Best UGC Script Structures for Ecommerce Video Ads

Ecommerce UGC wins when the hands beat the mouth. Demo-first opens on the mechanism — pour, stretch, click, wipe, blend — and only then explains it. The first frame is a verb, not a face saying 'so I found this'. In accounts we work with, this shape consistently holds viewers who would have classified a talking-head open as an ad. It also survives muted play, which is most of the impressions you are paying for.

Time it like this: 0–3s product in action, no logo; 3–8s the creator names the problem the demo just solved; 8–20s a second, closer demo plus one proof (number, time, comparison); last 5s a soft CTA. The second demo is not optional — one flash of product is a teaser, two is a demonstration. Keep the camera at hand height for the open, then cut to a face if you need scepticism. Do not reverse that order on a cold audience. If the first three seconds could be a still poster of the product working, you have the open; if they could only be a face, you do not, and you should steal a first frame from the B-roll instead of hoping the line will save you.

This is the default for physical products with a visible mechanism: kitchen, cleaning, beauty texture, gadgets, apparel fabric. It is the wrong default for supplements and SaaS, where there is nothing honest to show in three seconds. If you generate the ad, brief the B-roll first and the talk-track second, or the model will open on a smile. Pair with captions that describe the action ('watch the stain, not me') so sound-off still delivers the structure. On apparel and food, the demo is also a colour and texture check — if the on-screen unit does not match the PDP, you will pay for returns that look like a targeting problem. Lock the hero SKU in the brief before anyone generates a second take.

PAS is still the most reliable talk-track for products that solve a recurring irritation. The open names a specific weekly pain, the middle makes the current workaround feel stupid, the close is the product as the obvious replacement. It is not a story arc. It is a three-beat argument, and it should be over in half a minute. Hook rate is high when the pain is concrete and low when it is 'are you struggling with self-care'.

Specificity is the whole craft. 'My black tees were grey by wash five' is a problem; 'I was tired of low quality' is a slogan. Agitation should be one sentence, not a verse — extra twisting after the viewer already agrees feels like a lecture. The solve must be a demonstration, not an adjective: show the tee coming out dark, do not say 'premium cotton'. A PAS that never shows the product is just a rant with a URL. Read the 1-star reviews for the agitation line; they are ruder and more specific than anything a strategist will write in a doc.

Use PAS on replenishable products and on anything with a hated incumbent (the drugstore mascara, the noisy fan, the app they already pay for). Avoid it on identity and gifting products, where reminding people of pain is a poor mood for a purchase. If you run PAS in several markets, re-write the problem; do not translate the English irritation and hope it exists locally. A 15-second PAS is often enough: problem, one twist, demo. If you need 40 seconds, you are agitating for sport. Cut the second complaint and spend the time on the product in the hands.

The creator opens holding a phone, reads a customer sentence out loud — sceptical, specific, a bit rude — then looks up and says 'I wanted to see if that was true'. The rest of the ad is the proof. It borrows trust from someone who already bought, and it gives you a hook that is not a slogan. CTR is strong when the review is ugly-honest; it dies when the review sounds like it was written in the brand Slack.

Use a real review, lightly edited for length, not invented. Platform review and the comments will both clock a fake. Pick a line that contains a testable claim ('held through a 12-hour shift', 'didn't pill after three washes') so the demo has a job. Show the review on screen for a second, then get to the proof; lingering on a star rating is a wasted hook. Attribute it vaguely if you must ('someone said') but do not fabricate a name and a city. If the review names a time window or a wash count, that number becomes the only claim the rest of the ad is allowed to make — do not pile a second miracle on top of it.

This structure is a gift for AI UGC because the script is half-written on your product page. Pull three reviews, generate three ads, keep the one whose claim you can actually show. It is weaker on brand-new listings with five reviews, and it is dangerous in supplements or devices where the review implies a health outcome you cannot claim. When in doubt, pick a logistics or texture review, not a miracle. When reviews are thin, wait. A invented 'someone on Amazon said' is a faster way to lose the account than to pause UGC for a week. Use PAS or demo-first until the page has a sentence worth stealing.

The customer already has a reason not to buy. This structure steals that reason and answers it before the pitch. 'I thought this was going to be another cheap blender' is the hook; the blades on ice, the warranty card, the noise comparison are the body. It is the negativity-flip as a full script, not just an opening. Hold-through is high because the viewer wants to see whether you beat the objection they also have.

Pick one objection. Price, durability, fit, 'does it actually work', shipping — not all four. A 30-second ad that tries to clear a FAQ is a landing page. Name the objection in the customer's words, which you will find in 1-star reviews and in abandoned-cart surveys. Then show the disproof, do not argue it. A spoken 'but actually our quality is amazing' is the brand talking; a seam-stretch and a window test is the answer. Put the objection in captions as well as speech so muted viewers still get the structure; a doubt-flip that only lives in the VO is a talking-head the sound-off half never heard.

This is the structure to run when CPA is fine but conversion rate on the PDP is not — people click and then talk themselves out. Putting the objection in the ad pre-qualifies the click. It is the wrong structure when you do not yet know the objection; guess-based doubt sounds like a humblebrag. Mine support tickets for a week before you write. If two objections are tied, make two ads. Combining price and fit into one 25-second argument gives you a muddled CPA and no idea which doubt you actually cleared.

A true before/after is the most compressed proof in ecommerce, and it is also the most abused. The structure is visual: beat one is the problem state, beat two is the product in use, beat three is the after, beat four is the CTA. Words are captions, not a monologue. When the delta is real and the lighting matches, this shape stops thumbs that ignore talking heads. When the delta is graded in post, it trains distrust.

Match the frame. Same room, same time of day, same camera height, same crop. Anything else reads as a different photo, not a transformation. Time-stamps ('day 1 / day 14') help if they are true. Platform policies on before/after are strict in beauty, health and weight — if you cannot substantiate, do not run this structure, full stop. Household, apparel fit, and organisation categories are usually safer because the delta is mechanical, not medical. Shoot the before first, then use the product, then the after, in one session — coming back tomorrow is how the light changes and the proof starts to look graded even when it is not.

Hold the after long enough to be inspected. A five-frame flash looks like concealment. Then say one number or one constraint ('two coats, no primer') so the viewer knows what they would have to do. This structure recuts beautifully into a 6-second bumper: before, after, logo. Keep a 9:16 master with the talk-track for prospecting and the silent pair for retargeting. On retargeting, a 6-second after-only loop plus a captioned timestamp is often enough; save the full four-beat for cold. Do not run a medical-looking split on a cosmetic unless legal has signed the exact frames.

Unboxing is a Shop and TikTok native shape: the viewer likes the ritual, and you get to show pack quality, accessories and first use in one take. As a paid-ad structure it is mid. Hook rate depends entirely on the first frame (blade on tape, not a smiling face with a box) and the ad has to reach first use before second 12 or it is content, not a commercial. It is still the right shape when the unboxing is the proof — premium packing, a surprising accessory, a product that looks different in person.

Cut the gratitude. 'Oh my god it came' is not a hook. Open on the interesting physical detail: the seal, the weight, the extra nozzle, the colour in daylight versus the PDP. Get to use. A 30-second unboxing that never turns the product on is a shipping ad. Voiceover can be sparse; the sounds of the packing are part of the native feel. End on the first successful use plus a soft CTA, not on a table full of crumpled paper. If the accessory in the box is the reason people buy, hold it up before the thank-you energy starts — that clip is the ad, the rest is wrapping.

This is a weak prospecting structure for low-AOV replenishables — nobody needs to see your vitamin bottle come out of a mailer. It is a strong structure for devices, apparel, and anything with a 'looks cheaper online' problem. If you must generate it with AI, do not fake a branded box you do not ship; the comments will include photos of the real packing. Shop listings that already have a pack shot still benefit from a real unbox if the in-person colour surprises people. If it does not surprise them, skip this structure and demo the use instead of the cardboard.

Integration beats presentation for products that live in a routine: skincare, coffee, fitness, baby, kitchen. The script is a morning or a commute with the SKU as one beat among others. Done well, it is the least 'ad-like' structure on the list. Done badly, it is 25 seconds of lifestyle and 4 seconds of product, which is how you pay CPMs for a vlog. Pin a clock to the product beat or the editor will fall in love with the morning and forget what you paid for.

Pin the product to a timestamp. The SKU should be on screen by second five and doing its job by second twelve, even if the 'story' continues around it. Voiceover should sound like a diary, then snap to a concrete claim ('this is the only cleanser that doesn't leave a film before I commute'). If you cannot name a claim that sits inside the routine, you do not have an insert, you have a mood board. A simple rule: if you mute the ad, can you still tell which SKU is for sale by second eight? If not, you shot a vlog and added a URL.

Fatigue is slower on this shape because the wrapping can change (new room, new hour, new kid in frame) while the product beat stays. That makes it a good scaling companion to a demo-first winner. It is a poor testing structure: too many variables, too easy to misread a lifestyle preference as a product preference. Prove the product with ranks 1–4, then wrap the winner in a routine for extra life. When you refresh, change the wrapping (room, hour, extra person) and keep the product sentence identical so you can still read the test. Changing both is how this structure becomes un-attributable noise.

The catalogue script. A face, a pack shot, a list of attributes. It is how a lot of Amazon and Shop videos are still made, and it can work on high-intent placements and on retargeting where the viewer already wants the category. On cold UGC auctions it is the lowest hook-rate structure in the list, because nothing in the first three seconds is a story, a doubt or a demo — it is a spec sheet with lips.

If you must use it, put one surprising attribute in the open and move the rest to captions. 'This thing charges from a laptop' is a hook; 'key features include' is an ad. Show each feature as you say it. A list over a single pack shot is radio. This structure is also where brands dump compliance language — do that on-screen, small, not as the talk-track. If a spec is the only reason someone searches, put it in the first caption and show it; burying it at second 18 is how a battery number never gets heard.

Ranked last as a cold-traffic shape, not as a useless one. PDP traffic, Search, and retargeting of product-viewers will tolerate a list. Prospecting will not. If your current AI workflow defaults to this (URL in, bullet-list out), add a rule that the first sentence cannot contain a spec. Force a demo or an objection and keep the list for the caption track. A hybrid that sometimes earns a place on cold: one demo-first open, then a three-line captioned spec list over B-roll. The list is no longer the hook, which is the only thing that was wrong with it.