MakeRoll makes AI UGC ads for US brands on a monthly partnership. You send a one-page brief. We ship 20 to 60 new concepts a month, each with three hooks and a cut for every feed. Nothing reaches you until it clears a 95/100 quality gate.
No discovery call. You get a written scope, priced and dated, within two business days.
Every ad we make is AI-generated, and the people in them are fictional. We don't promise views, sales or ROAS, and none of the ads on our public case pages has run as paid media yet. We do commit in writing to volume, deadlines, the quality gate and a paper trail for every claim.
Put in your own numbers. The calculator shows how much revenue runs through your paid social each month, and what a small lift in it is worth. It's plain arithmetic on your inputs. We aren't forecasting anything.
If one ad in twenty wins, volume is how you find it. Creative share of sales comes from NCS, How Advertising Works (2017). Winner rate, spend share and launch volume from Motion's Creative Benchmarks 2026, which covers 578,750 ads from 6,015 accounts. US social ad revenue from IAB and PwC.
Each plan is quoted in US dollars in your written scope and billed monthly in advance. Every plan goes through the same gate and ships with the same claims review and rights sheet. Bigger plans buy more ads and more frequent drops, and Scale and Pipeline partner add a weekly read of your ad data.
Test the work on your own account before you commit to a month.
One time, credited against your first month
For accounts spending $50k to $200k a month. Top-quarter accounts that size launch about 69 ads a month.
Billed monthly, three-month first term
For accounts spending $200k to $1M a month. The top quarter of them launch about 135 ads a month.
Billed monthly, three-month first term
For brands and agencies that want the machine in-house, plus the output.
Billed monthly, six-month first term
Growth and Scale start with three months, then run month to month with 30 days' written notice. Pipeline partner starts with six.
Every ad ships with its gate report. If you find a delivered ad under 95 on that report, or one that breaks your approved claims sheet, tell us within 14 days and we'll remake it free. We don't guarantee performance.
Plans cover creative only. You run the ads in your own accounts and keep full control of spend, targeting and the ad data you share.
A team of autonomous agents researches, scripts, casts, shoots and assembles. Four gates sit between those stages, and any of them can reject the work before you see it. The rejection counts on steps 03 and 04 come from the logs of our published Booking Wolf run.
Your product, your audience, the offers you're allowed to run and anything legal has ruled out. We write back with the scope and start date. The first ten hook angles come in the same reply.
Agents read your site and your approved claims, then write the scripts, and each factual line points back to a source you can check. You approve the claims sheet before anything gets generated.
Characters are cast as still frames first. Face drift, stray limbs, fake status bars and wrong dates get caught here, where a fix is cheap.
Each clip speaks its scripted line on camera. A transcript check scores it against the script, and a clip that garbles a price, or gets more than about one word in seven wrong, goes back.
Twenty-five checks are measured again from each delivered file, including loudness, captions, safe zones and the first word's timing. Then a separate reviewer ranks the cuts without seeing the scores.
Each concept arrives as a 9:16 master and a 4:5 cut, with captions, end cards, the claims sheet, the rights sheet and the AI labels to switch on when you upload. The master runs the full gate, and the 4:5 cut gets its framing, captions and safe zones checked again.
Partners also get the production ledger for their own ads, down to each paid model call. We share cost breakdowns only after signing.
You share an ad-level export. We read hook retention, thumb-stop rate and cost per result, then build the next drop around what held attention. Growth gets this monthly.
A lot of AI UGC is a fictional customer saying they loved a product they never used. In the US that's a fake testimonial, and the FTC can fine you for each one. So our client formats never pretend to be your customers.
That's the maximum civil penalty per violation of 16 CFR 465, the FTC rule on consumer reviews and testimonials in force since Oct. 21, 2024. It bans a testimonial from someone who doesn't exist or never used the product. It also reaches whoever writes or creates it, and an agency counts.
In client work, our AI characters speak as presenters or in clearly staged scenes. They explain your offer. They never claim they bought it or that it changed their life, because the FTC Endorsement Guides bar presenting anyone as a customer who isn't one.
Prices, results and product facts come from your site or your written approval. The claims sheet travels with every drop so your legal team can check it line by line.
Every client ad with a realistic AI person carries a short on-screen line saying so. New York has required a clear disclosure since June 9, 2026, and TikTok rejects AI ads without its label, so we add the line and tell you which platform label to switch on.
We don't imitate real people. If you want your founder or a spokesperson in the ads, we need their written consent first, and the rights sheet records it.
Models, music and source material are logged for each ad. You get a clear list of what's safe to run in paid media and anything we'd hold back.
We're describing our process here, and it isn't legal advice. Your counsel should review claims for regulated categories such as health, finance and supplements before any ad runs. Rule text at the FTC, the penalty amount in the FTC's 2025 adjustment, and the New York law, S.8420-A, in Cooley's summary.
Three published case studies show the same pipeline that runs your drops, from the first prompt to the gate reports. None of these ads has run as paid media, so there's no audience data yet. Your pilot is where the first performance numbers come from.
These were internal test runs made before any client brief. Some use a selfie format where a fictional person talks as if they used the product, which client work doesn't do. Booking Wolf hasn't endorsed or commissioned these pages, and every person in the ads is an AI-generated fictional character.
Anything else goes to support@makeroll.com. We answer in writing within one business day.
Email support@makeroll.com with the subject line “Business partnership”. Include these five things and we'll do the rest in writing.
Your ad account needs new creative every week. We'll make it and gate it. You keep the paper trail.
