How we score and review courses

Reviews come from real students

Every review is written by someone who tells us they took the course. Reviews cover overall rating, value for money, content quality, beginner friendliness, instructor quality, platform support, and refund experience. We don't write reviews ourselves, and we don't accept payment from platforms or instructors to influence a listing.

Six categories, not one blended number

The Trust Score isn't a single formula — it's six independently scored categories, each weighted by how much it should matter: Evidence Quality, Student Satisfaction, Creator Trust, Course Quality, Platform Trust, and Value for Money. A category only contributes when we actually have real, independent data for it. When we don't, that category is marked not enough datarather than filled in with a guess — and it's excluded from the overall number rather than silently dragging it toward a neutral default.

Built to resist gaming, not just to look confident

A course with two 5-star reviews does not get a perfect score. Ratings are weighted toward the site-wide average until enough reviews exist to outweigh it (the same statistical shrinkage IMDb uses), recommend percentages use a confidence-interval lower bound rather than a raw ratio, and older reviews count for less than recent ones. Each review is also weighted by the reviewer's own track record — verified purchase, review history, and account signals — so a handful of coordinated or low-trust reviews can't swing a score the way genuine, independent ones do.

Coverage and Confidence are shown separately from the score

A high score doesn't automatically mean we know a lot about a course — so we say so. Coverage is the percentage of signals we actually have data for, out of everything the Trust Score could consider. Confidenceis a plain-language tier (never a fake percentage) describing how much we'd trust the current number holding up as more evidence arrives. It's possible — and honest — for a course to score 89 with only 41% coverage: the number is real, but there's more we don't yet know than a bare score would suggest.

Every score is auditable

The course page's “Why this score?” panel breaks the Trust Score down by category and, within each, by the individual signals that fed it — nothing is a black box. Reviews flagged for spam signals, likely duplication, or other trust discounts are shown with their reason, not silently dropped or silently counted at full weight.

A course moves through three states as evidence accumulates:

• Not enough data yet — no reviews yet, or too few to be meaningful. We show this instead of a number, because a score built on little or no evidence is worse than no score at all.
• Early signal — a small number of reviews exist. We show this as a qualitative signal, not a precise numeric score, because the picture can still shift significantly as more students weigh in.
• Verified score — enough published reviews exist for the score to be statistically meaningful. This is the number shown across the site and in search results.

What we never do

• We never fabricate, estimate, or interpolate a score when data is insufficient.
• We never let a platform's own rating stand in for our independent score.
• We never let payment or partnership affect a course's listing or score.
• We never show a fake-precision confidence percentage — confidence is always a plain-language tier.
• An LLM may help explain a score in plain language; it never calculates one.

Moderation

Submitted courses and reviews are checked before they go live — duplicate courses are merged, and reviews that don't look like genuine student experiences are held back. If you think something on the site is wrong, please let us know.