Retention analysis · before you record
Your script is leaking viewers at 0:23.
Paste it below. HookProof predicts your retention curve before you hit record, flags the exact lines where people bail, scores your hook against the first 3 seconds, and rewrites the leaks in your own voice.
0 / 15,000
Voice sample (optional)
Paste a transcript from a video that sounds like you. Hook variants and rewrites will match its cadence and vocabulary instead of defaulting to generic script-speak.
Drop-off flags
What it catches
Five patterns sink most retention curves. HookProof reads your draft for all of them and points at the line, not the minute.
0:00–0:03
Hooks that warm up instead of opening
Mobile viewers give a video about three seconds. A greeting spends all of them saying nothing.
cut
“Hey everyone, welcome back to the channel, hope you're all having a great week.”
keep
“This $40 mic beat my $400 one in a blind test. Here's the recording.”
0:20–0:40
The decision cliff
Over 40% of the viewers you lose decide in this window. If your setup is still running at 0:25, they leave during it.
cut
“Before we get into the comparison, some quick background on how I tested all of this.”
keep
“Test one. Same footage, both laptops, timer on screen.”
MID-ROLL
Lulls with nothing owed
A section that resolves cleanly is an exit ramp. An open loop carries people through the slow part because you still owe them the answer.
cut
“So that covers the lighting. Now let's move on to the audio setup.”
keep
“Lighting was the easy part. The audio decision I made next ruined the entire shoot.”
0:05–0:30
Intros that re-read the title
They clicked the title. Restating it for 25 seconds tells them the video still hasn't started.
cut
“In this video I'm going to be showing you exactly how I edit my videos twice as fast.”
keep
“Step one. Delete your timeline and rebuild it from the audio.”
LAST 0:20
Endings that apologize for ending
A trail-off takes your end screen down with it. The last line should hand the viewer the next video.
cut
“Anyway, that's pretty much it for this one, so yeah, if you enjoyed it maybe subscribe.”
keep
“That's the whole system. The video on screen shows how I script it in 20 minutes.”
Workflow
How it works
01
Paste the script
Drop in the whole draft, stage directions and all, and pick the format preset you're recording for.
02
Read the curve
The predicted retention curve marks each line where viewers are likely to bail, with the reason next to the flag.
03
Ship the fix
Take the rewrites that sound like you, patch the flagged lines, and record the stronger draft.
Plans
Free covers this week’s script
Creator covers a weekly upload schedule for $19 a month. Pro removes the meter entirely.
Free
$0
2 analyses a day, no signup. Enough to fix this week's script.
- 2 script analyses per day
- Retention curve with drop-off flags
- Hook score on every run
Creator
Recommended$19/mo
For weekly uploaders who script every video.
- 30 script analyses per month
- Hook score with 8 rewritten variants
- All format presets (long-form, Shorts, TikTok Shop)
Pro
$39/mo
For daily posters and multi-channel operators.
- Unlimited script analyses
- Everything in Creator
- Batch of 5 short-form scripts per day
FAQ
Fair questions
How accurate is the prediction?
It's a model trained on retention patterns, and it cannot see your thumbnail or your delivery. Treat it as an editor's gut-check you can run in 30 seconds before you commit three hours to recording, rather than a replacement for analytics. When it flags a line, read that line out loud. You'll usually hear why.
Do you train on my scripts?
No. Your script is sent to the model for the one analysis you asked for and never stored. In v1 there is no account and no database, so there is nowhere for it to live even if we wanted to keep it.
What formats does it handle?
Three presets: long-form YouTube, Shorts, and TikTok Shop, because a 9-minute tutorial and a 40-second hook video lose viewers in different places. Paste plain text, a teleprompter export, or bullet notes. If you can record from it, it can read it.
Why not use YouTube Analytics instead?
Use both. Analytics shows where viewers left after the video is public and the damage is done. HookProof reads the script while it's still a draft, when fixing a weak 0:25 costs one rewrite instead of one video.
Can I cancel?
Yes, any time. Subscriptions are monthly through Stripe and you keep access through the period you paid for. Right now the whole checkout runs in Stripe test mode, so no real card is ever charged.
What's the de-slop pass?
Rewrites come from a language model, and model writing has tells. The de-slop pass is a second step that strips them out, so the line you paste back into your script sounds like you wrote it on a good day. Pro runs it on every segment-level rewrite automatically.