How tutor listings rank.

The formula

We use a Bayesian-weighted average — a standard rating-system formula that blends each tutor's actual reviews with a platform-wide prior. The effect: a tutor with one perfect review doesn't outrank a tutor with fifty consistently great reviews.

n

Number of reviews this tutor has received.

avg

Their actual average rating across those reviews (1-5 stars).

m = 5

The "prior weight" — how many reviews we treat as enough to trust the average. Below this, the formula pulls toward C; above it, the tutor's own average wins.

C = 4.5

The platform-wide average rating. Recalibrated periodically once we have enough data.

DBS boost = +0.05

A small additive boost for tutors with a current DBS-Verified badge — adds 0.05 stars to the Bayesian score. Modest enough that a strong review record beats it; visible enough that DBS-Verified tutors get a fair lift over peers with similar review records.

A worked example

Four tutors, ranked by the formula above. Run the numbers yourself if you like — the constants on this page are the literal values used in the code.

• Ben: 50 thoughtful reviews — the average reflects steady consistent performance and dominates the prior.
• Anya: One five-star review — the formula trusts it less than fifty four-and-a-half-stars.
• Chiara: New tutor, no reviews yet. Sits at the prior, lifted slightly by the DBS-Verified badge.
• David: New tutor without DBS-Verified — sits exactly at the prior.

What's in the formula

• Reviews from real students

  Verified reviews — only students who paid the £20 unlock can write one — are the primary signal. The Bayesian weighting means consistency wins over a single perfect review.

• DBS-Verified status

  A flat +0.05-star bump on the Bayesian score. Manual DBS verification is a meaningful safety commitment, so it gets a lift — but it never overrides a strong review record (Ben outranks Chiara in the table above).

What's not in the formula

Signals we deliberately leave out, because they punish tutors for things that aren't about teaching quality.

• Response time

  We don't track how fast you reply. Real life doesn't run on SLA timers — you might be teaching, on holiday, or just choosy about who you take on.

• Hourly rate

  Default sort doesn't push lower-priced tutors above higher-priced ones. You set your rate; the student decides if it's worth it. (Students who want price-sorted results explicitly pick that sort option.)

• Pay-to-rank tiers

  No "boosted", "featured", or "pro" tier. We don't sell rank, full stop. The £20 unlock fee is the only money that changes hands on the platform.

• Internal favourites

  Admins can hide profiles for safeguarding violations or compliance issues, but never promote one tutor over another. The ranking code is the same for everyone, including us.

A note on city-based searches

On searches that name a specific city (e.g. /tutors/maths/in/manchester), in-person tutors based in that city appear first, then online tutors, then everyone else who matches. Within each tier, the same Bayesian formula above decides the order. Geographic relevance trumps a small ranking gap when parents are looking for nearby help.

A meritocracy, not an auction

Marketplaces have a habit of sliding into pay-to-play: boosted tiers, sponsored slots, premium memberships. The temptation is real because it scales. We're choosing not to, because the moment a tutor's rank depends on what they pay us, the ranking stops being about teaching quality and starts being about wallet size.

If we ever change the formula or the constants, you'll see it announced here first, with the full new ranking documentation in the same shape as this page. No quiet algorithm shifts, no opaque "editorial" thumb on the scale.