All courses › How we score these courses
How we score these courses
The rubric, the weights, and the things we refuse to do.
The six criteria
Every course is scored out of ten on the same six criteria, then weighted. Nothing is scored on vibes, and no provider sees a score before it publishes.
| Criterion | Weight | What earns a high mark |
|---|---|---|
| Currency | 24 | Content that reflects how AI search works this quarter. A course last revised in 2024 cannot score well here however good it once was. |
| Curriculum depth | 22 | Specific, sequenced, technically literate lessons. A syllabus of topic names scores lower than one with named mechanisms. |
| Practical application | 20 | Exercises, templates, audits, live builds. Something you do rather than something you watch. |
| Instructor authority | 16 | A verifiable record outside the sales page. Talks, published work, named clients, an agency that exists. |
| Value for money | 12 | What you get per unit of cost, judged against the rest of this index rather than in the abstract. |
| Support and community | 6 | Access to the instructor, to peers, and to answers when you are stuck. |
Currency carries the heaviest weight on purpose. This field turns over faster than any other corner of search marketing. A 2024 course on optimising for AI Overviews is not a slightly dated course, it is a course about a system that has since been rebuilt.
Ranking is editorial, scoring is arithmetic
The default order on this site is our editorial ranking. It reflects judgement about who a course is for and what it is best at, which no single number captures. The rubric score next to each course is pure arithmetic from the table above, it is calculated identically for every entry, and you can re-sort the whole index by it in one click.
Where our ranking and the rubric disagree, the course page says so and explains why. Two entries currently sit above their rubric position, and both carry a note giving the reason and the score they actually earned. We would rather show you the disagreement than hide it.
What we refuse to do
- No aggregate star ratings. We do not republish Udemy or Coursera star averages as though they were ours, and this site emits no
AggregateRatingmarkup anywhere. Our score is one editorial rubric, not a poll of users. Presenting it as user ratings would mislead readers and search engines alike, and it is exactly the shortcut that gets review sites penalised. - No rating markup on list pages. Review structured data describes one specific item, never a category or a list.
- No invented numbers. Where a provider does not publish a figure, the field reads not published. We never estimate a student count, a completion rate or a price to fill a gap in a table.
- No paid placement and no affiliate links. Nobody in this index pays us anything, so there is no commission that a higher ranking could earn.
- No silent conflicts. Where we have any relationship with a provider it is disclosed on the course listing and the review page itself, not buried in a footer.
How prices are recorded
We record the price we could actually reach at checkout on the date shown, not the price on the marketing page. Where a seller advertises different figures across its own properties, we publish the checkout figure and note the discrepancy on the review page. One membership in this index currently advertises three different monthly prices across its homepage, its about page and its checkout, and we say so.
Udemy prices are shown as a realistic range rather than a list price. The marketplace runs near continuous promotions, the crossed out figure is not a price anyone pays, and the real figure varies by country and by day.
What we could not check
We did not enrol in every paid programme in this index. Where our assessment rests on published curricula, provider documentation and independent reporting rather than first hand completion, the course page says so in its confidence rating and lists what could not be verified. A review site that implies otherwise is lying to you.
Re-review cadence
Every entry carries the date we last checked it. Courses not revised within twelve months are flagged on the listing. Programmes whose first cohort has not yet run are marked unproven and re-scored once one completes. Courses that close are kept in the index, marked closed, because a page that quietly disappears is how a reader ends up on a pirate mirror.
Last revised 2026.