The three promises
Retention · VidCC Research · 19 Oct, 2026
A click is a contract. The title and thumbnail tell a viewer how the video will feel, what it will be about and how it will unfold. Most drops that look mysterious turn out to be one of those three promises being broken.
Introduction
Retention advice often comes as tricks: a hook that begs viewers to stay, a countdown, a tease of something at the end. Tricks can move a number for a video or two. They don't explain why viewers leave.
A more useful way to see retention is as three jobs. Interest is why a viewer wants to keep watching. Engagement is what keeps them watching without effort. Satisfaction is what makes them come back for the next upload. This piece is about the first, and the promises interest is built on.
Key takeaways
- Every video makes three promises at the click: an emotional promise (how it will feel), an interest promise (what it will cover) and a structural promise (how it will unfold).
- Breaking any of them costs viewers. The interest promise is the most fragile: a scene that drifts away from the promised topic is where people quietly leave.
- The opening is where viewers check the promises first, which is why it carries so much of a video's loss. Nearly half of everything a typical video loses is gone within the first tenth of its runtime.
- Changing the title can change the structural promise, and with it the kind of video you are allowed to make.
Three promises, made in a second
A viewer reads a title and glances at a thumbnail in a moment, but in that moment they form a clear picture of what they're about to get. They expect a tone, a topic and a shape. The video then has to confirm all three, continuously.
| Promise | What the click sets up | How it usually breaks |
|---|---|---|
| Emotional | How the video will feel: funny, tense, calm, eye-opening | A long stretch in a different mood, or an explainer that turns melodramatic |
| Interest | The exact topic, and the scenes viewers expect to see | Side stories, tangents and background that aren't about the promised topic |
| Structural | How the video moves: a run of points, or one continuous story | A list forced into a narrative, or a story padded with list-like detours |
The emotional promise: hold the tone
If the packaging promises tension, the video should stay tense for the great majority of its runtime, roughly 80 to 90%. A short change of mood can be a welcome release. A long one feels like a different video, and viewers who came for the first one leave.
Educational videos make an emotional promise too. It is the feeling of discovery: the surprise of a fact that contradicts what you believed, or the moment something finally makes sense. That promise is kept with a light, direct tone, not with drama.
The interest promise: stay on the topic
The interest promise is the list of things a viewer expects to see, and it is the one most often broken. A creator adds context, a personal aside or a sponsor message, and for those seconds the video is about something the viewer didn't click for.
A side story can work, but only once the viewer is already invested in an open question. Without that, it stalls the video. Long breaks such as sponsor reads hurt less when they are woven into the main story instead of dropped in as a block.
The structural promise: list or journey
Every video is driven either by topics or by an experience. A topic-driven video delivers points: tips, facts, ranked items. An experience-driven video carries viewers through one connected story. The title tells viewers which one to expect.
"7 ways to fix a slow opening" promises a list, and turning it into a personal story will bore viewers who wanted the seven ways. "How I fixed my slow openings" promises a journey, and lets you tell the same material as a story. Same content, different promise, so choose the title that matches the video you can actually make.
Where a broken promise shows up
Viewers check the promises hardest at the start. In the 158 retention graphs we analysed, the median long-form video lost 48% of all the viewers it would lose within the first tenth of its runtime (95% CI 45–53%). Each later tenth cost only two to four points.
Later in the video, a broken promise rarely shows up at the exact second it happens. Viewers tend to give a segment the benefit of the doubt, then leave as it ends. When you see a drop, look back at the whole segment before it and ask which promise it stopped keeping.
What to try on your next video
Write the three promises down first
Before scripting, write one line each for the feeling, the topic and the shape your title and thumbnail set up.
Cut what isn't on the interest promise
Go through the edit and mark every stretch that isn't about the promised topic. Cut it, shorten it, or tie it to an open question.
Fit the title to the video you made
If the edit became a story, give it a journey title. If it became a list, promise the list.
About this research
VidCC Research keeps a private library of 2,151 findings on retention, storytelling, sound, packaging and analytics. Each one is tied to the source it came from and tagged by topic, and the library informs how we build and test VidCC's analysis. This article explains one part of it in our own words.
Where we cite numbers, they come from our analysis of 158 publicly posted YouTube Studio retention graphs. Studies that rely on them include the full methodology.