 30 by 200%, you get a chipmunk and no placement-ready asset. Aim for a version that can run in Reels, Shorts, and in-stream where 30 seconds is optional but a complete thought is not.

Start on the same first frame as the master so thumbnail tests stay comparable. Keep one B-roll cut on the action beat; lose the establishing wides. If the spoken script no longer fits, record a tighter read or trim captions to match the new timing — do not leave sentences hanging halfway through a claim. The 15-second file is often the volume workhorse, so treat it as a first-class ad, not a leftover you exported because a media buyer asked for 'something shorter.' Keep the same CTA verb as the 30-second so the landing page does not have to guess which promise won.

Export 9:16 first, then a 1:1 crop if the subject sits center enough for Feed. Do not stretch. If the crop beheads the product, skip square for this master and assign square to a static instead. Name the file with length in the title so media buyers do not upload the 30-second into a 15-second slot by habit. If legal copy was already tight at 30 seconds, it is tighter at 15 — cut a claim before you cut the disclaimer that makes the claim allowable. If the 15-second needs a new legal line, add it as a caption block rather than shrinking the whole card.

The six-second asset is a hook and a product reminder, not a compressed story. Use the strongest two seconds of hook, a beat of product, and a logo or offer end card if you have one. This is the YouTube bumper, the story interrupt, and the retargeting nudge for people who already saw the long cut. If it needs a setup sentence, it is not a bumper yet.

If the hook does not make sense without the next line, pick a different hook or add four words of on-screen text that complete the thought. Bumpers with a setup and no punchline teach nothing and annoy people who have already heard the 30-second. Keep motion in the frame — a hand, a pour, a cursor — so the six seconds do not look like a frozen display ad with a soundtrack. A still with music is a static; if that is all you have, export a static and skip the bumper slot. Test the bumper muted: if the product is unnamed by second four, change the end card before upload.

Do not autoplay the first six seconds of the master if those seconds are a slow walk-up. Many UGC videos waste the open on a breath. Start on the pattern interrupt. End on a readable product or URL treatment. If legal copy is required, it still has to fit; a six-second claim with unreadable disclaimer is a policy problem, not a clever crop. When the claim cannot be supported in six seconds, use the bumper as a reminder of a product they already know, not as a new hard sell. Keep a 6s version without extra claims if the disclaimer cannot be set in type large enough to read.

Export stills at the timestamps you marked: hook face, product proof, CTA packshot. Add a short headline that can be read at a glance, keep the product large, and leave safe space for placement UI. These statics fill Feed, retargeting, and Advantage+ extra slots so the account is not video-only when the platform wants a mix and your video frequency is already climbing.

Pick the frame where the product is sharp and the face is not mid-word. A talking-head still with a half-open mouth underperforms a still from the demo beat almost every time. Overlay one line, not a paragraph. If the video already had burned-in captions, hide them or recapture from a clean layer so the static is not a screenshot of a subtitle. Match brand colors lightly; a heavy template turns a UGC still into a banner that no longer matches the video sitting next to it in the same ad set. Export PNG or high-quality stills; a compressed pause-frame will look soft next to the video.

Make at least one 1:1 and one 9:16 static from the same frame family so you can pair them with the video in the same ad set. Do not ship a 16:9 letterbox as a 'static' unless the placement is actually horizontal. Name them by frame job (hook, proof, pack) so you can see in reporting which still carried the click when the video was not the winner. If two frames look the same, you do not have two statics — pick the sharper one and spend the time on a headline test instead. If the headline repeats the spoken hook word for word, shorten it — statics are read, not heard.

Turn three to five stills into a carousel that tells the same story in taps: problem, product, proof, offer. Then export a silent GIF or a looping 3-to-5-second preview and a thumbnail frame for placements that show a poster image before play. That is asset five and six from the same master, still without a new shoot, and only if the pictures actually change between cards.

Carousels fail when every card is the same face with a new slogan. Change the picture each card: before, hands, result, price or CTA. Keep type large. Put the offer on card one or two, not only on the last card people never swipe. If you only have two good frames, skip the carousel; a weak five-card set is worse than two strong statics, and it trains the algorithm on swipes that never become clicks. Write the card order down before you export so you do not shuffle the story in the upload UI. Cap the set at five cards even if you have extra stills; swipe fatigue is real on cold traffic.

For the loop, take a two-second action (cap twist, pour, cursor click) and let it repeat without a hard cut that flashes. Keep it silent. The thumbnail should be the first-frame candidate you would also test as a video thumbnail — high contrast, product readable, no tiny text. Klip Kanvas exports of the master in more than one ratio give you a cleaner source for these stills and loops than screenshotting a compressed upload from Ads Manager, which is how most 'repurposed' statics end up looking like phone screenshots.

Use one stem for all six files: product, hook ID, master date, then suffix `_30s`, `_15s`, `_6s`, `_static-proof`, `_carousel`, `_thumb`. Upload them as a set into the placements they fit. Do not throw all six into one ad as a kitchen sink and call it a test. The point is coverage and later isolation, not chaos in the creative column you cannot read a week later.

A clean first test is the 30-second versus the 15-second in the same placements with everything else locked. Statics and carousels belong in the mix placements or in a retargeting set, not as a surprise extra creative in a hook test. Bumpers go where six seconds is the unit. When something wins, you already have the sister assets to fill the other slots without waiting on production. If the 15-second wins, do not assume the bumper will; length is a variable you already isolated once — keep it isolated.

Store the marked timestamps in the project so a later editor can harvest a new static when frequency climbs, instead of commissioning a new video on day twelve. Retirement can happen per asset: kill the 30 if it fatigues and keep the stills. That is the operational reason to repurpose. One shoot, several lives, one naming system so you can still tell which life made the sale. A Drive folder named 'final_final_exports' is how six assets become one forgotten zip. When frequency climbs, harvest a new still from the same timestamps before you book another shoot.