Four stages. Full attribution. A defensible score.
The engine gathers consented alternative-data signals, weights them by their real predictive value against your book, produces a composite score with per-signal contribution attributed, and routes each application into a review tier your policy defines.
- Signal sources supported12+
- Standard weighting modelKreedite-01
- Score band0-100
- Default tier count3
Every score is composed - never conjured.
A Kreedite score is not a mystery number produced by an opaque model. It is a composition - each contributing signal enters with its own consent scope, its own weight, and its own audit trail. If your credit committee asks what a 71 means, the answer is not "the model said so." The answer is the four signals, their weights and their contributions - shown, not claimed.
- Per-signal contribution attributed on every score
- Weights tuned against your loan-book outcomes
- Recommendation tier, never an auto-decision
- Consent scope tracked per signal, per applicant
Five capabilities, one convergence.
What we read - and only where each source is consented and available.
How weighted signals combine into one composite score.
Every contribution attributed, every weight defensible.
Approve, review, decline-risk - configured to your policy.
Segment drift, tier migration, default correlation over time.