Jake Trinder runs over a hundred YouTube channels. His 15-minute training treats a video as five decisions in a row, where one wrong call sinks the lot. Here is the method, and the number behind each decision.
A YouTube video is a series system: the topic, the packaging, the hook, the body and the read afterwards all have to work, and any one of them can fail the video on its own. Make five decisions that are each 90% right and the chance all five hold is 59% (his illustration, not how YouTube mechanically works). So YouTube is a quality game, but the quality is in the decisions, not the camera or the edit.
The method: choose the topic from what the audience already chooses to watch (not what they say they need), package it on a pattern that keeps winning and know why it wins, open by removing the viewer’s doubt, deliver every condition the title promised, then read the result as a signal of where to look. Each decision has a number you can measure before you film.
Jake Trinder is the founder of Views To Clients and co-owner of VideoCore. The channels his team runs include Gary Vee, Ali Abdaal, ManyChat and Instantly. Everything below is his method as he states it in the video; the measurement notes are how VideoCore counts each step.
Jake starts with a principle from economics, bounded rationality: a decision can only be as good as the information it was made with. Most creators make the best call they can with the little they can see, and then blame the call. The fix is research before anything else, because you cannot find an opportunity you cannot see.
Measured: the field around your channel, counted from real uploads: which subjects won on which channels, which title constructions and thumbnail traits carry a higher share of winners, and what has not been tried in your field yet. That is what the research tabs on Find Ideas hold.
What people need and what they choose are not the same thing. Everyone says they want to eat healthy and then orders a takeaway. Economists call the gap revealed preference: you learn what people want from what they choose, so look at what your audience is already choosing to watch.
Measured: for any subject, the typical result of settled videos on it (the floor), the share that beat twice their own channel’s normal, the best result (the ceiling) and the smallest channel that won. Active demand is a date: how many of those winners are from the last 90 days.
Your video lands on a feed beside twenty others and the viewer can click one. The title and thumbnail have to get noticed first and clicked before the eyes move on, and they have to make a promise the video can keep, because a promise broken means low satisfaction and YouTube stops pushing it.
Certain ways of packaging keep winning because people respond to the same framing in similar ways. “The new rules of…” won in SEO, on LinkedIn and on YouTube, with a different topic and a different audience each time; only the framing stayed the same. That is a pattern. One winning video is not a pattern.
Understand why the framing works before using it. “Rules” means get these wrong and you fail; “new” says what you know may already be out of date; so a viewer in the market cannot not click. Copy a format without the reason and it stops working: in machine learning that is overfitting, and it is the commonest mistake Jake sees, an outlier lifted from a niche where the reason does not transfer.
The thumbnail has two jobs: get picked out of the feed, then get understood before the viewer scrolls. The title closes just enough of the gap to earn the click. A gut check: a good idea makes you nervous to mess it up.
Measured: every title construction and thumbnail trait, counted across independent channels with the share of winners that wear it, against the field’s base rate. Three channels before anything is called a pattern. VideoCore Lens shows the decode on any video you open.
The click did not just buy a view, it set an expectation. Every viewer arrives with two things, an expectation and a doubt. The hook’s job is to remove the doubt fast, with four questions answered in order:
The training’s own opening does all four in two minutes: give me 15 minutes and I’ll make you disgustingly good; over a hundred channels and the most viral video in the niche; every decision from idea to results.
Measured: the opening checks in VideoCore read a script or transcript for promise, proof, plan and persona and for how soon the first real value lands, counted the same way as the title constructions, so you can see which openings win in your field.
Most people plan a video by writing down everything they know and miss what the viewer came for. Software engineers solve this with acceptance criteria: before building anything they write every condition the program must meet, and it is finished when every condition passes, not when it looks finished.
Your title is the program. Decompose its promise: what has to be true by the end for the title to be true? Break each answer down again, to the smallest truthful components. That list is your outline. Every item is a single point of failure, and when they all pass you kept the promise.
The goal is satisfaction, not retention. The brain reacts to how good something is compared with what it expected. Jake’s training held viewers for about 12% of its length on average, which looks terrible on paper, and did over 800,000 views and added over $200,000 in monthly recurring revenue in a month (his own figures, quoted as his; VideoCore measured the view count at 818,672 on 6 October), because the right people got more than they expected. Satisfied viewers signal YouTube to show it to more people like them.
Measured: the Satisfaction Checker compares the packaging’s promise with what the video delivers; the retention tools say where viewers left, which is where to look, not the goal.
Early on you are predicting one thing: are the viewers satisfied? The strongest early signal Jake has found, which nobody talks about, is positive comments: the ratio of positive comments to views, and what the comments say (“in search for gold I found diamond”). Like-to-view and subscriber-to-view ratios back it up, with a view graph above the channel’s usual shape.
None of those numbers tell you what to do. They tell you where to look. A 3.4% click-through rate does not mean change the thumbnail; it means look at the thumbnail and ask whether that rate makes sense. Jake has had 2 to 3% thumbnails on highly successful videos. A sharp drop at five minutes means look at that moment: a direct call to action that turned 30% of viewers off can be cut in post, after publishing.
Measured: a young upload’s views against the channel’s own uploads at the same age, and likes and comments per thousand views against the channel’s own, captured at fixed ages so the comparison is honest. Failures still happen. If you know which decision was wrong, every failure makes the next video better.
VideoCore is built as this chain: one page per decision, a counted number at each, and the next step named. Nothing predicted; every figure says what was counted, how many and when.
VideoCore Lens shows a video against its own channel’s normal, with the title and thumbnail decoded, while you browse YouTube. No account, no credits.
Start free with your channel Get VideoCore Lens — free Run a free channel health checkThat every part of a video has to work for the video to work: the topic, the packaging, the hook, the body and the read afterwards. Any one of them can fail it alone, which reliability engineers call a single point of failure. Five decisions that are each 90% right leave only a 59% chance that all five hold, so a video can fail with nothing looking badly wrong on the surface.
Look at what the audience already chooses to watch rather than what they say they need. Search the problem your customers are trying to solve, find the floor (what a decent video on it gets, the median) and the ceiling (what the best get), check the views are real and the demand is still active, decide whether the floor is worth it, and then attach the subject to something the audience chooses in greater numbers to raise the ceiling.
The floor is the typical result a decent video on the subject gets, measured as the median of videos on it. The ceiling is what the best videos on the subject get. A good video can expect to land somewhere in that range, so the range tells you whether the subject is worth making before you film.
Attaching a subject to something the audience already chooses to watch in greater numbers, so the video stands out and reaches a wider audience. Jake Trinder’s example is turning a lead-generation video into “Alex Hormozi’s lead generation strategy”, which did 473,000 views and still ranks for the search term. The attachment has to be true and relevant.
Promise (confirm the viewer is getting what they clicked for), proof (why they should listen to you), plan (the steps you will take them through) and persona (who the video is for). They answer the four doubts a viewer arrives with, in order, and the hook’s job is to remove those doubts fast so the body can deliver the expectation.
A list of every condition that must be true by the end of the video for the title to be true, broken down into its smallest truthful components. The list becomes the outline, each item is a single point of failure, and the video is finished only when every condition passes. The idea is borrowed from software engineering.
No. Viewer satisfaction is the goal, and the brain judges a video against the expectation the click set. Jake Trinder’s training held viewers for about 12% of its length on average and still did over 800,000 views, because the right viewers got more than they expected. Retention tells you where to look; it is not the target.
Positive comments: the ratio of positive comments to views and what the comments say. Like-to-view and subscriber-to-view ratios back it up, together with a view graph above the channel’s usual shape at the same age. None of these say what to change; they say where to look.
Not on its own. A low click-through rate is a signal to look at the thumbnail and ask whether the rate makes sense for the audience and traffic source. Successful videos have run with 2 to 3% thumbnails. Judge the packaging against the channel’s own normal and against who is seeing it, not against a fixed number.