Data products
How opportunities are grouped, ranked on each of six real scorecard dimensions plus the composite, and diffed quarter over quarter into movers. Every step below runs exactly as written; nothing here is aspirational.
Composite cut
The full weighted composite score (all six dimensions), exactly as shown on every opportunity's public scorecard.
All six dimensions, weighted per the scoring methodology above.
Dimension cut 1
Median net hourly pay compared with the local minimum wage and the category median across the registry.
Weight: 25.0% of the composite.
Dimension cut 2
Total extraction rate: platform fees, withdrawal and instant-pay fees, plus unpaid onboarding time.
Weight: 20.0% of the composite.
Dimension cut 3
How often and how reliably earners are paid.
Weight: 15.0% of the composite.
Dimension cut 4
Task availability, deactivation rate, unpaid-work reports, and a human-reviewed appeals process.
Weight: 20.0% of the composite.
Dimension cut 5
Whether terms, fees, deactivation policy, pay disclosures, and an algorithmic-transparency statement are published.
Weight: 10.0% of the composite.
Dimension cut 6
Inverted — 100 means no signals found. Derived from risk flags, entity verification, and payment verification.
Weight: 10.0% of the composite.
Each opportunity is ranked within every (kind, region) group its own data actually supports: kind is the opportunity's real `kind` field; region is each of its real `geoCountries`, or the single group "GLOBAL" when it has none (fully remote, no country constraint). An opportunity available in more than one country is ranked separately within each — never averaged across them.
A "mover" is a real composite-score point change for the same opportunity between this period and the closest prior published period. An opportunity with no composite score in either period contributes no movers entry for that comparison rather than an invented delta; an opportunity newly meeting publication requirements, or no longer meeting them, is reported as "new"/"delisted" instead of a numeric delta.
This report reads only already-published registry data (`scorecards`, `pay_evidence`, `providers`, `opportunities`) — the same data a public scorecard page already shows. Passport-derived pay figures in that data were already suppressed at the source (a minimum of 20 distinct opted-in earners per cell) before ever reaching the registry; this report performs no additional aggregation of raw Passport data and therefore introduces no new suppression surface.
Version history