Skip to main content
Data sources

Where our ratings come from

Every IndexFair score is built from genuine user reviews and first-party measurement of the product. We don't accept submissions and we don't take payment for inclusion. We continuously re-evaluate which sources carry real signal — so the categories below are the backbone of our coverage, not an exhaustive list. We deliberately don't publish the full source list, so the input cannot be reverse-engineered and systematically targeted. The full derivation lives in the methodology.

Main sources

We read genuine user reviews across these source types, then corroborate them with an independent estimate of each brand's real-world traffic and with our own first-party measurement of the live product (game libraries, market pricing). A published score requires at least two distinct data sources — no single source determines a verdict on its own.

App stores

May include, for example: Apple App Store, Google Play

Verified-install reviews at scale — the closest thing to a representative cross-section of a brand’s actual users.

Public forums

May include, for example: Reddit

Unsolicited community reports, including unresolved complaints that may never reach a formal review platform.

Social media

May include, for example: X (Twitter), Facebook

Public posts and operator mentions, retained only as aggregate signals after provenance and quality checks.

Review platforms

May include, for example: Trustpilot, AskGamblers, Casino Guru

High-volume written feedback, read for the substance of each report rather than the headline star rating.

Regulator complaints

May include, for example: BBB, CFPB, UKGC/ADR, Reclame Aqui

Public complaint and ombudsman records, shown with jurisdiction and observation date. A report is a signal, not a finding of misconduct.

Independent test labs

May include, for example: AV-Comparatives, AV-TEST, SE Labs

Published benchmark facts, attributed and linked per test. These remain factual records rather than a fabricated headline score.

Sources we deliberately don’t use

Operator-supplied reviews

Reviews routed through a brand-owned channel cannot be verified for provenance.

Affiliate rankings & marketing copy

We read the genuine user reviews on affiliate and specialist sites, but never the rankings those sites are paid to show or the marketing around them — that output is downstream of someone’s commission.

Anonymous forums

Forums without account-age or post-history metadata collapse the signal-to-astroturf ratio below useful for the weight algorithm.

Editorial / journalism summaries

Already heavily summarised and slow to update. We aggregate raw user signal directly; secondary editorial layers re-introduce the biases we filter for.

How sources are weighted

Regulator data carries the highest single weight because it represents adjudicated outcomes — a complaint registered with the regulator survived both the operator’s own dispute process and the regulator’s intake filter. Genuine first-hand user reviews sit next, weighted by recency and by how reliably each source has tracked real outcomes over time. An independent traffic estimate corroborates them: sentiment that isn’t backed by a matching audience is trusted less.

General consumer-review platforms get the lowest weight — useful for cross-source validation but too easily flooded to anchor a score on their own. We publish the method, not a per-source scoreboard that could be reverse-engineered and gamed.

Read the full weight derivation in methodology §3 →

VPN review sources

For VPN brands we read genuine user reports from privacy-focused public forums alongside the source categories above. IndexFair has no affiliate relationship with any VPN provider; these reports are read for substance, not for any ranking. We weight them by recency and corroboration, and no single forum can move a score on its own.

Public forums

privacy-community reports

Unsolicited first-hand reports, including unresolved complaints. Promotional and referral-laden posts are filtered out before a report contributes to a score; uncorroborated sentiment is trusted less.

Methodology coverage

On each market ranking page we show a methodology coverage percentage. This tells you how much of the source set we plan to use for that market we currently have fresh data from — not a per-brand quality signal, a market-level honesty signal.

Current

The source has been refreshed within its scheduled window. Its signal feeds into every brand score in this market.

Lagging

The source has not been refreshed within its scheduled window — its data may be slightly stale until the next scheduled run.

Planned

We plan to include this source for this market but have not yet wired it up. It does not contribute to scores until it is current.

We show the source category (Review platforms, App stores, Public forums, Social media, Regulator complaints, Independent test labs) and freshness state — not individual source names or weights — so the score cannot be reverse-engineered and gamed.

Sources we plan to add

Planned additions
  • 01DE BZGA + GGL regulatory feeds · Phase 2
  • 02NL Kansspelautoriteit register · Phase 2
  • 03Additional app-store and social locales · Phase 2