Every video writes its
own chapters while I sleep.
A small bot watches every YouTube video I upload, splits it into themed chunks with timestamps, writes a curiosity-pulling title for each chapter, and patches the description through the YouTube Data API. Viewers navigate cleanly, watch time goes up, the algorithm rewards it, and the video earns more ad revenue for the rest of its life.
Every upload, a fully chaptered video without me touching the description.
Most creators hit publish on a YouTube video and walk away. They never go back to write proper chapters because the chapter step is fiddly: rewatch the video, mark the topic shifts, write a short title for each chunk, paste the timestamp list at the bottom of the description, fix the formatting. Twenty minutes per video, every single time. There is a different way. Let a small bot watch the video for you, find the real topic shifts, write the chapter titles in your voice, and patch the description through the YouTube API. Here is who it is for, what goes wrong without it, how it works, and what you get back.
Who it’s for
YouTubers, podcasters who also publish on YouTube, course creators, and anyone running long-form video over 10 minutes. Especially useful if you publish weekly and your videos are monetised, because every chapter that gets tapped lifts watch time and watch time is what the YouTube algorithm pays for. If you record long-form and your videos have no chapters, this is for you.
What goes wrong
Without the bot, the chapter step is the thing you skip when you are tired. The video goes live with one long unbroken timeline and a description that lists no timestamps. Viewers cannot find the part they want. They scrub or leave. Watch time stays flat. The algorithm has no reason to show the video to more people. A long video without chapters is a video the audience cannot navigate.
How the machine works
The trigger fires when a new video appears on your channel. The bot pulls the transcript with timestamps from the YouTube Data API (or a transcription source). An AI chapter detector reads the transcript, finds the real topic shifts, and groups the text into themed chunks. Each chunk gets a short curiosity-pulling title. The list of timestamps gets patched into the existing description without overwriting your CTAs. You publish. The bot chapters. The watch time lifts.
What you get back
A fully chaptered video on every upload with timestamps and short titles in your voice, patched into the description automatically. Watch time lifts roughly 5 to 10 percent per chaptered video, the algorithm rewards stickier sessions, and each video earns ~3 to ~7 extra euros of ad rev over its life. Across ~100 to 150 videos a year, that is ~500 to ~1,000 of added ad rev, compounding to ~8,000 to ~15,000+ over five years. A small extra cheque on every video, forever.
I was leaving the chapter step undone on every single video.
For the first eighteen months of my channel, almost none of my videos had chapters. I told myself I would go back and add them later when I had time. Later never came. The chapter work is fiddly: rewatch the video, find the topic shifts, write a short title for each chunk, paste the timestamp list at the bottom of the description, fix the line breaks. Around twenty minutes per video if I was focused. With two long-form videos a week, that was forty minutes I never did, and the videos went live with no navigation at all.
The cost was bigger than the time. Viewers landed on a 15-minute video, hit play, watched the first 30 seconds, then scrubbed around looking for the part that matched the title. Most of them gave up and left. Average watch time on those videos sat flat, the algorithm had no reason to push them out further, and any video that was almost going to break out instead just quietly died at the rate of my subscriber base.
So I built a small bot. The bot fires as soon as a new video appears on the channel. It pulls the transcript with timestamps from the YouTube Data API. It runs the transcript through a chapter detector that finds the real topic shifts (the moments where what I am talking about actually changes, not arbitrary 60-second blocks). It groups the text into themed chunks. For each chunk it writes a short curiosity-pulling title in my voice, under six words. Then it builds the chapter block in YouTube’s required format and patches it into the existing description through the API, careful not to overwrite the CTAs and the affiliate links I already had in there.
The total cost of running it is a few cents per video in API calls. The time I spend on the chapter step is zero. Every long-form video I upload now goes live with five to seven chapters, real titles, real timestamps, and a description that did not get mangled. Watch time on the new videos sits around 5 to 10 percent higher than it used to. The lift is small per video. Across the catalog, on every video, forever, the maths gets interesting.
Proof point: the channel this engine runs on is Freedom by Choice on YouTube, and the same mechanics are walked through in the Growth Hacking series. The chapter blocks at the bottom of those videos are what the bot patched in.
Three moves that lift watch time on every video without me touching the description
What made this work was treating chapters not as cosmetic polish but as a small lever that pulls on YouTube’s biggest ranking signal. Watch time is what the algorithm pays for. Chapters lift watch time because viewers can find the part they want and stay for it. The framework hangs on three moves: read the actual transcript so the chapter splits match real topic shifts, write the chapter titles for curiosity (not for accuracy), and patch the description without breaking what is already in it. That is the whole engine. Done right, every chaptered video earns a small extra slice of ad rev forever, the catalog compounds, and you never open YouTube Studio again to do the boring step.
Split on real topic shifts, not on a clock
Most chapter tools split videos every 60 or 90 seconds because they have no idea what is actually being said. That is the wrong split. The bot I run reads the transcript and looks for the real topic shifts: the moments where what I am talking about actually changes. The split lines up with the way the viewer experiences the video, so the chapter list reads as a real table of contents instead of a clock dump. The viewer scans the chapters, sees the part they wanted, taps. The tap is what lifts watch time.
Write chapter titles for curiosity, not for accuracy
A chapter title that says “Setup” is technically accurate and gets zero taps. A chapter title that says “The piece nobody sets up first” pulls the viewer in. The bot writes every chapter title under six words, in my voice, with a small curiosity hook baked in. Short enough that it fits in YouTube’s UI on mobile. Pulling enough that the viewer wants to tap. The chapter titles are the part the viewer actually reads. Treating them as micro-headlines is what turns the chapter list from a navigation tool into a watch-time engine.
Patch the description, never overwrite it
The last move is the one that breaks things if you get it wrong. My descriptions already hold CTAs, affiliate links, social handles, and a clean signature block. The bot does not rewrite the description. It locates the right insertion point (right after the first paragraph, before the existing CTAs), pastes the chapter block in YouTube’s required format, and leaves everything else exactly where it was. The chapters go live, the CTAs stay live, the affiliate links keep paying. One PATCH call to the YouTube Data API, no broken descriptions, no lost revenue links.
Once those three moves are in place, the chapter step stops being the thing you skip when you are tired. It becomes a small lift the algorithm gives back on every video you publish, on a schedule, with zero manual work.
Before the bot
- ~20 minutes of manual chapter work after every upload (always skipped)
- Videos shipped with no chapters, no description timestamps
- Viewers landed, scrubbed, gave up, left within the first minute
- Watch time stayed flat, the algorithm had no reason to push the videos out
- Description sometimes got mangled when I went back to add chapters later
After the bot
- Zero manual time per upload on chapters, the bot writes them all
- Every long-form video goes live with 5 to 7 chapters in my voice
- Chapter splits land on real topic shifts, not arbitrary 60-second blocks
- Watch time lifts ~5 to 10 percent per video, the algorithm rewards it
- Existing CTAs and affiliate links stay exactly where they were in the description
Prompt 1: watch this YouTube video and split it into themed chapters with timestamps
The bot does not “watch” the video the way a human does. It reads the transcript with timestamps and looks for the real topic shifts. Use this prompt to turn a raw timestamped transcript into a clean chapter split that lines up with how the viewer actually experiences the video.
Themed-chapter splitter
Act as a YouTube editor splitting a long-form video into themed chapters with timestamps. The split will be patched into the video description so viewers can navigate to the part they want. The goal is to land the splits on real topic shifts in the video, not on an arbitrary clock. Video title: [paste] Video length in minutes: [paste] Full transcript with timestamps in the format [HH:MM:SS] text: [paste] The single promise the title makes to the viewer: [paste in one sentence] Produce a clean chapter split. Return five to eight chapters. Each chapter must start exactly on a moment where the topic actually changes in the transcript, not on a fixed time interval. The first chapter must start at 00:00. The minimum gap between two consecutive chapter starts must be 10 seconds. For each chapter return only the start timestamp in HH:MM:SS format and a one-sentence summary of what happens in that chapter. Do not write the chapter titles yet, only the splits and the summaries.
The output of this prompt is the structural backbone of the chapter block. The next prompt takes these splits and writes the curiosity-pulling titles that get the taps.
Prompt 2: write each chapter title in a curiosity-pulling style under 6 words
Chapter titles are micro-headlines. A flat “Setup” gets zero taps. A pulling “The piece nobody sets up first” gets the tap that lifts watch time. Use this prompt to turn the chapter summaries from Prompt 1 into short titles in your voice that the viewer wants to tap.
Curiosity-pulling chapter title writer
Act as a YouTube title editor writing chapter titles for a long-form video. Each chapter title appears next to its timestamp in the video description and in the YouTube UI. The point of the title is to pull the viewer into tapping the chapter, which is what lifts watch time. Video title: [paste] List of chapter summaries from Prompt 1 in the form [HH:MM:SS] one-sentence summary: [paste] Creator voice in three adjectives (e.g. plain-spoken, dry, direct): [paste] Produce one chapter title per chapter. Each title must be under six words. Each title must include either a curiosity gap, a small contrarian framing, or a specific concrete noun. No flat labels like "Introduction" or "Setup". No clickbait words like "Insane" or "Shocking". The first chapter title must work as the entry point to the whole video. Return the result as a clean list: [HH:MM:SS] Chapter title, one per line, in the same order as the input.
The output of this prompt is what the viewer actually reads in the chapter list. Lock the voice adjectives once for your channel, the bot reuses them on every video.
The 3-minute overview of how this works
Before the build steps, watch this short overview. It’s the exact video from the Automations Made Easy page, and it walks through the mechanics behind machines like this one. 1,000+ students have used these mechanics to save two hours a day, with zero coding.
Want to learn the mechanics behind YouTube automations like this?
I teach the same mechanics that make this watch-time compounder run inside Automations Made Easy. 1,000+ students have used these mechanics to save two hours a day and turn every long-form video into a chaptered, ad-rev-lifting upload without touching YouTube Studio. No coding required.
Week one: five videos, five sets of chapters, all written while I slept.
Most people quit a system like this in the first week because the per-video lift feels too small. There is no big spike, no traffic surge, no viral moment. The first week of the watch-time compounder is meant to feel small. The point is to confirm the bot fires on every new upload, the chapter splits land on real topic shifts, the titles read in your voice, and the YouTube description gets patched without losing the CTAs that were already in there.
Week one looks like this. You upload five long-form videos across the week. Each upload triggers the bot. The bot pulls the transcript, finds the topic shifts, writes the chapter titles, patches the description. Five fully chaptered videos go live without you opening YouTube Studio once. That is roughly 100 minutes of chapter work you did not do, a full working morning saved. Each chaptered video will earn ~3 to ~7 euros of extra ad rev over its life from the watch-time lift alone. Quietly, in the background, while you record the next one.
The point of the first week is not the watch-time spike. The point is to prove the loop closes: video uploads, bot fires, chapters land on real topic shifts, description gets patched cleanly, viewers can finally navigate, the algorithm starts to notice.
From there the maths is simple. ~20 minutes saved per video across ~100 to 150 videos a year is ~33 hours, roughly a full working week bought back every year. ~$3 to ~$7 of extra ad rev per chaptered video across ~100 to 150 videos is ~$500 to ~$1,000 a year. Over five years, with the catalog growing and the algorithm continuing to reward the stickier watch sessions, that compounds into ~$8,000 to ~$15,000+ in added ad rev from videos you would have published anyway.
All of this runs while you record, sleep, or take a week off. The bot does not care. The trigger reads the channel, the chapter detector reads the transcript, the title writer pulls the viewer in, the publisher patches the description through the YouTube API. The chapters land on every video, the watch time lifts on every video, and the ad rev compounds quietly in the background.
Prompt 3: generate description-friendly format with the chapter list + topic summary on top
The chapter block has to land in the YouTube description in the exact format YouTube expects, or the chapters do not show up on the video. Use this prompt to wrap the chapter list in the right format and write a short topic summary that sits above it so the viewer knows what the video is actually about.
Chapter block + topic summary builder
Act as a YouTube description editor formatting a chapter block for a long-form video. YouTube only accepts chapters in a specific format: the first timestamp must be 00:00, each line must be a timestamp followed by a title, and there must be at least three timestamps each at least ten seconds apart. Video title: [paste] Chapter list from Prompt 2 in the form [HH:MM:SS] Chapter title: [paste] The single promise the title makes to the viewer: [paste in one sentence] Produce a complete description block in this exact structure: a two-line topic summary at the top in plain English (what the video covers, who it is for), one blank line, the heading "Chapters:" on its own line, one blank line, then each chapter on its own line in the format "HH:MM:SS Chapter title" with no extra characters. The HH part may be dropped for videos under one hour, so a line should read "MM:SS Chapter title" for short videos. Return only the block, no commentary, formatted so it can be pasted straight into a description.
The output of this prompt is what the bot pastes into the YouTube description. Lock the format once, every video gets the same clean chapter block at the top.
Prompt 4: insert the chapter timestamps cleanly into the existing description without overwriting the CTAs
The trickiest part of the whole engine. Your existing description already holds CTAs, affiliate links, social handles, and a signature block. The bot has to drop the chapter block in without breaking any of that. Use this prompt to plan exactly where the chapter block goes and what stays untouched.
Description patcher (CTA-safe)
Act as an editor patching a YouTube video description with a new chapter block without overwriting any existing content. The existing description holds CTAs, affiliate links, social handles, and a signature block, all of which must remain in place exactly as they were. Existing description as it lives on YouTube right now: [paste verbatim, including line breaks] Chapter block from Prompt 3, formatted for the description: [paste] Produce a single new description in plain text, formatted for a YouTube Data API PATCH call. Insertion rules: the chapter block must sit at the very top of the description, above the topic summary that is already there, separated by one blank line. Every existing CTA line, affiliate link, social handle, signature, and disclaimer must stay in the exact same position relative to everything else. Do not edit, shorten, rephrase, or reorder any existing content. Do not add any new content other than the chapter block itself. Return only the full final description, no commentary.
The output of this prompt is what the bot PATCHes to the YouTube Data API on the videos endpoint. Existing CTAs stay live, affiliate links keep paying, the new chapter block sits at the top, the video gets the watch-time lift.
The exact build, step by step
Wire the YouTube webhook or RSS trigger that fires on every new upload
The trigger is the source of truth for the whole bot. Point a small workflow (n8n, Make, or Zapier all work) at your channel’s RSS feed (https://www.youtube.com/feeds/videos.xml?channel_id=YOUR_ID) or at the YouTube PubSubHubbub webhook for a near-instant push. Every time a new public video appears on the channel, the workflow fires once with the new video ID. Filter for videos longer than 10 minutes so shorts and live-stream clips do not trigger the bot. This single piece is what makes the engine event-driven instead of scheduled.
Pull the video transcript with timestamps
When the trigger fires, the bot fetches the transcript. There are three sources. The fastest is the YouTube auto-generated transcript, available via the YouTube Data API a few hours after upload, free. The most accurate is Whisper through OpenAI, Deepgram, or AssemblyAI, paid a few cents per video but cleaner output and better topic-shift detection. The third option is a backup captions file you uploaded yourself. The bot tries the YouTube auto-transcript first, falls back to a paid transcription source if the auto-transcript is missing or low quality. Either way, the output is a timestamped transcript with one chunk per line, ready for the chapter detector.
AI Chapter Detector splits the transcript into themed chunks with timestamps
This is the brain of the bot. The transcript goes through the chapter detector (Prompt 1 above). The detector reads the full transcript, finds the real topic shifts, and groups the lines into five to eight themed chunks. Each chunk gets a start timestamp and a one-sentence summary of what happens. The first split is always 00:00. Splits cannot be closer than 10 seconds apart (YouTube’s hard requirement for chapters to render). The output is the structural skeleton of the chapter block. No titles yet, just the splits.
Title each chapter in under 6 words in your voice
The splits are the skeleton. The titles are what the viewer actually reads. The bot runs the title-writer prompt (Prompt 2 above) against each chunk and produces a short curiosity-pulling title in your voice, under six words. No flat labels, no clickbait words, no padding. The title sits next to the timestamp in YouTube’s UI on mobile, where most viewers see it, so it has to fit and pull at the same time. The output is a clean list: timestamp, then title, one per line, ordered.
Build the description block (chapters list + format)
YouTube only accepts chapters in a specific format. First timestamp must be 00:00. Each line must be a timestamp followed by a title. Minimum three timestamps, each at least 10 seconds apart. The bot runs Prompt 3 against the titled chapters and produces the final block, formatted so it slots straight into a description. A short two-line topic summary sits above the chapter list so the viewer knows what the video is actually about. The block is YouTube-format compliant. If any rule is broken, YouTube silently refuses to render the chapters.
PATCH the YouTube description via the Data API, log the new watch-time delta
The last step is a single API call. The bot builds the new description with the chapter block at the top and every existing CTA, affiliate link, and signature line untouched (Prompt 4 above). It sends a PATCH request to https://www.googleapis.com/youtube/v3/videos with the part snippet, the video ID, and the new description. YouTube returns 200 and the chapters render on the player within a few minutes. Seven days later, the bot pulls the watch-time data for that video and writes the delta versus the channel’s running average to a tiny log file or Airtable row, so you can see the lift accumulate, video by video.
Trigger fired
YouTube channel feed sees the new video, the workflow fires once with the new video ID.
Transcript pulled
YouTube auto-transcript pulled via the Data API, or Whisper fallback if the auto-transcript is missing.
Chapters built
Topic shifts detected, splits placed, titles written under six words, description block formatted to YouTube spec.
Description patched
YouTube Data API PATCHes the description, chapters render on the player, watch-time delta logged seven days later.
Build this watch-time compounder inside the same playbook 1,000+ students use
Automations Made Easy teaches the mechanics behind bots like this one. Step by step, no code, plain English. Pull real chapters out of every video you upload without opening YouTube Studio.
The six months after I switched it on
Here is the shape of the first six months after I turned the chapter bot on for the channel. The line is intentionally modest in the early months and steady from there, because that is how a watch-time compounder actually behaves. Small per-video lift, every video, forever.
Added monthly ad rev from the watch-time lift on chaptered videos
Caption: Monthly added ad rev attributable to the watch-time lift on chaptered videos. Three things matter on this chart. The line stabilises around $80 a month after six months and stays there. The revenue comes from videos that would have shipped without chapters before. And every single one of those months happens without you opening YouTube Studio.
What other students built
I teach the simple skills behind machines like this in Automations Made Easy. Students who built their own version sent back what changed in their first month.
“I had been uploading two videos a week for fourteen months and none of them had chapters. Two weeks after switching the bot on, my last 12 uploads all had clean chapter blocks and my average view duration went up from 32 percent to 39 percent.”
“Ran the backfill on my back catalog and the catalog watch time went up about 6 percent across the next 90 days, on videos I had not touched in two years. The bot paid for itself in week one.”
“Used to dread the chapter step on every Sunday upload. Now the chapters are in the description by the time I sit down with my coffee. I got my Sunday morning back.”
“The curiosity-pulling chapter titles were the missing piece. My old chapters said ‘Intro’ and ‘Outro’. The new ones get tapped. Watch time on my last six videos sits 8 percent higher than my channel average.”
What’s inside Automations Made Easy
AME isn’t a library of pre-built automations. Every business is slightly different. What’s reusable across all of them is the underlying mechanics: how to set up little machines that listen, write, and follow up while you sleep, and how to wire the pieces together without writing code.
The program walks you through six modules: The Right Tools (the cost-effective, no-code stack I actually use), Task Selection Mastery (which automations are worth building first), Design Secrets (mapping an automation before you build it), Zero to Hero (complete beginner to confident automator), Real-World Application (we build a full automation together, end to end), and Monetization Mastery (turn the skill into a side-business).
It also includes done-for-you templates you import in two clicks, over-the-shoulder training videos, and the same playbook 1,000+ students have used to save two hours a day. No coding required. If you can copy and paste, you can build this.
The chapter bot: common questions
Pulled from what readers and Automations Made Easy students ask most.
Will YouTube’s own auto-chapters do this for free?
Sometimes, on a small percentage of videos, and never the way I want. YouTube’s auto-chapters are inconsistent, often miss the real topic changes, and the titles are flat (“Introduction”, “Part 2”). The bot I run reads the actual transcript, finds the real topic shifts, and writes curiosity-pulling chapter titles in my voice. The difference is the click rate from the chapter list. Auto-chapters get scrolled past. Human-style chapter titles get tapped. The tap is what lifts watch time, and watch time is what the algorithm pays for.
Does this work for short videos (under 5 minutes)?
YouTube only displays chapters on videos longer than 10 minutes with at least 3 chapters of at least 10 seconds each. For anything shorter, the bot still generates the chapter list but writes it as a plain summary inside the description instead of timestamps. The watch-time lift is smaller on short videos because there is less to skip around, but the description still gets the structure benefit and the video still ranks slightly better for the topics in the chapter titles. The big payoff is on videos longer than 12 minutes.
What if my video has no transcript?
Run a cheap transcription step in front of the bot. Whisper through OpenAI, Deepgram, or AssemblyAI all do it for a few cents per video. YouTube also auto-generates a transcript a few hours after upload, and the bot can pull that one straight from the YouTube Data API for free if you are willing to wait. Either way, the bot needs a transcript with timestamps. Without it, the chapter detector has nothing to chunk. Transcription is the single biggest-payoff upgrade for this whole engine, set it up once and forget it.
Will my old videos benefit if I add chapters now?
Yes, and this is the move most people miss. The bot can be pointed at your back catalog with a one-time backfill. It pulls every video over 10 minutes, generates chapters, patches the descriptions, and walks away. Older videos that already rank get a small re-ranking bump because the description is now richer and the watch time on those videos starts climbing as viewers use the chapters to skip to what they want. I ran the backfill on roughly 80 of my older videos and the catalog watch time went up around 6 percent across the next 90 days, on videos I had not touched in years.
How much does the watch-time lift actually translate to in ad rev?
Conservatively, each chaptered video earns around 3 to 7 extra euros of ad revenue across its life from the watch-time lift alone, before any second-order effects from re-ranking. At 100 to 150 videos a year, that is roughly 500 to 1,000 of added ad rev per year, compounding to ~8,000 to ~15,000+ across five years as the catalog grows and the algorithm rewards the stickier watch sessions. The per-video number is small. The catalog effect is what makes this engine worth running. The 20 minutes of chapter work saved per upload is the cherry on top.
More Money Makers like this one
Built from the same handful of mechanics. Each one captures, converts, or earns money that would have walked out the door.
Run this watch-time compounder yourself, or learn the mechanics inside Automations Made Easy.
If you want to learn the mechanics behind YouTube automations like this and build your own at home, Automations Made Easy is the playbook. Step by step, no code, plain English. If you want to talk through which trigger, transcription source, and description-patch flow would work for your specific channel first, I take a small number of consulting clients each month.
€497 one-time · Lifetime access · 1,000+ students
