01
Some videos land, most don't, and you can't tell what separated them
You can see which ones worked. You cannot see why, which means you cannot do it again on purpose. Every upload is a fresh roll.
Long-form and Shorts · documentary scripts, and the retention architecture underneath them
Most scripts don't fail at the writing. They fail at the structure underneath — and that's diagnosable before a frame is shot. I write long-form and Shorts for documentary and essay channels, and build the editorial system they come out of.
Who it's for
For channels where the video makes an argument — documentary, essay, explainer. Faceless or on-camera; the structure is the same either way. Whether you've plateaued and can't see why, or you're growing and can't hold the quality as you scale, the script is usually where it's decided.
Every channel has a cliff. It is almost never where the creator thinks it is, it is written in before anything is shot, and it cannot be fixed in the edit.
Four things I hear from almost every channel that gets in touch. If none of them land, I'm probably not who you need.
01
You can see which ones worked. You cannot see why, which means you cannot do it again on purpose. Every upload is a fresh roll.
02
First draft came back wrong, so you fixed it. Then the second. By the fourth you stopped sending notes and just did it yourself again.
03
The drop is right there at forty seconds. What you don't know is whether that's the script, the edit or the thumbnail — so you change all three and learn nothing.
04
Each video is fine on its own. Together they read as a collection rather than a body of work, and nothing carries from one to the next.
If you write them yourself
Research, structure and draft take longer than everything else combined, and you still can't tell whether it's working until it's live. I take the writing; you keep the channel.
What you'd get →If you've outsourced before
The last writer produced one video at a time, so nothing carried and the voice drifted. The fix isn't a better writer, it's a system the writing comes out of.
How that works →If you write for clients
Editing capacity you can add in a week. A writer who can hold a client's register across a whole slate takes months to find. White-labeled, your name on the work.
Agency terms →Three openings, each written twice. On the left, what the same brief normally produces. On the right, what I shipped.
Hey everyone, welcome back to the channel. In today's video we're going to be talking about something that affects millions of investors around the world.
Studies show that people who check their portfolios frequently tend to make emotional decisions, and this can have a big impact on your long-term returns.
So if you've ever felt this way, make sure to stick around, because we're diving deep into the psychology of investing.
Seven words of greeting before the subject appears, then “millions of investors around the world,” which is everybody and therefore nobody. “Studies show” with no study. “A big impact” with no size. It asks the viewer to stay for a payoff it has not described, and in seventy-one words the only concrete noun is portfolios.
Refreshing your portfolio has never once made you money. You already know that. You will still check it four times before lunch.
You are not deciding anything. You are checking a wound.
In March 2020, at the peak of the panic, the app went down and stayed down for a full trading day. Locked out, unable to trade, people refreshed a screen that would not load, over and over, all day.
The first line is a flat contradiction about something the viewer did this morning, and the second closes the exit: you already know that. There is no lesson left to offer, so the only thing still worth watching for is why. The reframe then names the behavior honestly — this is not a video about panic selling. And the outage is not colour, it is the proof: the people refreshing that screen could not trade even if they had wanted to, which is the only way to show the compulsion existed separately from the decision.
71 words before71 words afterSame length. One gestures at a feeling; the other names the thing you did.
Artificial intelligence is transforming the modern workplace at an unprecedented pace. In this video we'll be exploring how AI is revolutionizing performance management across industries.
Recent studies suggest that a growing number of managers are leveraging AI tools to streamline personnel decisions, and the implications for employees could be significant.
Stay tuned as we unpack what all of this really means for the future of work, and for your own career.
Read it once more and notice what it is: a script about AI, written the way AI writes. Transforming, unprecedented, revolutionizing, leveraging, streamline, implications, unpack, the future of work. Eight tells in seventy-one words. It also says “AI” twice before it mentions a single person, which tells a viewer who does not work in technology that this video is for somebody else.
Nobody wrote your last performance review. That is not a figure of speech, and it is not the part that should worry you.
The written warning was never protecting the company. It was protecting you.
It only worked because it was expensive. A manager who had to document three real failures needed three real failures to have actually happened. Producing one now takes seconds, and the page does not say which.
Six words that sound false, followed by a refusal to soften them. “That is not a figure of speech” forecloses the comfortable reading, and “not the part that should worry you” opens a second gap before the first one has closed. Then the reversal lands on the viewer's side: the document they resented was the one protecting them. Note what never appears in seventy-one words. No tool is named, no technology is explained, and the word AI is not used once — which is exactly why this holds a viewer who does not work in tech.
71 words before71 words afterSame length. One explains a technology; the other hands you a document you have already been given.
We've all been there. You promise yourself you'll put the phone down, and before you know it, hours have flown by. Sound familiar?
In this video we'll be diving deep into the fascinating science of why our brains are wired this way.
Studies show that short-form content is rewiring our attention spans. Let's jump right in.
“We've all been there” is among the most-used opening lines on the platform, which makes it the one line guaranteed to sound like every other video. “Sound familiar?” asks the viewer to agree before anything has been said. Then it promises science twice and delivers none, and calls its own content fascinating rather than being it.
At 11:47 you were still scrolling. You had meant it at eleven, and that is the problem.
The decision didn't fail. It competed.
On the other side was a feed that pays out on no schedule you can learn, the same design that keeps a hand on a slot machine. Willpower was real. It got outbuilt.
It opens on the failure, not the intention — 11:47 lands before the viewer is told what it means, so recognition happens before persuasion is attempted. Then the hook turns: you had meant it, and that is the problem. Meaning it is supposed to be the defence, so calling it the problem opens a gap the viewer cannot close on their own, and the only way to close it is to keep watching. The four-word reframe answers it, and hands them an opponent they can picture — a slot machine in the hand — instead of gesturing at “the science.”
56 words before56 words afterSame length. The rewrite gives you an opponent you can picture.
Most writers open a blank document and start typing. Each script becomes a fresh act of judgment, the voice drifts, and the channel never becomes one thing the algorithm can profile.
LAYER 01
What your channel argues that its competitors don't, where topics sit on the advertiser-category boundary, the retention shape your format has to hold, and the packaging logic that decides whether a video gets clicked at all.
Niche-agnostic by design — built across behavioral psychology, personal finance and enterprise tech. The architecture doesn't change between niches; the research and vocabulary get rebuilt every time.
LAYER 02
Every script comes out of Layer 01, which is why video twelve is recognizably the same channel as video one. Consistency stops being something you police in review and becomes something the process produces.
Delivered as a full production package — script, b-roll brief, editor brief, packaging variants and upload metadata.
This holds whichever alternative you're weighing it against — a cheap writer, a tool, or yourself at one in the morning. The problem was never who typed it. It's that nothing carried between one video and the next.
A script is where this ends up. Most of the work that decides whether a video holds attention happens before a word is written.
01
What your channel argues that its competition doesn't, and which topics have demand nobody has served.
02
Who already covered it, what angle they all missed, and whether the idea can carry a video at all.
03
The retention shape the format has to hold. Where the premise lands, where recognition happens.
04
Written to that structure, in your channel's register, with the production brief attached.
05
Read the retention graph at day 28. What held, what didn't, what changes next time.
Most writers hand over prose and leave the interpretation to whoever builds the video.
In your channel's register, sectioned so each part states what it's doing, formatted to paste straight into your voice tool. Sources for every factual claim.
Shot-by-shot direction described as behavior rather than category, with overlay specs, motion graphic build notes, and a reveal-marker table naming the moments the edit has to break pattern.
What the video argues, the assembly order, where pace shifts, and what must never be trimmed. Your editor starts without a call.
YouTube tests up to three title and thumbnail pairs and promotes the winner automatically. Most writers deliver one, which means you're guessing. You get three distinct hypotheses about why someone clicks, one marked recommended.
Description, chapters written as real search phrases, tags, sources and the pinned comment. Nothing left to write on upload day.
A real Blueprint, anonymised and partly redacted: the honest read, the four-gap analysis, and a full worked rewrite of one channel's opening, before and after.
PDF · 5 pages · no email requiredMost engagements never define this, which is why they end in an argument about whose fault the numbers are.
The Blueprint is charged once, and credited in full against your first four scripts. Every script after it is faster because the thinking is already done.
Taking on a limited number of new channels at a time. One writer, and the whole point is that the voice holds.
$199 one video
$750 one-time
$650 per video
$1,350 / month
Written to the same system, on the same voice. Either derived from a long-form you've commissioned, or conceived standalone. Sold in packs because one Short in isolation rarely tells you anything.
Published US rates for the same three purchases, so you can see what you are and aren't paying for.
| The purchase | A US agency | Here |
|---|---|---|
| Getting started — channel audit, positioning, content plan | $2,500 – $10,000 | $750 |
| Per month, channel outsourced | $4,000 – $12,000 | $1,350 – $4,500 |
| One long-form script | $400 – $1,200 | $650 |
Two of those are not the same purchase, and I'd rather say so than let the table imply otherwise. An agency retainer buys production, editing, thumbnails, ad management and an account manager between you and whoever actually writes. The writing is one line item inside it. If you already have an editor, most of that spend is on a layer you don't need.
The third row is the honest like-for-like, and note that it is not the cheap option — it sits mid-band for experienced US writers. What you're not paying for is the agency around them.
A long-form video costs you a week — research, writing, recording, editing, packaging. The script is the part that decides whether that week was worth spending.
Rates scale with runtime, research depth and turnaround. Fifty percent upfront, two revision rounds included.
Your editing capacity is probably covered and your writing capacity probably isn't. That's the part I take — briefed once per client channel, holding one voice across the whole slate, your name on the work and mine nowhere near it.
Scripts arrive with shot direction and an assembly order attached, so your editors cut instead of interpreting prose. Non-solicitation as standard.
Talk about capacityDeepest in personal finance, enterprise technology and AI, behavioral psychology, health research and business — those are the niches I've built in. The method is structural though, so it transfers to economics and industry, consumer culture, science and engineering. Anything where the video makes an argument.
What I don't write: true crime, partisan politics or current-affairs commentary, medical or health advice, and anything reaction, vlog, gameplay or personality-led. The first three sit inside YouTube's limited-monetization categories or carry legal exposure, which makes them a bad trade for you regardless of how well they're written. On health I report what the research found — I never write guidance.
No, and be careful with anyone who promises it. Views depend on your thumbnail, your edit, your upload timing and your channel's existing profile. What a script moves is how long people stay once they've clicked — and how many of them get far enough to be worth anything to you.
It's better with them. A screenshot of one video's retention graph roughly doubles what I can tell you. But the audit works from the public video alone, so it's optional, and I won't chase you for it.
Long-form scripts, four working days. Shorts, two. The audit comes back in three. Blueprints take about a week — most of that is watching your catalogue and your competitors properly, which isn't a step worth rushing.
Two revision rounds are included, scoped as corrections within the agreed angle. If the angle itself changes, that's a new script. Most of the risk is handled earlier than that though — the free teardown and the audit both exist so you can judge how I think before you commit to anything.
For an audit, three things: your channel link, one video you thought should have done better, and — optionally — a screenshot of that video's retention graph. That's it.
For scripts, more: analytics access or screenshots, the three videos you're proudest of and the three that disappointed you, who your competition is, what's off-limits, and who else has to approve a script before it's filmed. I ask for all of it once, in one document, rather than dripping questions at you for a week.
Yes — working that out is the job, not a prerequisite for it. Most channels that come to me have no defined structure, which is usually part of why the uploads don't read as one channel.
For one to three long-form scripts it doesn't much matter. From the fourth it does, so the Blueprint comes first, and the $750 is credited in full against those first four scripts. Eight scripts without a defined structure is eight separate acts of judgment, and by the fifth I'd be guessing at decisions I made in the second. That's the exact problem I'd have just finished diagnosing for you.
Both. Long-form is where most of the work sits, because that's where structure has the furthest to fall — but Shorts are written to the same system and the same voice, either pulled out of a long-form you've commissioned or conceived standalone.
They're sold in packs of four or eight. One Short on its own rarely tells you anything: the format is volatile enough that a single result is noise, and you need a run before the pattern means something.
They stay private. I don't publish client work, name clients without written permission, or reuse your positioning for anyone else — including channels in your subject area. Anything I publish publicly is anonymised past recognition, or it isn't published.
Research uses official, documented interfaces only. Nothing is collected by working around a login or an access control, and I delete material you've sent whenever you ask.
Yes, white-labeled, with non-solicitation as standard. Volume rates from $375 a script, on a committed monthly minimum. Details here.
One page back within 48 hours: where retention breaks, which layer it lives in, and what your packaging is currently signalling. It tells you what is wrong. What to do about it is the audit. No call, no obligation — if it's useful on its own, keep it and we never have to speak.