YouTube Ad Script: Five Beats and the Ten-to-One Ratio

How do you write a YouTube ad script people do not skip?

Five seconds. That is all a skippable YouTube ad is sure to get.

Then a Skip button shows up in the corner.

Most people write a YouTube ad script like a TV spot. A logo, a warm hello, a slow build. The viewer is gone before the good part arrives.

We write it as a race against that button. Every second has one job. The jobs come in a fixed order.

This method is for offers that have to sell on day one. A great ad with basic targeting beats a weak ad with perfect targeting.

So fix the script before you touch the settings.

Build the ad as a timeline, not a smooth story

Most people want a YouTube ad to flow from start to finish.

Nice music. One clean story. One ask at the end.

That is the wrong goal. Think of a plain car with a rocket bolted on. Ugly, but it gets there fast.

Why ugly? Because the crowd changes while the ad plays.

At ten seconds, almost everyone is still there. By forty, the eager ones have clicked and the bored ones have skipped. At ninety, you are talking to a smaller, warmer group.

So each part of the script talks to whoever is still watching. That is why the ask comes more than once.

The clock has three phases. The first five seconds are awareness: viewers cannot leave yet, so they just see what this is.

Six to thirty seconds is attention. The job is to keep them watching long enough to want the click.

After thirty seconds comes focus. This is where clicks are won or lost.

The five beats of a YouTube ad script, second by second

Here is the full shape. Five beats, one job each, for one person talking to the camera.

Beat one, the hook, 0 to 5 seconds. A strong statement or a question. It usually says the opposite of what the viewer thinks.

Beat two, the expansion, 5 to 30 seconds. Make the hook real in a few sentences: a benefit, a picture, or a use case. Once viewers start thinking, they stop reaching for the button.

Beat three, proof, about 30 to 60 seconds. One line on who you are. Then two or three customers.

Beat four, the ask plus the answer. Say what they get when they click, and why it fixes the problem from the hook. This is your call to action.

Beat five, the close, about 1:30 to past 2:00. Ask again, with a reason to act now.

Two minutes sounds long. It is not long for the people who chose to stay.

Two limits on this shape. It is built for a person on camera, so image and voiceover ads need a different plan. And it is built for offers that must pay back fast, not software that earns its money over years.

Treat it as a map, not a law. The order matters most. The exact seconds can move.

What should the first five seconds of a YouTube ad say?

Say the opposite of what your viewer believes. That style of hook usually works best.

For example: "Want to lose weight fast? Keto is not the answer."

The viewer thought keto was the answer. Now they are not sure.

That doubt is a small crack. The next 25 seconds widen it.

Four hook types to rotate: a bold statement, a question, a worry that never leaves their head, or a belief they gave up on that can actually change.

Where do you find those beliefs? Your support inbox. The same wrong idea shows up in emails every week, and it belongs at the front of your ad.

Here is how you pay, under Google's rules. On cost-per-view bidding, you pay when someone watches 30 seconds, or a shorter ad to the end, or interacts with it.

So a viewer who skips at second four costs you nothing. So does one who leaves at second 29.

That makes the first 30 seconds a free filter. People who were never going to buy can leave before the meter starts. You pay for the ones who stay.

The hook still has one job: keep the right person. A clear hook keeps them and lets the rest go, for free.

One catch. That rule only holds on cost-per-view bidding; on other bid types you can pay every time the ad is shown. The settings box near the end has the details.

More on why openings work at all: the first three seconds and hooks built on change plus hope.

Use three customers to answer three different doubts

Proof comes at about the 30-second mark. They stayed. Now you have to earn their belief.

Start with one authority line: who you are and how many people you have helped. Use your real number.

Then two or three customers. The rule that matters: each testimonial answers a different doubt.

Take a credit repair ad. One customer had an old collection mark that had sat there for years. Another had a score under 650 and thought it could not move in two months.

A third finally bought a car. Three customers, three doubts, three answers.

Three people saying "it worked" is one testimonial read three times. Three people knocking down three doubts is a real argument.

Our rule on top: real customers, their real words, and results they actually got. The FTC bans fake reviews and testimonials, and a made-up customer is not a shortcut. It is a legal risk.

Need more kinds of evidence than customer stories? See the proof grid.

Ask more than once, because the crowd keeps changing

Beat four joins the ask to the answer. Tell them what they will see after the click, and why it solves the problem from the hook.

Be literal. "Click below to watch the free training on the three steps" beats "learn more" every time.

Then ask again in the close. And again. Each ask gets its own reason, because each ask is for a different slice of the crowd.

Pushy? Not to the person who tuned in at minute one. For them, it is the first ask.

Plenty of sample closes lean on "last and final chance" urgency. We do not write that unless it is true.

Our close uses a true reason. A real deadline, a real price change, or the plain cost of waiting another month.

Fake urgency gets the click. Then it loses the buyer at the next page.

Test the hook first: five ads, one body

Shoot one body and five openings. That is your first batch.

Only the first 5 to 10 seconds change. Everything after stays the same. So any difference in results comes from the hook.

Why start there? One hook can make or break the whole ad. It is also the cheapest part to reshoot.

Pick the top one or two winners. Then shoot a second batch of five that changes only the 10 to 30 second section.

That middle section is where click rate moves. That second batch is where the best ads tend to show up.

This is plain A/B testing: change one part, keep the rest still. We use the same logic in one-variable ad testing.

How to waste a shoot day in four easy steps. Step one: write five completely different ads. Step two: launch all five at once.

Step three: crown a winner after one day. Step four: have no idea why it won. Congratulations, you bought a damn expensive coin toss with a camera crew.

Want a steady stream of tests after this? Our creative testing engine shows how to keep it going.

What is a good view rate to CTR ratio? About ten to one

View rate should be about ten times your click-through rate. Ten to one is the golden ratio.

Two definitions first. View rate is views divided by impressions, in Google's own definitions. Click-through rate is clicks divided by impressions.

So divide your view rate by ten. That is the lowest click-through rate you want.

Here is an approximate example from one advertiser's account. Treat these numbers as rough, not a measured result.

View rate was roughly 21.4 percent and click-through rate roughly 1.88 percent. A tenth of 21.4 is about 2.1. So that ad was a little short.

Higher is better. Ads that hold this ratio tend to keep working. It is a rule of thumb, not a Google benchmark.

What does Google say? That low clicks are normal in video campaigns. Clicks are not usually the main goal.

That is fair for brand ads. For an ad that has to sell today, there is no sale without the click.

So read the two numbers together. Google says a low view rate can mean the start of your ad was weak, so fix the hook. A good view rate with a weak click rate points at the middle, which is the job of your second batch.

When the cost per sale runs too high, read your numbers top down. Think of a grade point average. You cannot raise it directly; you raise it one class at a time.

Cost per sale is the average. Click rate and view rate are the classes. Cost per click, cost per view and cost per thousand impressions are the tests inside each class.

Most of the time, the problem lives one level down. For the full tree from feed to sale, see reading ad data.

Fix a dying YouTube ad by changing one word

Good ads stop working over time. When one does, do not start over. Look for the one word that is out of date.

For example, one finance ad was doing great, then it faded. Near the start, the person on camera pointed at a date when something big would happen.

That date had passed. So the advertiser found the next date that mattered just as much.

They swapped that one word and ran the same ad again. It took off again.

Every market has an anchor like that. It is the thing on everyone's mind.

In finance it is a date. In credit or dog training, it might be a news story or a change in the law.

Put your market's anchor in the hook. Then move their attention from that event to your offer. We cover this in the news hook.

Check every week. Compare this week to last week in your ad account, column by column.

Spending a lot? Check every three days too.

For example, one account showed click rate up 50 percent and cost per click down 20 percent. Good news. Now ask why: a new audience, or a better ad?

The settings this method touches are below, in one place.

Run it with AI

The prompt below runs this whole method. Paste it into your preferred AI and answer eight questions.

It writes five hooks, one body and a plan for your second batch. Give it your view rate and click rate, and it checks the ratio too.

It never invents a customer, a result or a number. Missing one? It asks.

This week: first, read ten support emails and write down the wrong belief that repeats. Second, run the prompt and shoot one body with five hooks. Third, after a week, divide your view rate by ten and hold it up against your click rate.

Want a second pair of eyes on the script? Send it to us; you are not signing up for anything.

The skip button is not your enemy. It is a stopwatch that shows you where the script broke.

Five seconds to earn the stay. A ten-to-one ratio to know it worked.