Resource Center›How scoring works
Guide 03 · about 8 minutes

The five beats, and the score before you spend.

Two ideas run through the whole product: every ad is built from five parts, and every option gets a confidence number before a penny of media goes behind it. Here's what both actually mean, and what to do when the numbers aren't flattering.

The 5 Beats of Conversion™

Ads that work tend to do the same five things in the same order. That's the whole idea behind the five beats — not a creative theory, just the structure the builder assembles every ad from.

1. Hook

The opening seconds that stop the scroll. A claim, a question, a problem named out loud, something visually surprising. If this beat is weak, nothing after it gets watched — which is why it's first and why it matters more than the rest combined.

2. Problem

Someone naming the frustration your product fixes, in their own words. This is the beat most brand-produced footage is missing completely, because brands are trained to talk about benefits rather than pain.

3. Solution

The product doing its job, with someone explaining what it does. Real hands on a real product, not a packshot.

4. Social Proof

A real person vouching for it. Results, a before-and-after, or simply "I've used this for three months and here's what happened." This is the beat that makes multiple creators saying the same thing read as consensus rather than as an ad.

5. Call to Action (CTA)

Telling the viewer what to do next. Usually the shortest beat, and usually the easiest gap to fill if you're missing it.

Where you'll see this in the app

Your clips live under the Beats tab in your Library, each tagged with the beat it serves. When the system proposes an ad concept, it shows you which clip it has chosen for each of the five sections. On a finished ad, the same information appears as Beats used.

How your clips get tagged

When you upload a source video, the system watches it and listens to what's said. It cuts the video into individual clips, and for each clip it decides which of the five beats that clip serves — and how confident it is about that decision.

That confidence matters. A clip where someone says "honestly I was ready to give up on this problem entirely" is obviously a problem beat. A clip where someone says "yeah, so, anyway" isn't obviously anything, and the system says so rather than guessing.

Clips it's confident about get tagged and become available to the builder. Clips it isn't confident about come back uncategorized, and sit out until you tell it what they are.

The Conversion Confidence Score™

This is the part nobody else does, and it's simpler than the name suggests: before you spend anything, every option you're choosing between carries a number.

When you reach the concept stage of the builder, you don't get one ad — you get several proposed concepts, each one a different way of assembling your clips across the five beats. Each concept card carries a Confidence percentage.

What that number is for is ranking. It exists so that when you're looking at four ways to cut the same footage, you're not picking on taste alone. You can see which assembly the system thinks holds together best, and put your budget behind that one instead of splitting it across guesses and waiting a week for the market to tell you.

What it is not

It is not a prediction of your ROAS, and nobody can honestly give you one of those before the ad runs. It's a relative measure — useful for comparing these options against each other today, not for comparing one brand's ads against another's.

Reading the number

Treat it comparatively, not absolutely. The useful question is never "is 78% good?" — it's "which of these four is highest, and is the gap big enough to care about?"

What you're seeingWhat it usually meansWhat to do
One concept clearly ahead of the others The system found a strong assembly — usually because one of your clips is a genuinely good hook Build that one. Consider building the runner-up too, as a test against it
All concepts clustered closely together Your footage is consistent, and the choice between them is genuinely down to your judgement Pick on brand fit and build. The score isn't going to decide it for you
All concepts low Usually a library problem, not a concept problem — a beat is thin or missing across all your footage See what to do about a low score below

A useful habit: build the top concept, run it, and then compare what actually happened in your ads manager against what the score said. After three or four ads you'll have a feel for how the number maps to your account specifically, which is worth more than any general rule we could give you here.

What to do about a low score

A low score is not a verdict on your product. It's almost always telling you that one of the five beats is missing or weak across your whole library — and that's a footage problem with a cheap fix.

  1. Find the thin beat Open your Library and look at the Beats tab. Count how many clips you have for each of the five. There's usually one with almost nothing in it, and it's usually problem or social proof.
  2. Shoot or source one short clip to fill it You don't need a reshoot. You need one person, on a phone, saying the missing thing out loud for twenty seconds. A customer, a founder, a member of staff who genuinely uses the product.
  3. Add more creators, not more footage of one Ten clips from one person gives the builder one voice to work with. Three clips each from three people gives it far more, because it can take the hook from one and the proof from another. This is the single highest-leverage change most brands can make.
  4. Check for uncategorized clips You may already own the footage you need — it just isn't tagged. Assign beats to anything sitting uncategorized and regenerate.
  5. Regenerate concepts With the new material in the library, run the concept step again. You should see the numbers move.
If you can't shoot anything

Creator Sourcing exists for exactly this. Real creators, briefed on the specific beat you're missing, with the content cleared for paid use.

Uncategorized clips

When the system isn't confident enough about which beat a clip serves, it doesn't guess — it leaves the clip unlabeled in your Library and marks it uncategorized.

Two things worth knowing about these:

  • They're invisible to the builder. An uncategorized clip won't be used in any ad until it has a beat. It isn't deleted and it isn't lost — it's just sitting out.
  • You can categorize them yourself. Open the clip's preview and assign the beat you think it serves. Your choice is used from then on. If you know a clip is a great hook and the system didn't spot it, you're allowed to be right.

It's worth a five-minute pass through your uncategorized clips before you conclude your library is thin. Brands regularly find they already had the missing beat, unlabeled.

Common questions

What counts as a good score?

There isn't a universal threshold, and we'd be making one up if we gave you one. Use it to rank the concepts in front of you: the highest of the set is the one to back. Over a few ads you'll learn what a typical range looks like for your own footage.

Does the score predict my ROAS?

No, and be skeptical of any tool that claims to. It's a pre-spend ranking signal to help you choose between options, not a forecast of performance. What it does buy you is that you stop splitting budget evenly across four guesses.

Can I override how a clip was tagged?

Yes. Open the clip and assign the beat yourself. This is the intended fix both for uncategorized clips and for anything you think was filed wrongly.

All my concepts scored low. Is my product the problem?

Almost never. It's nearly always that one of the five beats is missing across your library — most often the problem beat, because brand footage tends to skip straight to benefits. Add one clip of someone naming the frustration, and regenerate.

Do I have to build the highest-scoring concept?

No. You choose which concepts to build, and you can build more than one. The score is advice, not a gate. If a lower-scoring concept is a better fit for your brand, build it — and consider building the top one as well so you have something to test it against.

Does every ad have to use all five beats?

Social media ads are built around the full five. Product videos for ecommerce pages use a shorter structure, since someone already on your product page doesn't need as much setup. The builder handles this difference for you based on which ad type you chose.

Didn't answer it? Ask a human.

Email support@videoforce.ai and tell us what you were trying to do. Screenshots help.

Back to the Resource Center