The first three seconds decide the other ninety

Almost nobody stops watching a video at the end. They stop at the start — before they have consciously decided anything. If you are trying to work out why a video underperformed, the answer is usually in a window shorter than the time it takes to read this sentence.

Open the retention graph on any video that did not work. The shape is nearly always the same: a cliff in the first few seconds, then a gentle slope. Editors spend most of their time on the slope. The cliff is where the audience actually went.

This is not a trick of the algorithm. It is how people behave when the cost of leaving is zero. A viewer scrolling a feed is not deciding whether to watch your video — they are deciding whether to keep scrolling, which is the default. Your opening has to interrupt a behaviour that is already in motion.

A hook is a promise, not a gimmick

The word "hook" has been flattened into meaning something loud at the start. A jump cut, a scream, a countdown. Those can work, but they are not what makes an opening hold.

A hook works when it sets up a gap the viewer wants closed. It tells them, quickly and specifically, what they are about to get and why it matters to them. Everything after that is you keeping the promise.

That is why loud openings so often fail. Noise creates attention but no expectation. The viewer looks up, finds nothing owed to them, and leaves — often a second or two later than they would have otherwise, which is why the graph shows a small bump before the cliff.

If your opening could be pasted onto someone else's video without changing anything, it is not a hook. It is a preamble.

Four openings that consistently hold

1. Lead with the result, then rewind

Show the outcome first — the finished room, the number, the before and after — then go back to how it happened. The viewer now has a reason to sit through the process, because they have seen what the process produces. This works especially well for anything transformational: renovations, fitness, design, business results.

2. State the disagreement

Open by contradicting something your audience already believes. "You do not need more content, you need fewer, better posts." The tension does the work. The viewer stays to find out whether you can defend the claim, or to disagree with you — both of which count as watching.

One caveat: the claim has to be one you actually go on to support. A contradiction you abandon reads as bait, and viewers remember it.

3. Name the specific person

"If you run a service business and your calendar is half empty" filters hard and holds harder. You lose the people who were never going to convert, and you keep the ones who feel addressed. Broad openings feel safe and perform badly, because nobody in particular is being spoken to.

4. Drop them mid-action

Start inside the moment rather than before it. Not "today we are going to install this," but the installation already underway, hands already moving. Context can arrive ten seconds later, once the viewer has committed. Most footage has a better starting point about thirty seconds after the one that was shot.

The openings that quietly kill videos

• The animated logo. Every second of branding at the front is a second spent on someone who has not decided to care about your brand yet. Put it at the end, where it lands on people who stayed.
• "Hey guys, welcome back." This is a greeting for an audience that already subscribed. To everyone else it is dead air. Existing fans will forgive its absence; new viewers will not forgive its presence.
• Context before conflict. Long setup — who you are, what this series is, what happened last week — delivered before the viewer has any reason to want it. Earn attention first, then spend it on context.
• A hook that oversells. If the opening promises more than the video delivers, retention collapses in the middle instead of the start, and the comments turn hostile. This is worse than a weak hook, not better.
• Silence. A second of quiet at the top reads as a loading screen. Audio should be present on the very first frame.

How to find the real hook in footage you already have

Most of the time the hook is already in the raw file — it is just not at the front. A practical pass:

• Watch the raw footage and mark every moment where you personally leaned in. Those are candidates.
• Ask what the single most surprising sentence in the whole recording is. Consider opening on it, even if it appears at minute nine.
• Cut the first thirty seconds entirely and watch what is left. Very often nothing of value was lost, and the video now starts at the right place.
• Watch your own opening with the sound off. If it is unreadable without audio, it will fail for the large share of viewers who watch muted.

Test one variable at a time

If you are changing openings, change only the opening. Same topic, same body, same thumbnail. Two videos with different hooks and different everything else tell you nothing, and most creators draw confident conclusions from exactly that kind of comparison.

Give it a fair sample. A single video is noise. Three or four with the same hook pattern start to be a signal — and once you find a pattern that fits your audience, it usually keeps working far longer than people expect.

The uncomfortable part

Strong openings mean cutting material you worked hard on. The best take is often not the one that should go first. The considered introduction you recorded three times usually belongs in the bin.

That is the actual skill: being willing to throw away good footage because it is in the wrong place. The video does not owe anything to the order in which it was shot.

---

We cut retention-first edits for creators and brands. If you want a second opinion on where your videos are losing people, book a call — we will look at your last three uploads and tell you what we would change.