How to read the scores. Quality is how good the teaching is on its own merits, 1 for filler and 10 for exceptional. Depth is specificity and worked examples, the difference between "make good thumbnails" and a repeatable process. Fit is relevance to your channel: expert-led, B2B, selling high-ticket services, not chasing AdSense or a mass-market audience.

How much to trust it. Each lesson was read by DeepSeek three times and the median taken, which cost about ten cents for the whole course. The three votes landed within about one point of each other on average, so the ordering is stable, not noise. What it cannot judge is whether Shane's advice actually works, only how well it is taught and how relevant it is to you. 28 of the 133 items in the course are intros, summaries, or quizzes with no video, so they were skipped rather than scored, and they are not in the table below. Rechecked against the live classroom on 28 July 2026, which is where the extra lessons since the first pass came from.

One caveat on precision. Twenty-one lessons were re-scored on 28 July after a better transcript parser recovered text the first pass had dropped, and one lesson that had no captions at all was transcribed and scored for the first time. Most of those scores moved by a point or less, which is inside the model's own disagreement between votes: treat a one-point gap between two lessons as a tie, not a ranking. What the refresh did change is the shortlist below, which gained two lessons that had been sitting just under the bar.