Why retention dies at second 3 (and your hook isn't the problem)
The cliff in your retention curve is almost never the hook. It's the second beat. Here's what I see when I look at the videos that hold.
Open your worst short. Pull up the retention curve. I'll wait.
If yours looks like mine used to, there's a steep cliff between second 2 and second 4. Then a slower bleed. Then a long tail of the maybe-30% of viewers who were always going to finish anyway.
Every coach, every newsletter, every breakdown thread tells you the cliff is the hook. Fix your hook. Punch it up. Bigger claim. Faster cut.
That's wrong. Or at least it's been wrong for most of the videos I've looked at this year.
The hook brought them in
If your retention starts at 78% at second 1, the hook worked. They watched the first second. They didn't swipe before the video loaded. The thing that pulled them in did its job.
The cliff at second 3 isn't them rejecting the hook. It's them realizing the hook was the whole video.
The second beat is what's missing
In the shorts that hold past second 5, something changes between second 2 and 4. A new angle. A new claim. The reveal of what the first claim actually meant. A second visual.
In the shorts that crash, the hook is delivered and then the creator just... starts explaining the hook. Which is exactly what the viewer's brain was already doing in their head. So they leave. You stopped being entertaining and started being predictable.
This is a writing problem, not an editing problem. No amount of zoom-punches and whip cuts will save a second beat that's just an unpacking of the first.
A video I bombed last month
Hook was strong. Retention at second 2: 81%. By second 4: 44%.
The transcript of seconds 2 through 4 was — I'm not making this up — "and the reason this matters is because..."
Of course they left. I told them the reason was coming. Then I told them again. I never actually said the reason until second 7. By then there were 30 of us in the room.
The fix wasn't a new hook. The fix was deleting seconds 2 through 6 and going straight from the hook to the payoff. Same script, three seconds shorter, retention at second 4 climbed to 71%.
I don't have a clean rule for this except: when you're writing the second beat, the thing you wrote is probably wrong, and the thing that should be there is the thing you were saving for second 8.
The boring meta point
Retention is a writing structure problem. Editing covers for it some of the time. The reason hooks get all the airtime in the creator-advice world is that hooks are easier to talk about than structure — you can A/B test a hook, you can't really A/B test a beat map without making two completely different videos.
But the shorts that hit big almost never have just-a-good-hook. They have a hook that opens a question the viewer can't answer in their head, then a second beat that takes a different angle at the same question. The second beat is where retention actually gets made.
Stop fixing your hooks. Start cutting your second beats in half.