Trust grows through work.

How trust is calculated

Seller trust combines delivery outcomes (40%), timeliness (20%), verified buyer feedback (25%) and agreement conduct (15%). Buyer trust combines review responsiveness (60%) and conduct (40%). Missing metrics are excluded and remaining weights are normalized.

Only confirmed live payments from the last 365 days count. Test transactions, self-dealing and known shared organizations are excluded. Each counterparty contributes at most one outcome per calendar month. A numeric score needs at least five eligible outcomes, three counterparties and two metrics with three observations each.

Evidence loses half its weight every 90 days. Each metric starts with two positive and two negative prior observations, so a few successful jobs cannot create a perfect score. The observed percentages above show actual weighted evidence before that adjustment. Confidence describes evidence volume and coverage, separately from the score.

Open disputes do not lower conduct scores. Only recorded decisions do, and affected people can request a correction. An observed missed deadline may still affect delivery metrics. Wallet wealth, nationality, identity documents, followers and token ownership are not scoring inputs. Scores describe marketplace behavior, do not guarantee future performance, and never automatically exclude new users.

Payment proves that an agreement was funded; it does not prove two accounts are independent. These controls limit obvious repeat trading, not every form of collusion. Inspect the sample size, deliverables and counterparty history when assessing a service.