A hook generator writes plausible openings. It cannot tell you which family they belong to
Nine families of opening across 2,499 Reels, with clear rates from 22.7 percent down to 15.7 percent. What to do with a generated hook before you film it. 2026-10-10.
What a generator is doing
You give it a topic; it returns openings that read like openings. That is genuinely useful against a blank page, and the output is usually fluent. What it is not doing is checking its suggestion against anything that happened.
So the question is not whether to use one. It is what to do with the ten lines it gives you, because they will not be equally good and nothing in the tool knows which.
The families, and how differently they perform
Our grouping of Scrptly's free-text opening labels, first match wins; 472 Reels matched none and are excluded. Treat the top three as tied on median views: the clear-rate column separates them more reliably.
| Opening family | Reels | Median views | Beats 3x |
|---|---|---|---|
| Curiosity gap | 484 | 175,847 | 20.5% |
| Pattern interrupt | 177 | 169,411 | 18.9% |
| Visual first | 495 | 163,257 | 22.7% |
| Question | 101 | 151,934 | 17.8% |
| Number or list | 105 | 95,972 | 15.7% |
| Bold claim | 210 | 95,620 | 18.3% |
| Story or confession | 189 | 88,459 | 18.6% |
| Direct address | 37 | 65,076 | 21.6% |
| Problem solution | 229 | 60,012 | 18.9% |
The spread that matters is the right-hand column. Visual-first openings clear the outlier bar 22.7 percent of the time; number and list openings manage 15.7 percent. A generator asked for hooks will cheerfully hand you five of the second kind, because they are the most recognisable thing a hook looks like in writing.
The thing generated hooks cannot do
A visual-first opening is a camera decision, not a sentence. 495 Reels in our sample open that way and they have the best clear rate of any family. You cannot prompt your way to one, because there is no line to write: the frame does the work and the words come second.
That is the structural blind spot. A text generator will never suggest "open on the wardrobe from the inside, say nothing for two seconds", because its output space is sentences.
A workflow that uses the generator for what it is good at
- Generate ten. Volume is the thing it is genuinely good at.
- Label each one by family using the table above. This takes a minute and is the whole value.
- Delete every number and list opening unless the number is the actual point of the video.
- For the two or three you keep, ask what the first frame shows while the line is spoken. If the answer is "my face, talking", you are leaving the strongest family unused.
- Check length. Whatever the hook, a reel under 15 seconds is starting in the weakest band in the sample.
Figures on this page come from 2,499 Instagram Reels across 526 accounts, analysed by Scrptly and deduplicated so no reel is counted twice. Read 2026-10-10.
See the openings real creators in your niche used, grouped and measured, instead of generated guesses.
Analyze a creator freePut your own idea through the same checks before you film it.
Create content