UGC ads work on cold audiences: what the lift tests show, and how to test yours
UGC ads work best as cold-audience ads that look like posts: Meta's conversion lift studies show a lower cost per incremental sale, while most of the famous percentages trace back to vendors grading their own work.
UGC ads are paid video ads made in the style of User Generated Content: one person, a phone camera, a product and a few spoken lines. Ask whether they work and the first page of Google answers yes with a pile of numbers: four times the clicks, half the cost per sale, a 38 percent lift. We read that page on 13 October 2026, and almost none of those numbers had a method attached.
Do UGC ads work, measured how, by whom, and for which ads? The short version: yes, on cold audiences, for products people buy without a long think, and when the ad looks like a post down to the last frame. The long version comes with its sources, the cases where the format loses, and a test you can run on your own account.
Do UGC ads work?
Yes, mostly in cold prospecting: shown to people who have never heard of the brand, a UGC ad usually buys a sale more cheaply than a polished brand film. The best evidence for that comes from conversion lift studies, which measure the sales an ad caused rather than the sales it happened to sit next to.
That answer has three conditions, and the rest of this post is about them. The audience has to be cold, because a viewer who already knows the brand is past the introduction a stranger needs. The product has to be an easy decision, the kind someone buys from a phone on the same evening. And the ad has to pass as a post, since the whole advantage of the format is the second or two in which the viewer has not yet noticed it is an ad.
Outside those conditions the format still runs, it just stops having an edge. That is the honest shape of the answer, and it is also why "do UGC ads work" is a question you settle per product, with a test, and settle again every few weeks as the ads tire. The UGC ads guide covers what the format is and what it costs.
Where do the famous UGC numbers come from?
Mostly from companies that sell UGC, publishing their own case studies. Opascope, a performance agency, made that point in its guide to UGC ads on Meta (last updated 17 July 2026): it traced the 35 percent, 38 percent, four-times and 25 to 50 percent figures that circulate in the category back to vendor case studies with no primary source behind them.
The numbers could still be right, and nobody outside those companies can check them. A case study is a survivor: the brand whose test went well gets the write-up, and the ten whose tests went flat do not. It also tends to compare a fresh UGC ad with a tired brand ad, so part of the lift belongs to novelty. And it reports click-through rate (CTR) more often than cost per sale, because clicks move first and look bigger.
Opascope's own figures need the same discount, since they come from accounts it manages, and the guide says as much. Its most useful number is a cost: Meta's cost per thousand impressions (CPM) rose 30 to 40 percent from 2024 to 2025 across the accounts it runs. No creative format absorbs a rise like that on its own.
Hey there :) clicking this would help us so much!
What is the strongest evidence that UGC ads work?
Meta's conversion lift studies on partnership ads, which are creator videos run as ads from the creator's own handle. In a white paper Meta published on 7 October 2026, 20 conversion lift studies from advertisers in Europe, the Middle East and Africa, run from January to June 2025, found that campaigns with at least 30 percent partnership ads paid a 45 percent lower median cost per incremental purchase than business as usual.
Three details make this stronger than a case study. A conversion lift study holds out a control group that never sees the ads, so it counts the sales the ads caused and leaves out the ones that would have happened anyway. Twenty studies make a sample, where a case study is one survivor. And Meta reported a median, so one spectacular brand cannot carry the result. The same paper reports a 35 percent higher median incremental return on ad spend.
Now the discount. Meta sells the format it measured, ran the studies in its own tooling, and stated the 45 percent result at 92 percent confidence, below the 95 percent most analysts expect; PPC Land, which read the 54-page paper and wrote it up on 11 October 2026, flagged all three. Partnership ads are also a specific case: creator content on a creator's handle, mixed into an existing campaign. The evidence is good that this mix lowers the cost of a sale. For a UGC-style video run from the brand's own page, the effect has to be measured in your own account, which the test further down is for.
Why do UGC ads work when they do?
Because for a second or two they do not look like ads, and that is when the viewer decides whether to keep watching. A person talking into a front camera in a real room matches everything else in a TikTok, Reels or Shorts feed, while a studio shot with a logo animation announces itself before the first word.
We make AI-generated UGC video ads, and this is the rule we turned into code instead of leaving it to taste. A Go Rocket video reads as a real person talking about their experience: no call-to-action pill, no buttons, no badges and no price tags on the video. The only text on screen is the words being spoken plus one hook line, and the call to action is the presenter's own closing sentence. Every one of those overlays is something an ad tool would happily add, and every one is a tell that pulls the video out of the feed!
The other half is structure. Our arc is fixed: hook, problem, proof, payoff, call to action, with each role's place set by the video's length instead of chosen by a model, and the hook lands inside the first two seconds. Opascope's testing points the same way from the other side: in its accounts, storytelling visual hooks beat talking-head openers 86 to 14, which says the first seconds decide more than the UGC label does.
When do UGC ads not work?
They lose their edge with warm audiences, with expensive or considered purchases, and when a single video is asked to carry a whole campaign. Opascope names cold direct-to-consumer prospecting, mid-priced ecommerce, problem-aware audiences and skeptical categories as the places UGC beats studio work, and the cases outside that list are where its advantage fades.
Retargeting is the clearest one. Someone who visited the product page last week has met the brand already; another stranger telling them about it adds little, and a clear offer or a polished demonstration often does more. Considered purchases are the second: software sold to a procurement team, a mattress, a holiday. That buyer wants specifications, comparisons and proof, and a 15-second reaction from a stranger is the wrong instrument for the job.
The third failure is self-inflicted. A UGC ad that claims something the speaker never lived, "I've used this every day for six months", turns the format's main asset, believability, into a liability, and a scripted testimonial like that can also break the Federal Trade Commission (FTC) endorsement rules that our post on UGC legality covers. Our script validator rejects any script that claims usage history or measurable personal results, because a generated presenter has no history to report.
How do you test whether UGC ads work for your product?
Run one clean comparison: the same audience, budget, objective and landing page, with the creative as the only difference, judged on cost per purchase once the learning phase is over.
Start by checking that conversions are counted the same way on both sides, because a tracking difference will happily pose as a creative difference. Then make several UGC variants instead of one, at least three hooks on the same product, since the hit rate is low: Opascope puts it at 5 to 7 percent of creative angles on Meta, and Motion's Creative Benchmarks 2026, which we used in our post on UGC marketing, found that 4 to 8 percent of creatives win. Leave each ad set alone until it exits Meta's learning phase, which takes about 50 optimization events; before that, the cost per result swings for reasons that have nothing to do with the video.
Judge on cost per purchase or, if your account qualifies, a lift study, because CTR moves first and flatters any new video. If UGC wins, keep the winner and replace it before it tires: in Opascope's experience hooks fatigue within two to four weeks at scale. If it loses on a warm audience and wins on a cold one, you have found the most common result there is, and you can split the budget accordingly. A test like that costs a few hundred dollars of media and settles the question better than every statistic on the first page of Google!
Do AI UGC ads work as well as filmed ones?
As tests they often do, and a published comparison is close: in aubado's Meta test of 220 ads, published in March 2026, AI-made ads clicked less than human creator ads and still returned more per ad dollar. We went through that test, and the three defaults that ruin AI ads, in our post on AI UGC ads.
What AI changes is the price of the testing step above. Three hooks from a human creator cost three fees and a few weeks; generated, they take minutes, which makes the honest workflow, many variants with most of them killed, affordable for a founder with one product page. A generated presenter makes the believability rule stricter, since it can react to a product and can never testify about one. If you want to try that on your own page, our UGC ad generator starts from the link to your product page.
The human creator stays the right call when the ad needs a real customer's story, and nothing generated should pretend to be one!
FAQ
UGC ads are most effective on cold audiences for products people buy after a short think, where Meta's 2026 lift studies found a 45 percent lower median cost per incremental purchase for campaigns that mixed in creator ads. With warm retargeting audiences and expensive considered purchases the edge shrinks or disappears.
For a brand selling to people who have never heard of it, yes, as long as it tests several UGC ads at once and judges them on cost per purchase. For retargeting and high-priced products, a clear offer or a polished demonstration often does better.
A UGC ad is filmed like a post, one person talking into a phone in a real room, while a commercial is produced and branded from the first frame. UGC usually wins with cold audiences, and commercials hold up better with people who already know the brand.
Most UGC ads run 15 to 30 seconds, and the first two seconds decide whether anyone watches the rest. Go Rocket makes its videos in 15-second segments, so the hook, the problem, the proof and the close each get only the time they need.
Paid social ads still work, though they cost more: Opascope saw Meta's cost per thousand impressions climb 30 to 40 percent from 2024 to 2025 in the accounts it runs. The ads that still pay for themselves tend to look like posts and get tested in volume.