4.1 Credit bureaus
The four bureaus
Section titled “The four bureaus”| Bureau | URL | Coverage strengths |
|---|---|---|
| TransUnion CIBIL | cibil.com | Largest consumer file; strongest brand; consumer + commercial bureau |
| Experian Credit Information Co. of India | experian.in | Strong consumer; growing commercial |
| Equifax Credit Information Services | equifax.co.in | Consumer + commercial; strong inquiry / fraud feeds |
| CRIF High Mark | crifhighmark.com | Strongest MFI / MSME coverage; commercial bureau |
All four are RBI-licensed under CICRA. Lenders must be members of all four and submit monthly data to each (per CICRA + RBI directions).
What each bureau report contains
Section titled “What each bureau report contains”Consumer bureau (for individuals — proprietor, director)
Section titled “Consumer bureau (for individuals — proprietor, director)”- Borrower demographic — PAN, name, DoB, gender, mobile, email, addresses.
- Score (
300 – 900range typically; CIBIL TransUnion score widely used; other bureaus have similar scales). - Account summary — total accounts, active, overdue, closed.
- Per-account detail — lender, type, sanction date, sanction amount, current balance, DPD history (24–36 months), status (active / closed / settled / written off).
- Enquiry history — every enquiry by every lender in the past 24 months.
Commercial bureau (for entities)
Section titled “Commercial bureau (for entities)”- Entity demographic — name, PAN, GSTIN where available, addresses, directors.
- Score / risk indicator (vendor-specific).
- Total credit facilities — sanctioned + outstanding by lender.
- Per-facility details with DPD history.
- Director linkages — common across entities.
- Enquiries.
MSME / commercial reports (CRIF, CIBIL MSME, etc.)
Section titled “MSME / commercial reports (CRIF, CIBIL MSME, etc.)”- Same as commercial but with deeper MSME segment cuts, GST-derived turnover estimates (where bureau ingests GST), and MSME-specific risk indicators.
API flow (pull)
Section titled “API flow (pull)”- Lender submits enquiry with borrower’s identifiers (PAN minimum; mobile, DoB, address recommended).
- Bureau performs match (PAN is strongest; demographic fields supplement).
- Bureau returns report (usually XML or JSON).
- Lender parses, stores, uses for decision.
- Bureau records enquiry against borrower’s file.
API flow (submission)
Section titled “API flow (submission)”- Lender generates monthly file per CIC’s standard format from its loan book.
- Submits via SFTP / API.
- CIC processes; returns ack file with errors flagged.
- Lender corrects errors and re-submits where needed.
Pricing proxy
Section titled “Pricing proxy”| Action | Approximate cost (orientation; verify with vendor) |
|---|---|
| Consumer report pull | ₹15 – ₹50 per report (volume tiers) |
| Commercial report pull | ₹100 – ₹500 per report |
| MSME report pull | ₹100 – ₹500 per report |
| Soft pull / pre-eligibility | Lower; vendor-specific |
| Monthly submission | Usually no per-submission fee (cost is in membership) |
| Membership fee | Annual, varies |
Aggregators (Karza, Perfios) typically wrap all four bureaus behind a single API with consolidated billing.
Consent requirement
Section titled “Consent requirement”- Explicit borrower consent at application — language to specify each bureau and purpose.
- For soft pulls (pre-eligibility), some bureaus allow soft-pull APIs that don’t show as full enquiries on the borrower’s report.
Implementation complexity
Section titled “Implementation complexity”- Direct integration with one CIC:
~3 – 6 weeksengineering for pull + submission + parsing. - Via aggregator:
~1 – 2 weeksfor pulls; submission still on lender side. - All four CICs in parallel: cumulative effort if direct; near-zero incremental via aggregator.
Failure cases
Section titled “Failure cases”- No-hit — PAN doesn’t match any record (could be new-to-credit or PAN mismatch).
- Partial-hit — bureau matches but with low confidence; report quality lower.
- API timeout — bureau slow; queue + retry with idempotency.
- Submission file reject — schema / field-format errors; fix and resubmit.
- Score discrepancy between bureaus — common; policy must specify which to use or how to combine.
Fallback
Section titled “Fallback”- No-hit borrower — route to manual underwriting; use AA / GST / Tally as primary signals.
- API down — sandbox copy with cached version for in-flight applications.
- Multiple bureaus disagree — policy decision; document.
Compliance implications
Section titled “Compliance implications”- CICRA — mandatory membership and submission.
- Digital Lending Guidelines — explicit consent for bureau pull recorded.
- DPDP — bureau pull is processing personal data; basis is consent.
- Borrower dispute resolution —
30-daySLA.
Build vs buy
Section titled “Build vs buy”- Pull: Buy. Use one CIC direct + aggregator for the others, or aggregator-only.
- Submission: Build. The format is stable, the data is yours, and submission file generation rarely changes.
- Parsing: Buy (via vendor / aggregator’s structured output) or build (slower).