- Home
- How we rank
How we rank
Every guide on this site starts from the same question: of all the videos that answer this search, which one should someone actually watch, and why? Here is exactly how that gets decided.
1 · What gets scanned
For each topic we search YouTube through its official Data API and pull metadata for a hundred-plus candidate videos across several phrasings. A topic only becomes a page if that scan finds real material behind it: at least eight videos long enough to teach something, from at least four different channels, with a meaningful audience and at least one upload in the last three years. Topics that fail are not published — and there is no page telling you they failed.
2 · What gets analysed
From that pool we take a shortlist, capped at two videos per channel so one prolific creator cannot dominate a ranking. Each finalist is read in full: its timed transcript, its chapters, and its most-liked viewer comments. A video whose transcript we cannot retrieve is not ranked, because we will not judge content we have not read. When a finalist drops out, the next candidate takes its place.
3 · How they are ranked
Four criteria, weighted:
- Accuracy (40). Do the claims hold up, and do they agree with the other videos on the same task? Where videos contradict each other, we say so and explain which reading we follow.
- Pedagogy (25). Order, pace, and whether the reasons are explained or only the steps.
- Coverage (20). Does it finish the job, including the parts that are easy to leave out?
- Freshness (15). Interfaces, prices and recommendations age; a confident video can be confidently out of date.
There is rarely one winner. Most guides name a best overall, a fastest, and a best for beginners, because those are different videos more often than not.
4 · What we flag
Criticism on this site is specific and timestamped, or it is not published. A missing safety step, a figure contradicted by a qualified commenter, an unmarked sponsor segment — those get flagged with the moment they occur. "This channel is bad" is not a finding.
5 · How the pages are made
The synthesis is machine-written from the transcripts and comments described above, then checked by a separate automated reviewer that scores it on grounding, specificity, originality and whether the page would replace watching the video. Pages below the threshold are not published — that gate has already rejected work on this site and sent it back for correction. A human reviews before anything goes live.
Nothing here is written from memory or from other people's summaries. If a claim is on a page, it came from the evidence collected for that page.
6 · Corrections
If you made one of these videos and something here is wrong — a mis-timed reference, an outdated flag, a ranking based on a version you have since fixed — tell us and we will correct it, or re-run the analysis against your current video. Rankings change when the evidence changes; that is the point of dating every page.