Every guide hands you the same hook-problem-solution skeleton. Here is how to write UGC ad scripts that convert: fill the blocks with your customers' own words.
You brief three creators with the same tidy script template. Hook in the first three seconds, name the problem, show the product, drop one piece of proof, end on a CTA. The videos come back on time. The structure is textbook. And all three die in testing inside a week, with the comments politely explaining why: "sounds like an ad."
I have written, bought, and buried a lot of these scripts. The template was never the problem. The words inside it were.
The takeaways
A working UGC ad script runs 20 to 40 seconds and moves through five blocks: a hook in the first three seconds, the problem named in the viewer's terms, the product introduced as the mechanism that fixes it, a short demo, then one piece of proof and a single CTA. That is the whole skeleton.
Every guide on page one of Google agrees on this, and they agree because it is right. It is plain direct response compressed into vertical video. But notice what that agreement means: the structure is now a commodity. When everyone bidding against you starts from the same five blocks, the skeleton stops being an edge. Your edge has to come from somewhere the template cannot give you.
From your customers, in their original phrasing. The useful sources are the places people complain without a marketer in the room: Reddit threads, Amazon reviews, App Store and Google Play reviews, niche forums. Collect quotes, and when you collect them, keep the exact wording. Two to four verbatim phrases per quote is the level of fidelity that matters, because the phrasing is the asset.
Here is the difference in practice. A template-filled problem block says: "Struggling to stay motivated with your workouts?" A research-filled one says: "I stopped logging on day four because it felt like homework." The second line came from a one-star review of a competitor. It is specific, it is mid-thought, and a viewer who has lived it recognizes it instantly.
I spent months at a €150k-per-month ad budget learning this the slow way. Scripts written from my own head plateaued. Scripts assembled from review language kept finding new hooks long after I had run out of clever ideas.
Borrow a sentence a customer already wrote. The most reliable hooks I have run were lifted, nearly verbatim, from a complaint: first person, oddly specific, starting mid-thought. "I returned two of these before I figured out what the problem was" beats anything a brainstorm produces, because no copywriter invents that sentence.
A creator can also deliver a found sentence more credibly than an invented one. It scans as something a person would say, because a person did say it.
When you have a pile of complaints, rank them by how much each one hurts, not by how often it shows up. A pain that appears in 40% of reviews but reads as mild annoyance makes a weaker hook than one that appears in 10% and reads as a small crisis. Intensity is what stops a thumb.
Schwartz's five stages run from unaware (does not know the problem exists) to most aware (knows your product, needs a reason to act now). A problem-aware viewer and a most-aware viewer should never get the same script.
For the problem-aware viewer, the hook names the pain and the script spends its time on the mechanism. For the most-aware viewer, the pain is old news; the script opens on the offer or a sharp proof point and gets to the CTA fast. Writing one script per stage sounds like more work. In practice it is less, because you stop trying to write a single script that does five jobs and fails at all of them.
Tag each quote you collected with the stage its author was in. The quotes then sort themselves into scripts.
Shoot one body with three or four swapped hooks and let the auction vote. One caution from my own bruises: the hook that leads after two days is often just the lucky one. I wrote up how I tell a winner from a coin flip if you want the longer argument; the short version is to decide your conversion threshold before the test starts.
The research grind above (five sources, exact phrases, awareness tagging) is the part I ended up building into Adscalr. It pulls quotes from Reddit, Amazon reviews, both app stores, and custom forums, keeps two to four exact phrase markers per quote, maps each quote to a Schwartz stage, and ranks the pain map by intensity. That research feeds its script generation, and a copywriting critic scores every concept against six direct-response principles, quoting the offending line when one fails.
I want to be plain about scope: Adscalr writes UGC scripts and creator briefs and builds motion storyboards. It does not render finished video. A creator still shoots it, and a human approves everything before it spends a cent. If the research-to-script loop is your bottleneck, that is the ad-creation side of the product.
This is the thinking behind Adscalr.
See the product →