Your click-through rate is fine. Five percent, maybe better. People are choosing your video out of a feed, which is the hard part, and everyone tells you the hard part is the thumbnail. So you did that part right.
Then the retention graph falls off a cliff before the thirty-second mark, and you are back to guessing.
The diagnosis you will get, every time, is the most repeated line on r/NewTubers. One creator posting exactly these numbers got it word for word: your title and thumbnail are pretty good with a 4.5% CTR, but once they see the video, for some reason, you're not delivering on the promise of the thumbnail and title, so they leave. Asked what the worst error a new creator can make is, the top answer in that thread was the same thing — not delivering on the promise your title and thumbnail are making, and then delivering it thirty seconds' worth of substance after five minutes of talking about random stuff.
The advice is correct. It is also impossible to act on, because it never tells you which promise broke, or what second it was supposed to land on. This article is about answering those two questions on your own video, before you upload it.
What is a packaging promise, exactly?
It is every expectation a viewer is holding at the moment the video starts playing. Not what you meant. What they are now waiting for.
Most creators can name one — the topic. But a package makes four kinds of promise at once, and three of them are invisible to the person who wrote it:
- Stated. Whatever the title literally says. "Bank fees." "I tried it for 30 days." This is the one everybody sees.
- Counted. Any number in the package. "The 3 fees" is a contract, and viewers keep score against it whether you remembered to or not. Deliver two and the top comment writes itself.
- Implied. The frame your words create. "(A Retired Manager Explains)" promises insider authority and a mechanism actually explained — not a claim asserted by a narrator. "Never see" promises the thing is currently hidden and about to be revealed to you.
- Visual. Whatever the thumbnail points at. A red circle on a document promises you will be shown that line. An arrow promises a destination. This is usually the heaviest promise of all, because the image is what earned the click, and it is the one creators never write down anywhere.
A typical video makes five to seven of these. The creator is tracking one. The gap between those two numbers is where retention goes.
Why can't you see the gap in your own video?
Because you already know what is in it. You cannot un-know the payoff, so you cannot feel its absence.
The workflow makes it worse. The script gets written on Tuesday, when you are thinking about the subject. The title and thumbnail get written on Friday, when you are thinking about the feed and what will make a stranger's thumb stop. Two different documents, two different headspaces, days apart, and at no point do the two get laid side by side.
So when you rewatch your own opening, the four minutes of setup do not feel like four minutes of nothing. They feel like context, because you know what they are setting up. The viewer does not. They are holding a promise and a stopwatch, and everything that is not the payoff reads as delay.
This is why other people can spot it on your channel in seconds and you cannot spot it on your own after five viewings. It is not a skill difference. It is an information difference, and the fix is mechanical: write the promises down, then go find them in the script.
How fast does the payoff have to arrive?
The heaviest promise has to be paid off inside the first thirty seconds. Not addressed. Paid.
The reason is arithmetic rather than taste. YouTube's judgement of a video is formed on the first tranche of impressions, and the signal that matters most in that window is whether people who clicked stayed. As one creator put it in a thread on this exact problem: every millisecond you're not delivering on the promise of your title and thumbnail, you're losing viewers.
What counts as delivery is stricter than people expect. Saying "today we're going to talk about three fees" is not delivery — it is a repetition of the title with a delay attached. Delivery is naming the first fee, or showing the circled line, or saying the sentence that only an insider could say. It is the thing itself, arriving.
There is a second, quieter failure the timing rule catches: the promise that is delivered brilliantly, at 9:40. That is not a broken promise. It is a wasted one. The strongest sentence in the video ran to the eighteen percent of the audience who were still there, and it was the sentence that would have kept the other eighty-two percent.
What do you do when the promise really is broken?
There are exactly two moves, and picking the wrong one is what makes creators think the fix requires a reshoot.
- Move the payoff forward. Cut the intro, the subscribe ask and the throat-clearing history, and put the thing they clicked for at 0:00. Most videos have a beat that already works as an opener sitting somewhere between minute four and minute ten. This is a copy-paste, not a rewrite.
- Change the promise. If the video genuinely does not contain what the package promised, do not add it under duress. Repackage around what the video does contain, and pick the strongest thing in it.
That second move is where the surprise usually is. Nearly every script has a buried lede — a concrete, actionable, faintly secret line sitting at minute twelve that nobody clicked for because it was never advertised. "If you call and ask, most banks will refund that fee going back six months. They have the authority. They just don't offer it." That is a stronger package than the one that was built, and the footage already exists.
Both routes end in the same place: a package the video actually delivers. One moves the video toward the promise. The other moves the promise toward the video.
What does the real output look like?
Here is the actual output from the sample run — a faceless senior-money channel, one 14-minute video, its title, its thumbnail text and its script:
# PROMISE STATUS PAID AT 1 three fees, counted - all 3 BROKEN - 2 "never see" - hidden from you PARTIAL 4:11 3 a retired manager, on the inside LATE 9:40 4 "THEY HID THIS FEE", circled BROKEN - 5 "explains" - how the fee works PAID 4:35 5 promises made, 2 paid off. First payoff at 4:11.
The title promised three fees. The script contains two. Nobody had counted.
0:00-0:14 channel intro + subscribe pays off nothing 0:14-4:11 banking history, caution pays off nothing 4:11 fee #1 named first payoff A viewer clicked because three fees were hidden from them. For 4 minutes the video does not mention a fee.
The drop-off, located to the second — and fixable by moving one sentence, not by reshooting.
It also found the buried lede: a line at minute twelve saying most banks will refund six months of that fee if you simply call and ask. Never promised anywhere in the package, and a better title than the one the video shipped with.
How do you run it yourself?
You paste one prompt into Claude Code and it builds the tool for you. It is a dark dashboard, pre-filled with the sample above, so it works on the first run.
It has a Settings panel for your own API key, so you can run it on every video before you upload — and on the last five you already published, which is usually where the first surprise is.
Grab it below. Drop your email and the prompt is on the very next page, free. Paste it in, then run it on your own title, thumbnail and script.
Can you turn this into a side hustle?
Yes — and it is one of the simplest ways to make money with AI. You do not have to use this tool only for your own work. You can run it for other people and charge for it.
Here is the model. Local businesses need Paste your title, your thumbnail text and your script. Get every promise your packaging makes, whether the script actually pays it off, the exact second it lands, and the rewrite that closes the gap — before you upload., but they do not have the time or the skill to do it well. You do. So you run the tool, hand them a finished result, and charge for the service. Many people charge $500 a month per client for work like this.
The best part is the cost to start: a free prompt — it pays for itself on the first job. The tool does the heavy lifting in minutes, so your margin is high and you can take on more clients without more hours. To get your first client, reach out to a few local businesses you already know. Do one for free, show them the result, and ask who else needs it.
FAQ
Do I need my analytics for this?
No. It reads the package against the script, so you can run it before the video is published and before there is any data to read. That is the point — retention data tells you people left, and only after it is too late to matter.
Does it judge my writing or my topic?
No. It only judges whether the script delivers what the package promised, and when. A well-written video can score badly here, and that is the useful case: the video is fine, the mismatch is the problem.
How does it know what my thumbnail promises?
You describe it in one line — the words on it and what the image points at. That description takes ten seconds and it is the part creators never write down, which is why the heaviest promise in most packages goes unaudited.
What if it says my promise is broken but I can't reshoot?
Then you take the other route. It writes an alternative package built around the strongest thing already in the script, including the timestamp where the existing footage pays it off. Same video, a promise it can keep.
Can I reuse it on every upload?
That is the point. Enter your API key once and run it on each video before it goes out. It is a reusable app, not a one-time output.