Hook Rate vs Watch Time: What to Optimize First on Shorts
Picture a creator, call her Nia, on a Tuesday morning with a cold mug and a Short about a kitchen timer. The view count is not zero. The comments are. She cannot tell whether people left in the first moment or stayed, got the point, and had nothing to say. Those are different problems, and they ask for different edits.
Hook rate and watch time answer two questions. The first is whether a viewer chose to stay once the clip was already playing in a feed. The second is how much of the clip they actually watched after that choice. Fixing the wrong one wastes an afternoon. You rewrite a script that was fine, or you polish a middle that almost nobody reached.
What YouTube actually names
On YouTube, the Shorts content tab in Studio includes a metric called "How many chose to view." YouTube's help text says it highlights the percentage of times that viewers viewed your Shorts versus swiped away. That is the closest official label to what creators casually call a hook rate. It is a ratio of people who stayed against people who swiped, not a grade, and YouTube does not publish a universal "good" percentage for it. Compare a Short with your own recent Shorts in the same format. A number borrowed from a stranger's screenshot is not a standard.
Separate from that opening choice, YouTube describes how viewers engaged as the percentage of times viewers stayed to watch past the initial seconds of a Short. Among people who stayed, Studio also reports average view duration and average percentage viewed. YouTube's help page defines average percentage viewed as the average percentage of a video watched among those who stayed to watch, calculated from engaged views and their watch time. Average view duration is the average time watched among those who stayed. Read the definitions on YouTube's Shorts content-tab help and on YouTube's content-performance help before you argue with a graph.
Other apps use their own labels for an early exit and for time watched. The names will not match YouTube's. Treat each app as its own notebook. Do not average a TikTok watch-time figure with a YouTube "chose to view" percentage and call the blend a strategy.
When the opening is the problem
If people swipe away and almost nobody reaches the useful line, the middle of the script is not the bottleneck yet. Look at the first picture and the first spoken sentence.
The first picture
The first picture should already be the subject. A logo sting, a slow zoom onto a ceiling, or a hand reaching for a phone burns the moment when the feed is deciding. Motion helps only when the motion is the point: a timer hitting zero, a book closing on a sentence, a product failing in a way the viewer recognizes. Random camera shake is not a hook. It is noise.
The first sentence should carry the job of the clip. Name that job before you edit: teach one move, tease one consequence, prove one claim you can show, or land one joke. If you cannot say the job in a single line, the opening will wander while the viewer leaves. Write the line on the script, then film or generate toward it.
Here is an illustrative example. A novelist opens on the cover and says the title. A stronger opening for the same book is the choice the character is about to make, said in ordinary words, with the title held until the viewer knows why it matters. A coach opens with "Hey guys, quick tip." A stronger opening is the myth the client walked in believing, stated as a sentence the viewer might have said that morning. A shop opens with a logo. A stronger opening is the object in use, already failing or already fixed, so the viewer knows which problem the next ten seconds will answer.
When watch time is the problem
A drop after the opening
Sometimes people stay and then leave at a specific beat. That pattern means the opening did its job. Rewriting the hook again will not repair a middle that stalls.
Play the clip with your finger on the moment you would swipe if you did not make it. Common leaks, described without pretending they are universal laws:
- A second idea arrives before the first one pays off. One clip, one job.
- An apology or a throat-clear: "this might be obvious, but." Cut it. The viewer already chose to stay.
- B-roll that does not show the thing you just said. A sunset over a sentence about a receipt does not illustrate the receipt.
- A caption that repeats the whole script in a block, so the eye is reading a paragraph while the voice is on sentence two.
Average percentage viewed tells you how far stayers got. It does not tell you which word lost them. For that you still have to watch the cut. If your tool shows a retention curve, mark the drop with a word from the script, not with a mood. "They left when I listed three benefits" is a note you can edit. "The energy dipped" is not.
Change one variable, then look again
A useful test keeps the body and swaps the opening, or keeps the opening and tightens one beat in the middle. Changing the hook, the music, the caption color, and the ending in the same upload gives you a new clip and no lesson.
Write the variable in a line you can find next week: "Same script, first frame is the broken latch instead of my face." When you review, you are asking whether that frame changed the chose-to-view share, or whether the average percentage viewed moved. If both moved, you still only learn a little, but you at least know which number you meant to touch. The weekly habit for that review lives in Analytics Review Cadence for Short-Form Channels.
Do not chase follower count as the score for this test. Follows are a later decision. A clip can hold attention and still not earn a follow if you never said who the channel is for. That is a profile problem, not a hook problem.
Sound, silence, and the same opening
Many people meet a clip with the sound off, on a train or in a meeting, and only unmute if the picture earns it. The caption's first line has to carry the same claim as the spoken hook, in fewer words. If the voice says the consequence and the caption says the title, silent viewers and hearing viewers are watching two different videos. Caption Styles That Keep Silent Viewers Watching is the practical pass for type size, line breaks, and emphasis.
If you make faceless clips, the opening still has to be a picture of the subject, not a picture of the tool. Faceless Shorts is the workspace path for drafting that kind of cut and reviewing it before you export. The review is the point. A generated opening line is a draft until you watch it the way Nia should have watched hers: once for the swipe, once for the stay.
A pass you can finish before lunch
Pull the three Shorts you posted most recently. For each one, write two words: opening clear or opening muddy, body held or body leaked. This week, rewrite only the muddy openings. Leave the held bodies alone. Next week, if the chose-to-view share on the new versions looks more like your better clips, then look at a leaked body and cut one beat. That order keeps you from polishing sentences nobody hears.
You might be thinking you need a bigger sample before you touch anything. You do need more than one clip, because a single post can be unlucky. You do not need a month of paralysis. Three clips in the same format are enough to see whether your openings are consistently muddy.
FAQ
Should I optimize hook rate every time? No. If viewers stay and then drop in the middle, another hook will not teach you anything. Fix the beat where they leave.
Is a high chose-to-view share enough? It means the opening earned a chance. Average percentage viewed tells you whether the rest of the clip used that chance. Comments and saves tell you whether the idea was worth keeping. None of those numbers promises distribution.
Do I post the test and the control on the same day? Give them similar conditions if you can, and do not change the posting time, the caption style, and the topic in the same breath. One variable.
The next clip you open is already telling you which edit to make, if you separate the people who never stayed from the people who stayed and then left.