Fair Price Index

A venue's Fair Price Index (FPI) is a weighted average of verified customers' ratings, pulled towards the category average when the sample is small, with a lower bound that tells you how sure we are. It runs from 0 to 100 and answers one question: do enough real customers think what they got was worth the price?

How to read it

The index is the point estimate. Above 80 means that the large majority of verified customers got value for their money.
The lower bound matters more than the point estimate: it reflects how many ratings there are and how much they agree. A small sample penalises itself.
Confidence tells you how sure we are: limited below 20, good up to 49, high from 50 effective ratings.
The price range (€, €€, €€€) comes from receipt amounts compared with the category median. An expensive restaurant can have a high index: we don't measure whether it's cheap, but whether it's worth it.

Each purchase is rated on the five survey questions, weighted 40/25/15/10/10. Example: 9, 4, 4, 5, 4 → 83.5.

The five measures

1 · Overall value · scale 0–10
Was the price fair for what you got?
40%
2 · Quality · scale 1–5
Did the quality justify the price you paid?
25%
3 · Quantity / size · scale 1–5
Did the quantity, time or size justify the price?
15%
4 · Transparency · scale 1–5
Was the final price clear in advance?
10%
5 · Would buy again · scale 1–5
Would you pay the same price again?
10%

The formula

The Fair Price Index of a venue is a weighted average of the ratings from verified customers. With small samples it is pulled towards the category average, and a lower bound shows how certain we are. Every purchase is rated on the five survey questions, weighted 40/25/15/10/10. Version: fpi_v1.2.

purchase score s = 0.40·Q1 + 0.25·quality + 0.15·quantity + 0.10·transparency + 0.10·would buy again Q1 “overall value” 0–10 → 0–100; the others 1–5 → 0–100 time weight w = 0.5 ^ (age in days / 120) after 365 days: 0 account weight a = 0.4 → 1.0 by account age and variety effective sample n_eff = W² / Σ(w·a)² W = Σ w·a raw index FPI_raw = Σ w·a·s / W index FPI = (n_eff · FPI_raw + 25 · C) / (n_eff + 25) C = category average lower bound FPI_LB = FPI − 1.28 · σ_w / √n_eff “Fair Price” badge n_eff ≥ 40 and FPI_LB ≥ 80 and 60-day hold and no open report

The five measures. Overall value 40% · Quality 25% · Quantity / size 15% · Transparency 10% · Would buy again 10%. Example: 9, 4, 4, 5, 4 → 83.5.
Decay. A rating loses half of its weight in 120 days. A venue that improves sees it in the index within weeks.
Prior. A venue with 5 ratings cannot look more certain than one with 500. 25 “virtual” ratings at the category average pull small samples towards the average.
Lower bound. The badge is awarded on the lower bound, not on the point estimate. A small sample penalises itself.
New accounts. At most 20% of a venue’s weight can come from accounts younger than 30 days.
Confidence. limited below 20, good up to 49, high from 50 effective ratings.

The methodology code is open, with public test vectors, so anyone can compute the index of any venue. A review of the weights is planned after the first 5,000 ratings and will be released as a new version with an explanation.

When a venue gets the “Fair Price” badge

An effective sample size above 40, a lower bound above 80 for 60 days, and no active integrity report. The badge is removed automatically when the conditions are no longer met; the QR code on the sticker always leads to the live status.

The methodology code is open, with public test vectors, so anyone can calculate the index of any venue. A reassessment of the weights is planned after the first 5,000 ratings and will be released as a new version with an explanation in the Trust Center.

See the index live

Every company and every venue in the directory has an index, a lower bound and a confidence level.

See the companies Trust Center