How to make UGC style ads without a creator

Shoot the situation, not the spokesperson

The reason a creator video works is not the creator. It is that the video is set somewhere real and starts in the middle of something.

So replace the person with the place. Hands, not faces. The counter, the van, the shelf at the wrong time of day, the phone with the message on it. These are cheap to make, they are what your customer's week actually looks like, and none of them asks anybody to believe a stranger is endorsing you.

A useful rule: if a shot needs a face to work, the script is doing too little. Rewrite the line so the picture can be of the situation instead.

If you are stuck for what the situation looks like, write the line first and then ask what was in the room when it happened. Four shots cover most businesses.

• The before. The desk, counter, inbox or van at the moment the problem shows up. Six in the evening, not nine in the morning.
• The object with the problem attached. The spreadsheet with forty tabs. The box that arrived damaged. The phone showing the fourth missed call.
• The room after. Same place, different state. This does the work a results claim would do, without making one.
• A detail nobody would bother to fake. A scuffed floor, a handwritten note, weather through a window. This is the shot that makes the other three believable.

Leave out anything with writing in it you did not control, any screen with text small enough to read, and any close view of fingers doing something intricate. Generated footage gets those three wrong often enough that framing away from them is faster than fixing them.

Write it to be read, not heard

Most vertical video is watched muted, so the words go on the screen and they have to survive being read in about two seconds each.

That is a different craft from writing narration. Short lines. One idea each. No sentence that needs the one before it to make sense, because somebody may have started watching halfway through.

The shape that works for thirty seconds is roughly: something true about their week, the thing that makes it worse, what changes, what it feels like afterwards, then what to do. Five or six lines. The first is the only one most people will read, so it gets the most work.

Four mechanics decide whether a line survives being read.

About seven words, one clause. If a line needs a comma to work, it is probably two lines.

Give it the time it needs and no more. Read it out loud at a normal pace and add half a second. A line still on screen after it has been read is dead air, and dead air is a scroll.

Put the words where the platform is not. Every vertical feed covers the bottom of the frame with the caption, the account name and the buttons, and the top with its own controls. The middle band is the only safe one.

Do not write what the picture already says. If the shot is a kitchen in the evening, the line does not need to mention the evening. Use the words for the part a picture cannot carry.

Then read the whole thing on a phone, at arm's length, in daylight. Text that looks generous on the monitor you wrote it on is often unreadable in the only viewing condition that counts.

The opening line is the whole thing

Two ways to write it, and only one of them works.

Writing about yourself: "We make weekly meal kits for busy families." Nobody is looking for you.

Writing about them: "It is Wednesday night, and dinner came back untouched again." That is a thing somebody recognizes before they have decided whether to care about your company.

The second is harder to write from nothing, which is why it helps to start from research about who your customer actually is rather than from a description of your product. If you have ever written down what customers say when they complain, that file is where your first lines are.

If you do not have that file, an hour builds a usable one. Go through your support inbox, your sent messages to customers and the reviews of the two competitors nearest you, and copy out every sentence where somebody describes the problem in their own words. Do not tidy the wording as you copy. Tidying is where their language turns back into yours.

What you are hunting is not the complaint. It is the clause before it, where they set the scene. Somebody who writes we were three days from the deadline and the file would not open has handed you an opening line and a first shot in one sentence.

Five shapes that work once you have the raw material.

• The time stamp. It is Sunday night and next week's schedule still is not done.
• The count that is too high. Fourth time this month.
• The thing they say to themselves. I will do it properly next week.
• The small humiliation. Asking a client you like for the same invoice twice.
• The flat observation, no adjectives. Nobody opened the report.

None of them is a question. Openers phrased as questions, are you struggling with, do you ever find that, read as advertising instantly, because nobody describes their own week in that form. The longer version of this exercise is in market research for startups.

Make more than one, and change one thing

A single video tells you nothing. The feed is noisy enough that one result is mostly luck.

Make three that differ in exactly one way, usually the opening line, and run them against each other. Same scenes, same music, same ending. When one wins you know why, which is the only useful outcome of a test.

This is the part that was impossible with a creator, because three versions meant three shoots. It is the real advantage of making them without one, and most people waste it by making a single video and staring at it.

Vary in this order, because you only learn from one change at a time. The opening line first, and it is usually the only variable that moves the result by much. Then the first shot, the same words under a different picture. Then the instruction at the end, because the opening decides who watches and the ending decides what the watchers do. Music, type and pacing last.

Start every version at the same time, because a video published three days later is competing in a different week. Then leave them alone. Reading the numbers at hour two and pausing the loser is how founders declare a winner from noise. Stop when one is ahead by a margin you would still believe if you saw it a second time. If two are close, they are the same video and the gap is chance.

What to look at afterwards

Two numbers, in this order.

How many people are still watching at three seconds. This is the opening line's score and nothing else. If it is low, no amount of fixing the rest will help.

How many finished. This is the script's score. A strong opening with a weak middle shows up here as a cliff halfway through, and the fix is usually that you spent two scenes on the same idea.

Everything else, including likes, tells you less than either of these and takes longer to read.

Two things to add about reading either number.

Judge against your own last video, not a published benchmark. Retention figures quoted online are averages across every account, category and length, and they move whenever a platform changes. Your own previous video is the only comparison that holds those steady. Keep one sheet: date, opening line, length, how many were still watching at three seconds, how many finished, and what happened afterwards. Six rows in, you will know more about your audience than any benchmark can tell you.

Decide the sample before you look. A few hundred views is a mood, not a measurement. Write down in advance how many you will wait for and what result would make you change the script, then hold to it, because afterwards everything looks like a trend.

Where the drop sits tells you what to change. In the first two seconds it is the opening line or the first shot, and nothing else is being read, so nothing else is the problem. In the middle, two beats are saying the same thing, and cutting one usually moves the cliff. At the end, with nobody acting, the instruction was either vague or asked for more trust than thirty seconds buys, which is what a demo booking asks for and a page does not.

Above both numbers, once you are spending anything: what it cost to get one real conversation or one signup. A video with poor completion and three good conversations beats a video everybody watched and nobody acted on. Watch time is the diagnosis. The conversation is the result.

What the format is actually doing

The look is worth understanding, because every decision further down follows from it.

The format is named after user generated content, meaning video made by ordinary customers rather than by an agency. What a viewer reads in it is not a claim, it is a report. Somebody in a kitchen with a phone is a peer describing their week. A studio commercial is a company describing its product. The first gets the benefit of the doubt for about two seconds longer, and two seconds is most of what you are fighting for.

Two things do the work. The video matches the feed it sits in, so it is read as content first and judged as advertising second. And recognition persuades where an argument cannot, because nobody is talked into anything in thirty seconds. They are reminded of something they already believe about their own week.

Which tells you what breaks the spell, and it is almost never the missing face.

• A logo on screen in the first second.
• Music that swells.
• Color so even it could only have come from a grading tool.
• The advertising voice in writing: introducing, proudly, designed for.
• A shot composed like a magazine cover, which is the one thing nobody films by accident.

Any of those announces what the viewer is watching before the words get a chance. Almost everything else in this guide is about protecting those first two seconds.

How to write a script that does not sound like an advertisement

Most founder scripts fail the same way. They are accurate, they are about the product, and they sound like every message the viewer has already trained themselves to skip.

The test takes five minutes. Read the script out loud and mark every sentence whose subject is your company. Rewrite each one so the subject is the viewer, or the thing in front of them. Usually nothing else has to change.

We built a scheduling tool that saves small teams hours every week.

You moved the same meeting four times this week.

Both are about scheduling. Only one is about the person reading it.

Six phrases that mark a script as an advertisement.

1. Introducing. Nobody introduces anything in conversation. Start in the middle of the situation.
2. Solution. The word a company reaches for when it does not want to say what the thing does. Say what it does.
3. Easily, simply, effortlessly. They promise the reader's experience before the reader has had it. Show the shorter version of the task and let them draw the conclusion.
4. Superlatives. Best, fastest, number one. Each needs evidence you have no room for, or it is quietly ignored.
5. A stack of benefits. Three of them read as a slide, not a story. Pick the one that hurts most and spend the video on it.
6. Learn more. It asks for a decision without saying what is on the other side. Say what happens when they tap.

Rhythm gives a script away faster than any single word. Advertising copy is written in balanced, finished sentences. People are not. Two short lines, one longer, then a fragment. A line that is three words. If every line is the same length it reads as written, even when every word in it is right.

And write the script before you think about pictures. Founders who start from the footage end up describing the footage, and what comes out is a slideshow with narration.

A thirty second skeleton you can fill in

Here is a structure you can fill in this afternoon. Six beats, one shot each, about thirty seconds.

1. Recognition, three seconds. One line about their week over the shot of the situation. No logo, no company name, no setup. This beat decides whether the other five are seen, so give it more rewriting than all of them together.
2. The cost, four seconds. Not the problem again in different words. What the problem costs them. Late invoices is the problem. Chasing a client you like for money is the cost.
3. The turn, five seconds. Something changed. Factual and small. This is where scripts start shouting, and it is the worst possible moment for it.
4. The proof, six seconds. One concrete detail that could only be true. A number you actually hold, a step that used to be five steps, a real screen from your own product. Vague here and the whole video becomes an assertion.
5. The after, five seconds. The same room in a different state. Quiet, and no claim.
6. One instruction, four seconds. A single action, said plainly, held long enough to read twice.

Two notes on the shape. The join between beat two and beat three is where a video is either honest or not, because that is where companies habitually promise what they have not shown. And beat six is one instruction. A video offering a website, a discount code and a follow gets none of the three.

For a fifteen second version, cut beats three and five, never beat two. The cost is what makes anybody care, and it is the first thing founders delete.

Claims, and the line not to cross

The advantage of the format is that it reads as a person's report rather than a company's claim. That is also where the exposure is, because a report carries an implied endorsement, and an endorsement has rules attached to it.

One line, and stay on the right side of it.

A situation is yours to describe. Nobody has to verify that dinner came back untouched on a Wednesday. It is a scene, not a representation about your product.

A result is a claim, and a claim needs evidence you actually hold. Saves six hours a week. Cuts costs by a third. Doubles replies. If a customer asked you to show your working on any of those, you should have something to send them. If you do not, it does not go on screen, and putting it on screen in smaller type is not a fix.

A person who does not exist cannot recommend you. Here it stops being a marketing question. A generated figure talking to camera about how much they love your product is a testimonial from nobody. The Federal Trade Commission's guidance on endorsements, influencers and reviews turns on whether an ordinary viewer would take something as a genuine opinion from a real user, and a convincing synthetic customer clears that bar without trying. Other countries take a similar view. Shooting the situation rather than the spokesperson keeps you out of the question entirely.

Two habits that cost nothing. Label generated footage wherever a viewer could mistake it for a recording of real events, because every major platform provides the setting and expects it to be used. And get permission in writing before a real customer appears, including their words, their premises and their logo, since a verbal yes on a call is not what you will want when the video is running as a paid advertisement.

None of this makes the video weaker. The scripts that work are specific about the situation and modest about the result, and modest about the result is what a founder can actually defend.

Where to spend, in order

None of this needs software. A script in a notes file, four shots taken on your own phone during an ordinary working day, captions added in the editor already on that phone. Founders skip this because it feels beneath the ambition, and it is the fastest way to find out whether the script works, which is the only genuinely hard part.

Spend in this order, and only once the step above has stopped being the thing in the way.

1. Nothing. Your own footage, three versions, one changed line. This answers whether anybody recognizes themselves in your opening.
2. A small paid push behind the winner. Enough to reach past your own followers, who are the least representative audience you will ever test on.
3. Generated footage, once the script is settled and you need scenes you cannot film, or eight versions rather than three.
4. A real creator, last. Handing a proven script to a creator is a different transaction from asking one to invent your message, and it costs less.

The trap is buying step three before doing step one. Generated video of a script nobody recognizes themselves in is a weak advertisement made faster, and the speed makes it worse, because you will publish forty of them before you notice.

When the script is settled and the making is what is in the way, the video tool inside LaunchValid writes the beats from research about your own market rather than from a description of your product, shows you the script before anything is rendered, and lets you remake one scene instead of the whole video. What it cannot do is tell you what a customer said on the phone last week. That sentence is still the most valuable thing you own, and if you want this format as a habit rather than a single test, faceless video marketing covers the cadence.