A CIBIL score of 0 does not mean a borrower is not creditworthy — it means the bureau has no data to generate a score from. In India, an estimated 190 million adults are credit-invisible: they have bank accounts, pay utility bills, and run small businesses, but have never taken a formal loan and therefore have no bureau score. The Bureau AI does not treat a thin file as a failed file — it reads the signals that exist and builds a credit picture from what is actually there.
What a thin file is — and what it is not
A thin-file borrower is not the same as a poor-credit borrower. A poor-credit borrower has bureau data that shows a negative history — DPD events, write-offs, settlements. A thin-file borrower has no bureau history at all, or a history so limited (one credit card opened 8 months ago) that the CIBIL scoring model cannot generate a reliable score. CIBIL returns these cases as either "NH" (No History) or "0" — both of which trigger automatic decline logic in most credit policy systems, regardless of the borrower's actual financial behaviour.
The Bureau AI treats "NH" and "0" scores as information gaps rather than credit failures. An information gap requires additional signals — not a decline. The AI activates the thin-file pathway, which gathers and weighs five categories of alternate signals that are predictive of repayment behaviour and are available through sources that do not require prior formal credit history.
A live thin-file profile: what the Bureau AI assembles
The five alternate signal categories and their predictive weight
The alternate signal table: what each source reveals and its limitation
| Signal Source | What It Reveals | Predictive Strength | Access Method | Limitation |
|---|---|---|---|---|
| Account Aggregator (AA) | 18-month bank behaviour: income regularity, balance trend, obligation payments, zero-balance events, NACH debits | Strong | Borrower consent via AA framework · RBI-regulated | Requires borrower consent; some banks not yet AA-connected |
| EPFO member passbook | Employment tenure, salary band, employer stability, contribution regularity — verified from government source | Strong | Borrower-shared EPFO UAN · Or DigiLocker pull with consent | Salaried employees only; informal sector excluded |
| Utility payment history | Electricity, water, LPG, piped gas payments — obligation fulfilment history outside formal credit | Moderate | Bank debit match to known utility providers · State utility APIs where available | Utility bills may be in landlord's name; not all utilities have digital payment trails |
| Telecom behaviour | Mobile plan tenure, bill payment regularity, plan tier — proxy for financial stability and commitment fulfilment | Moderate | Telecom bureau (TRAI-regulated) or consent-based operator data | Strong for postpaid subscribers; prepaid tells much less |
| GST returns (SE borrowers) | Business vintage, turnover trend, filing regularity — primary income signal for self-employed thin-file borrowers | Strong (for SE) | Borrower-shared GST credentials or GST system API | Only for GST-registered businesses; micro-enterprise below threshold excluded |
| Rental payment via bank | Consistent rental payments are a proxy for obligation management discipline in the absence of formal credit | Moderate | Bank statement pattern matching · Regular credits to same landlord account | Only detectable if rental is paid via bank transfer, not cash |
The thin-file lending conditions: what gets added at approval
A thin-file approval is not the same as a standard approval. The Bureau AI recommends conditions that build credit history while managing the incremental risk of lending to a borrower with no bureau track record. Typically: a reduced loan amount (50–70% of standard maximum eligibility, increasing after 12 months of on-time payments); NACH-mandated EMI collection (not standing instruction) to ensure payment discipline; mandatory credit bureau reporting from month 1 (so the borrower is building a tradeline record); and a 6-month portfolio review at which the Bureau AI re-pulls the bureau — by that point the borrower will have a thin CIBIL score based on the institution's own reporting, and the reduced amount restriction can be reviewed.
The credit-invisible borrower is not a risk — they are an opportunity that requires a different assessment method
The lender that refuses a thin-file borrower does not eliminate the credit risk — they transfer the borrower to an informal moneylender at 36% and lose a customer who may go on to take four more loans over the next decade. The lender that assesses Kavitha's 18 months of bank behaviour, her 24 months of EPFO contributions, and her 36 months of on-time utility payments — and approves her at a reduced ticket with conditions — acquires a customer at her first formal credit interaction. First-loan customers who perform well have the highest lifetime lending value in any portfolio. The Bureau AI makes it possible to identify the ones who will perform — using the signals that exist, not the signals that a bureau score requires.