The proven structure for a YouTube video: hook in the first 15 seconds, setup by second 45, then three to five payoff blocks with a rehook every 60 to 90 seconds, and finally a smooth transition into the end screen. This structure keeps the retention curve flat, and watch time is the strongest signal you can send the algorithm.
Read the retention curve before you plan
YouTube is the only major video platform that shows you to the second when viewers leave. You find the curve in YouTube Analytics under “Engagement.” Watch for three patterns:
- The crash at the start. Losing 20 to 30 percent in the first 30 seconds is normal. More means: the hook does not deliver on the promise made by title and thumbnail.
- Stair steps in the middle. Each step marks a spot where you cashed in a promise without making a new one. That is exactly where rehooks belong.
- The drop at the end. It shows when viewers mentally check out, usually well before the outro. Almost nobody sees anything after that point.
Look at the curves of your last three videos before planning the next one. Then you know where your structure tears and can work on exactly that spot. Use the curve of your best video as the reference. Watch time counts for more on the platform than clicks: a video many people watch to the end gets more impressions.
The dramaturgy: four phases, concrete times
Promises like “the perfect video in 3 steps” fall short. The proven structure has four phases, and each has a time mark. All four can go straight into your next video.
Phase 1: Hook (0:00 to 0:15). The opening confirms the promise from title and thumbnail and plants an open question. No greeting, no jingle, and you do not need to introduce yourself either. The first 5 seconds decide whether someone clicks away. Seven proven formulas are in the guide to YouTube hook formulas.
Phase 2: Setup (0:15 to 0:45). Give exactly the context the rest needs: the starting point, the rules of the experiment, what the viewer should be able to do at the end. Keep it tight; spend two minutes explaining here and half the audience is gone before the content begins.
Phase 3: Payoff chain (from 0:45). Break your topic into three to five blocks and arrange them so each block cashes in a partial promise and sparks curiosity for the next. Every block needs its own value. Example of a transition: “That gets the camera running. But the picture is only half the battle; the audio decides whether anyone stays.”
Phase 4: Rehook every 60 to 90 seconds. A rehook is a mini-hook in the middle of the video: a teaser (“in a second you’ll see what went wrong”), a change of perspective, or a new visual meant to spark curiosity. In a 10-minute video, that is six to eight deliberately placed spots. Tip: write the rehook lines first; many YouTubers leave them to the edit and waste exactly these spots.
The end-screen transition (last 20 seconds). “Thanks for watching” is the click-away signal for many viewers. Deliver the last payoff instead and announce the next video in the same breath, while the end screen is showing: “The structure stands. How to write the title for it, I’ll show you here.” Remember: the end screen overlays your picture, so plan the last 20 seconds without important on-screen elements.
Calls to action: few, in the right place
More than two calls to action per video dilute both. A “don’t forget to subscribe” without a reason delivers next to nothing measurable; tie the ask to a concrete benefit (“I upload a new tutorial every Tuesday”). A comment question works better at a spot with high retention than at the end, for example right after the strongest payoff. That way your community gets a question it actually still sees. If your channel is part of a larger marketing chain, say with a newsletter or social media as the next step, the same rule applies: one goal per video, clearly announced.
Chapters: structure visible in the player
You create chapters with timestamps in the video description. Mind three rules: the first line starts with “0:00,” you need at least three chapters, and each must be at least 10 seconds long. Then the chapters appear as sections in the player and as jump marks in Google Search. Name each chapter by its benefit (“Set up audio in 2 minutes” instead of “Part 2”) so viewers can click straight to the spot they need. A link to further material goes directly under the timestamps in the description. Side effect: if you stay cleanly structured while speaking, you get a usable automatic transcript, which helps discoverability in search.
Tutorial, storytime, listicle: three formats, three orders
The skeleton stays the same; what changes with the format is the order of the payoffs.
Tutorial and explainer. Result first: show in the first 20 seconds what comes out at the end, then the steps in working order. Chapters are mandatory here, because many viewers rewatch the video later and want to jump straight to one step.
Storytime. Break the chronology. Enter at the peak of tension, then jump back to the beginning and tell toward that moment. Rehooks come from hints (“what I didn’t know yet”). The most common mistake: starting chronologically at day 1 and hoping someone sticks around until day 30.
Listicle. Sort the points by tension, not by logic. Open strong, put the best point second to last, and tease it in the hook (“number 4 surprised me the most”). Number them visibly: viewers like progress indicators because they show how much is left.
Your channel probably mixes several formats. That is fine, as long as each video stays consistent within its format: videos that jump between tutorial and storytime produce the worst curves in practice.
Length and aspect ratio: the structure sets both
The most common planning question: how long should the video be? The answer is in your block list. Three payoff blocks of 90 seconds each plus hook, setup, and end-screen transition make a video of about 6 minutes; five blocks make 9 to 10 minutes. Do not stretch a 6-minute topic to 12 minutes; the curve visibly punishes every filler minute. Technically, classic videos upload in 16:9 landscape, so 1920 × 1080 pixels or more; vertical belongs to Shorts, and those follow their own dramaturgy with a hook in the first second.
The structure template: build once, use often
Building a successful YouTube channel is about repeatability. Build one template per format, something you can create in an afternoon and then just fill in: hook idea, setup bullet points, three to five payoff blocks with minute marks, rehook lines, end-screen announcement. If you want to upload regularly, this saves one to two hours of planning per video and makes you steadily better, because your own content becomes comparable. That applies from the first video: even if you are just starting your channel and barely know your audience yet, you benefit, because even the first curves reveal something about your niche. A recognizable structure across many videos also builds loyal viewers: they know what they will get before they click.
This is what a filled-in template for a tutorial looks like: 0:00 hook (show the end result), 0:15 setup (starting point and materials), 0:45 block 1 (base settings), 2:15 block 2 (the mistake almost everyone makes), 4:00 block 3 (fine-tuning), 5:30 end-screen announcement. Six lines, and half the work is done before you word the first sentence.
From structure to video script
Structure comes before text. First plan the blocks with minute marks, then decide per block what to say and show, and only then write it out. A video script does not have to be a word-for-word manuscript: for tutorials, bullet points per block are enough; storytime benefits from written-out transitions. And because a perfect video gets no clicks without the right title: the formulas for that are in the guide on writing YouTube titles.
Building structure in the eloqole Studio
In the eloqole Studio, the structure comes before the text: hook variants and title in the same step, then the outline with payoff blocks, only then the script. The retention check scans the finished text for the spots where viewers typically bail, for example missing rehooks after minute two.