How the Credda score works
Your Credda Reliability Score reflects your track record as a professional, built from real commitments, verified outcomes, and platform data. Scores range from 0 to 100 and update deterministically as your record grows.
Your score comes from a few things about how you deliver work, each weighted by how well it predicts reliability, plus one small bonus, capped, for having at least one past job confirmed by a third party (education never counts).
The fewer commitments you have, the more cautious the score is; it settles as your record grows. The exact factors and weights in use right now are published live in the scoring model below. Scores are recalculated automatically whenever a commitment outcome changes, a platform syncs, or a dispute is resolved.
New users start at 20 (Provisional); trust is earned, not granted. Confirmed outcomes reach full confidence at roughly six. Unverified activity needs about ten before it lifts your score at all, and on its own it cannot carry you into the top bands. A verified failure still counts against you straight away.
Scoring factors
Completion Rate
Factor 1
UpFulfilling every commitment you log
DownCanceling or abandoning commitments
On-Time Rate
Factor 2
UpDelivering before or on the deadline
DownLate deliveries; the longer the delay, the greater the impact
Dispute History
Factor 3
UpMaintaining a clean, dispute-free record
DownCounterparty-filed disputes
Verification Depth
Factor 4
UpA larger share of your record independently confirmed by a counterparty
DownOutcomes that stay self-reported, with no third party to confirm them
These weights are not a description of the formula. They are the formula, read from the scoring engine when this page loaded. The endpoint is public and needs no key, so you can open it yourself and check this table against it: https://api.credda.io/api/v1/scoring/model
Score bands
Fairness & transparency policy
- No human, and no AI, decides your score.
- The score is a pure, deterministic function of your recorded commitments. No Credda employee can modify it, and AI is never in the decision loop. Every change is algorithmic and logged.
- Disputes are resolved deterministically.
- When a commitment outcome is contested, the result is decided solely from the tamper-proof commitment record by Credda’s deterministic dispute engine, not by a reviewer’s judgment. The same record always yields the same outcome.
- AI is advisory only.
- Credda’s AI, the Trust Coach and evidence summaries, explains your record in plain language and suggests what would move your score. It never sets, changes, or decides a score; that stays with the deterministic formula.
- Score history is immutable.
- All resolved commitments are permanently recorded. A score that drops can only recover through new fulfilled commitments; history cannot be deleted.
- We are cautious with a thin record, on purpose.
- A record with only a few events is held near the Provisional starting point until enough confirmed evidence exists, so a high score cannot be manufactured from a handful of events. The caution only works one way: a verified failure counts against you straight away rather than being averaged out.
Trust Model · Scoring formula published live · Deterministic disputes · AI advisory-only