The 3-second rule on Meta

Everyone repeats that the first three seconds of a video are what decide whether someone stays or keeps scrolling. That's true.

The problem is the number you were taught to check it with. Meta deleted it.

Until a while ago there was a metric that counted how many people watched at least three seconds of your video. Today, if a system asks Meta for it, it doesn't get a zero or a warning: the whole query fails with an error. It isn't that the data comes back empty. It's that asking for it breaks everything else you were asking.

I found out the worst way, which is when something that worked stops working with no explanation. And it seemed worth writing down, because a lot of people are quoting a rule and measuring it with an instrument that no longer exists.

The rule still stands. What changed is how you check it.

What to look at instead today

Meta left two things in its place.

The first is called ThruPlay, and it counts the people who watched 15 seconds or the whole video. It's useful for knowing who really paid attention, but it arrives late: it tells you nothing about the opening, which is exactly what you want to know.

The second is the useful one, and it's probably already in your table without you ever having looked at it: plays to 25%. How many people reached the first quarter of the video.

In a twenty-second vertical video, the first quarter is the first five seconds. Not exactly three, but it's as close to the opening as you can get today, and for deciding whether your start works it's more than enough.

The math is simple: divide plays to 25% by impressions. That gives you the percentage of the people who were shown the video who stayed long enough to reach the first quarter.

How to build an opening that stops someone

Before you measure it, you have to make it. And there's less magic here than gets sold: the openings that work are almost always one of six things.

Lead with the price, not the product. "This costs 40 thousand pesos a month." It works when the price is your advantage, or when it's so far from what's expected that it forces a look.

Name the problem out loud before offering anything. Not "let us introduce," but "you've spent three months paying for something you don't know is working." Whoever recognizes themselves stays.

Show someone else saying it. A customer talking lands differently than you talking about yourself. It's the same sentence, but it isn't the same sentence.

Open a question that forces people to stay for the answer. Careful with this one: the line between curiosity and clickbait is thin, and on the wrong side people come in and leave, which is worse than if they'd never come in.

Show it working immediately, with no introduction. No logo, no "hi, we're." The screen doing the thing, from frame one.

Start with something that doesn't match what the person expected to see. That's what makes the thumb stop on its own, because the brain takes a moment to work out what it's looking at. That moment is all you need.

Then comes the order of everything else: problem first, benefit first, story, or testimonial. Any of them works. What doesn't work is changing two things at once.

That's the rule that gets broken most. You reshoot, you change the opening and the ending and the copy, it comes out better, and you have no idea which of the three changes helped. One variable at a time. It takes longer and it's the only way to learn anything.

How to tell whether yours hooks

You have three stretches to check, in this order.

1. The opening

If fewer than 1 in 5 people reach the first quarter of your video, the problem is the opening.

And what matters there is what not to do: don't raise the budget. Putting more money into a video that stops nobody is paying to show more people something that has already proven it doesn't work. Money doesn't fix a bad opening, it just spreads it around.

An ad with this happening is giving away most of what you spend before anyone understands what you sell.

2. The midpoint

If the opening hooks but people drop before the halfway mark, the opening promised something the rest didn't deliver.

Don't touch the start here: it's doing its job. What's failing is what comes after, which probably takes too long to get to the good part or changes the subject.

3. Who you're showing it to

I ran into this one with my own campaigns and hadn't been clear on it before.

I have the same video running in two different campaigns. Same file, same first frame, same everything. In one, 1 in 4 people reach the first quarter. In the other, 1 in 10.

The video didn't change. What changed is who Meta showed it to, because one campaign is optimized for people to do something on the site and the other for them to click. With different objectives, Meta goes looking for different people.

So the same opening can pass and fail at the same time. It's an observation from two of my own campaigns, not a lab experiment, but it makes one thing clear: before remaking the video, look at who you're putting it in front of. The video may be fine and the problem may be the guest list.

The numbers you see aren't the same in Reels

A necessary caveat. The thresholds above are for the feed. In Reels the scroll is faster and more brutal, and everything runs five to ten points lower.

If you measure your Reel with the feed's yardstick, you'll conclude you have a content problem you don't have. Compare each format against its own.

When not to shoot another video

We get to the part that nobody charging you for production has an interest in telling you.

There's a case where the opening hooks, people stay, and you still don't sell. When that happens, the problem isn't the video. It's further down: in your offer, or in the page they land on.

I'll tell you mine, because it's the one I know best.

I've had an ad running for a month. 1 in 4 people reach the first quarter, and of those, almost 2 in 3 make it to the midpoint. Both numbers are comfortably above the threshold. As a creative, it works.

In thirty days it spent 196,663 pesos (about 215 dollars) and brought in 23 scans. Each one cost me 8,551 pesos, about 9 dollars. Of 1,324 people who clicked, 23 scanned.

If I were in the business of selling you video production, this would be the perfect moment to tell you we need to reshoot. But the numbers say otherwise: thirteen hundred people stayed watching, understood, and clicked. The video did its job. What fails is what happens after the click — the page they land on.

Shooting another video wouldn't have fixed anything. It would have spent two weeks and another budget to arrive at the same place, with the difference that I'd now believe the problem was solved.

That's why the order matters so much: first the opening, then the midpoint, then who you're showing it to, and only then — if all of that is fine — you stop looking at the video and start looking at your site.

Frequently asked questions

Is the 3-second rule still valid?
As a principle, yes: the first few seconds decide whether someone stays. What no longer exists is the 3-second metric Meta used to measure it. Today you check it with plays to 25% of the video.

Where do I see how many people watch my video's opening?
In Ads Manager, in the per-ad view. You have to add the "video plays to 25%" column from the customize columns menu: it isn't there by default. Divide that number by impressions.

What's a good retention percentage on Meta?
In the feed, at least 1 in 5 reaching the first quarter. If you get to 1 in 4, you're in good shape. Below 1 in 7, the opening is the problem. In Reels, subtract five to ten points from all of the above.

Does the same video work for Reels and for the Feed?
It works, but it isn't judged the same way. The same video can perform very differently depending on where and to whom it's shown, so compare each format against its own yardstick before concluding the creative is bad.

If my video retains well but I don't sell, what do I change?
Not the video. If people stay watching and click but don't buy or leave their details, the problem is in the offer or the landing page. Remaking the creative in that case is spending on the one thing that was already working.

If you want to know whether your problem is in the ads or further down, scan your site free: I'll tell you in plain terms what your marketing is measuring today and how your operation looks next to your sector's. No agency and no airplane-cockpit dashboards.