Signals a traditional file cannot see - read only where consented.
Kreedite reads four families of non-traditional signals. Each source is scoped to a specific consent, tracked per applicant, and revocable at any point. Coverage varies by market - the engine is explicit about which signals were and were not available for a given score.
- Signal families4
- Consent scopePer source
- Signal availability shownOn every score
Four signal families, four separate consent scopes.
A signal is only considered if the borrower has explicitly consented to that source. Missing sources are not silently penalized - they are simply reported as unavailable, and the score is composed only from the signals that were consented and readable.
Payment cadence, on-time behavior and continuity across electricity, gas or water bills held in the applicant's name - a strong proxy for consistent, non-luxury financial responsibility.
Prepaid recharge frequency, top-up amount stability and account activity duration - a meaningful behavior signal in markets where mobile is the primary financial rail.
Inflow regularity, cash-flow variance, spending-to-income ratio and seasonality on any account the borrower has consented to link - never used to infer specific merchant behavior.
Outcomes on group-lending or joint-liability products where such records exist and can be verified - highly predictive in microfinance contexts, treated as a first-class signal.
Coverage varies. Honesty about coverage does not.
Kreedite deploys per region against the sources actually available and legally accessible there. The engine states its coverage per-market rather than claiming universal reach.
Availability is highest in markets with digitized utility billing. In fragmented utility markets, signal density is lower and the engine reduces this source's weight in the composite.
Widely available across mobile-first economies. Requires opt-in via the operator or aggregator route. Not available where regulation prohibits mobile-data use in credit assessment.
Available only where a group-lending or joint-liability record exists and can be verified through the originating institution - deep predictive value in microfinance contexts.
- Every score explicitly lists which sources were read and which were unavailable.
- Weights adapt to the signal set actually present - no source is invented to fill a gap.
- Consent scope for each signal is stored alongside every scored decision for audit.
Bring an anonymized sample. We will walk through which signals we read, which we could not, and what the composite says - transparently, in one working session.