Most Reels that underperform are not bad videos. They are decent videos with a weak opening. The viewer decides whether to stay before your first sentence finishes, so the edit you spent an hour on, the b-roll, the payoff at the end — none of it gets seen.
The first second is not decoration. It is a filter, and you can build it on purpose instead of hoping the rest of the video is good enough to carry it.
What the first second is competing against
A Reel does not open in a vacuum. It opens in a feed where the previous video has just ended or already been abandoned, the thumb is already moving, and the sound may or may not be on. Your opening is competing with momentum that started before you got there.
That framing changes what the hook has to do. It does not have to be clever. It has to interrupt. Those are different jobs, and clever openings often fail at interrupting because they need a second or two of setup before they make any sense.
Nobody outside Instagram knows how early retention is weighted in distribution, and anyone quoting you an exact number is guessing. But you do not need the weighting to act on it. If people leave in the first second, the video cannot accumulate anything else — no watch time, no shares, no saves, no profile visits. Early retention gates everything downstream no matter how the ranking works.
The four levers a hook can pull
Almost every opening that works is doing one or more of four things. Naming them turns the hook into a design decision rather than a vibe.
- Visual. Something on screen is unusual, mid-action, or visibly about to change. Movement beats a static talking head. A hand already reaching for something beats a person standing still deciding where to start.
- Verbal. The first words carry information or tension. "This is the part everyone gets wrong" works because it promises a specific correction. "Today I want to talk about" promises nothing.
- Textual. A line on screen that is readable at a glance, for the large share of viewers watching muted. Short, high contrast, and positioned away from the interface.
- Contextual. The viewer recognises the situation instantly — a familiar setting, a familiar object, a familiar frustration. Recognition is fast. Explanation is slow.
Weak openings usually pull zero levers: a static frame, a slow greeting, a logo animation, or a sentence that only makes sense once you have heard the next one. Strong openings pull two or three at once. You say something specific while something moves, and a short line of text carries the same claim for people with the sound off.
A checklist for your opening frame
Run the first second of your next video against this list before you publish. It takes about thirty seconds and catches most of the damage.
- Pause on frame one. Is anything in it interesting without sound and without the caption? If not, you have no visual and no textual lever.
- Read the first spoken words out loud. If they are "hey guys", "welcome back", or "so I wanted to talk about", cut them and start on the next sentence.
- Check where your on-screen text sits. The interface covers the bottom of the frame and part of the right edge. Text placed there is text nobody reads.
- Make sure the subject is identifiable in the first second. Not the conclusion — the subject. A viewer who cannot tell what the video is about has no reason to wait for it.
- Confirm the video actually delivers what the opening implies. An opening that overpromises buys a second and costs a follower.
- Watch it once on your own phone, at arm's length, with the sound off. That is the real viewing condition, not your editing timeline.
A worked example
Take a video about pricing for a small service business. The original opening is the creator sitting still, saying: "Hey everyone, welcome back. Today I want to talk a little bit about how I approach pricing my work."
That pulls nothing. The frame is static, the words carry no information, there is no text, and the actual subject does not arrive until around second four.
Rewritten: the creator is already writing a number on a sheet of paper, and the first words are "I raised my rate forty percent and lost one client." On-screen text reads "raised my rate 40%". Same video, same content, same person, same length.
Look at what changed. There is movement in frame one. The first words contain a concrete claim and a mild tension — a big increase, a small loss. The text repeats the claim for muted viewers. The subject is unmistakable within a second. Nothing was invented; the interesting true thing was simply moved to the front.
That is what hook writing almost always turns out to be. Not inventing a hook, but finding the most interesting true detail in the video and refusing to bury it under an introduction.
The hook has to match the video you actually made
A hook is a promise, and the rest of the video pays it. Mismatches show up as a decent first second followed by a cliff at second three, which is a specific and fixable pattern.
If your video is a demonstration, open on the demonstration, not on you describing that a demonstration is coming. If it is an opinion, open on the sharpest version of the opinion. If it is a story, open in the middle of the incident, not at the background. In each case you are cutting the runway and starting at take-off.
One caveat: do not front-load so completely that the video is over in the first second. A hook opens a loop. It should not close it.
Testing openings instead of guessing
You cannot judge your own opening reliably, because you already know what the video is about. Use the data instead.
In Insights, compare average watch time against video length across your recent Reels. When average watch time is a small fraction of an already short runtime, you have a first-second problem, not a pacing problem. Where a retention curve is available, read its shape: a cliff at the very start is a hook failure, a steady slide through the middle is a structure failure. People routinely try to fix one with the other and conclude that nothing works.
To isolate the variable, take a video that performed acceptably, keep the body identical, and re-cut only the opening. Post it later as its own Reel. It is not a controlled experiment — distribution varies for reasons you cannot observe — but do it across several pairs and you get a usable read on which openings hold people.
Keep a running note of the openings that held. After a couple of months you will find your account has two or three formats that reliably work, and they are usually not the ones you assumed.
Tactics that buy the second and cost you the account
- Fake interruptions like "wait, don't scroll" train people to distrust every opening you write.
- Bait the video never resolves is the quickest way to lose someone who had already found you.
- Loud noise unrelated to the content earns a stop and an immediate exit, which is worse for you than not being watched at all.
- Borrowed hooks from a completely different niche pull in an audience that will never buy, follow, or watch anything else you post.
The distinction is simple. A good hook is a promise you keep quickly. Anything that gets the stop without the promise is borrowing against your own account.
Rewriting openings is the cheapest high-leverage editing you can do, and it applies to everything you have already made. Start with your last ten Reels, find the three with the lowest average watch time that you still believe in, and re-cut the first second of each.
