How the rankings work
1. Evidence, not stars
Star ratings mostly measure how much students liked a tutor. TutorRank instead reads every testimonial on a tutor's public profile and keeps only the ones that cite a concrete result: "went from a C to an A on her tests", "quiz scores rose from 70 to 90". Each one becomes a before → after data point on a 100-point scale.
2. Letter grades become points
A = 95, B = 85, C = 75, D = 65, F = 55; a plus adds 3 and a minus subtracts 3 (B+ = 88, A− = 92). Percent scores are used as-is. So "C to an A" = +20 points.
3. Tests count more than quizzes
Each data point is weighted by what it measures: test grades × 1.0, overall class grades × 0.9, quiz grades × 0.8.
4. Harder subjects count more
A gain in Calculus asks more of a tutor than the same gain in Algebra 1 — higher courses presuppose everything beneath them. Each data point's weight is multiplied by its subject's level on the curriculum ladder, 0.1 per level: Elementary Math × 0.6, Middle School Math × 0.7, Algebra 1 × 0.8, Geometry × 0.9, Algebra 2 × 1.0, Trigonometry × 1.1, Statistics × 1.1, Precalculus × 1.2, Calculus × 1.3. Testimonials that don't name a course count × 1.0. These multipliers are editorial judgment, published here in full; the raw before → after numbers on every tutor page are not affected by them.
5. High-end points are harder to earn
Raising a 90 to a 100 means eliminating your last mistakes — harder than fixing your first ones. Each point gained is therefore weighted by the band it's earned in: 60–70 × 1.0, 70–80 × 1.25, 80–90 × 1.5, 90–100 × 2.0. A 90 → 100 counts as 20 weighted points; a 60 → 70 counts as 10. Gains that span bands sum their parts. Scores below 60 are floored at 60 first — very low baselines are statistically noisy, and without the floor one dramatic anecdote would outweigh consistent evidence.
6. Many wins beat one big win
A tutor's Impact Score for a subject is (weighted evidence count)0.8 × (average weighted gain)0.2. The count is deliberately dominant: ten modest gains outrank a single dramatic story. With the same amount of evidence, bigger weighted gains win.
7. Value, and a price floor
Value = Impact Score ÷ hourly rate: how much proven improvement each dollar buys. Tutors who don't list a price aren't ranked — without a rate, their value can't be compared. We also rank only tutors charging $60/hr or more: this board is for choosing among serious, established tutors, and bargain-rate listings are a different search.
A bias you should know about: this favors Wyzant
Wyzant reviews routinely cite a real grade (“went from a C to an A”). Preply reviews almost never do — of 822 Preply testimonials we read, 6 cited a concrete grade, versus 84 of 10,236 on Wyzant. Because our method scores nothing else, the rankings are populated almost entirely by Wyzant tutors. This is a difference in review culture, not in teaching quality: a great Preply tutor is simply invisible to a method that demands numbers. Full detail on the disclaimer page.
What gets excluded
Vague praise ("helped so much!"), ambiguous extractions, grades that went down, low-confidence readings, gains entirely below the 60-point floor, and tutors with no listed price. Excluded testimonials are stored, and every tutor page shows exactly which ones scored and which were excluded, so you can audit any ranking yourself.
Honest limitations
Testimonials are self-selected and unverified; a tutor with few reviews isn't necessarily worse, just less proven. Data is a sample from Wyzant and Preply, refreshed manually. Always read the original profile before hiring.