KYC Audit for Loan Files: Sampling, Exceptions and Template
Last reviewed: November 2026
A KYC audit for loan files is a post-disbursement review of sampled loan files that tests whether each borrower's Know Your Customer (KYC) record is complete, valid, consistent and correctly reported. It catches the gaps that slipped past login: a missing photograph, an expired passport, an address that differs from the loan agreement, or a record never uploaded to the registry.
This guide is for chief compliance officers and heads of credit operations at non-banking financial companies (NBFCs), housing finance companies (HFCs) and banks. It answers the questions audit teams ask most, from Reserve Bank of India (RBI) rules and sample sizes to exception codes, closure and a worksheet you can copy.
Quick Answer
- What it tests: whether each sampled borrower's KYC record meets seven rules in the RBI's NBFC KYC Directions, 2025, from identity documents to registry upload and record retention.
- How many files: about 185 files give 95% confidence for a 2,000-file population, with high-risk files usually checked in full.
- What it records: every exception, coded as missing, expired, mismatched or process, so trends reach the board audit committee.
What Is a KYC Audit for Loan Files?
A KYC audit for loan files re-checks finished files after disbursement, unlike login checks that stop forged documents at the door. Forgery at onboarding is covered in our guide to forged KYC documents in loan applications; this audit asks whether the file that was booked meets the rules today.
How Is a KYC Audit Different From a Concurrent Audit?
A KYC audit tests one subject across a sample, while concurrent audit for NBFCs reviews operations broadly and continuously. The two connect: paragraph 12(4) of the RBI's NBFC KYC Directions, 2025 requires a "Concurrent / internal audit system to verify compliance with KYC / AML policies and procedures", where AML means anti-money laundering.
Which RBI KYC Rules Should a Loan File Audit Test?
A loan file audit should test seven rules from the Reserve Bank of India (Non-Banking Financial Companies – Know Your Customer) Directions, 2025 (RBI/DOR/2025-26/361, 28 November 2025, updated 29 December 2025).
| Requirement | Para | What the Auditor Checks |
|---|---|---|
| Officially valid document (OVD) list | 5(1)(xiv) | Passport, driving licence, proof of possession of Aadhaar number, voter identity card, job card under the National Rural Employment Guarantee Act (NREGA) or National Population Register letter |
| Deemed OVD for address | 5(1)(xiv)(b)–(c) | Utility bill under two months old or similar; current-address OVD collected within three months |
| Permanent Account Number (PAN) or Form 60 | 23(5) | PAN, its e-document or Form 60 on file |
| Risk categorisation | 20, 41(1) | Low, medium or high category recorded; reviewed at least every six months |
| Periodic updation | 42 | High risk every 2 years, medium every 8, low every 10 |
| Central KYC Records Registry (CKYCR) | 63 | Upload within 10 days; KYC identifier communicated; updates within seven days |
| Record retention | 47 | Identity records kept five years after the relationship ends |
What Does a KYC Lapse Cost a Lender at RBI Inspection?
KYC gaps that reach an RBI inspection turn into monetary penalties and public press releases. On 17 July 2026 the RBI announced a ₹2.70 lakh penalty on Muthoot Vehicle and Asset Finance Limited because it "failed to put in place a system of periodic review of risk categorisation of accounts".
The bigger cost is remediation. In this illustrative case, an audit finds a 7% exception rate in a 40,000-loan book, which is 2,800 files to recall, re-collect and re-upload at about 20 minutes each: roughly 930 staff hours.
How Many Loan Files Should a KYC Audit Sample?
Sample enough files to estimate the exception rate within a set margin, using the standard sample size formula with a finite population correction.
n₀ = Z² × p × (1 − p) ÷ e²
n = n₀ ÷ (1 + (n₀ − 1) ÷ N)
Here Z is 1.96 for 95% confidence, p is the expected exception rate, e is the margin of error and N is the number of files disbursed in the audit period. With p = 5% and e = 3%, n₀ = 3.8416 × 0.05 × 0.95 ÷ 0.0009 = about 203 files.
| Files Disbursed in Period (N) | Files to Sample (95% Confidence, 5% Expected Rate, ±3%) |
|---|---|
| 500 | 145 |
| 1,000 | 169 |
| 2,000 | 185 |
| 5,000 | 195 |
| 10,000 | 199 |
Should High-Risk Files Be Sampled Differently?
High-risk files should be sampled separately, usually at 100%, because their periodic updation cycle is two years and their exceptions matter most. An illustrative stratified plan: all high-risk files, the table sample for medium and low risk, and every file sourced by a channel with past fraud findings from your risk containment unit sampling.
Are your auditors still reading KYC scans line by line?
See DocuEye Extract KYC Fields From Sampled Loan Files
Run DocuEye on a batch of your own audit samples, or talk to us about your KYC audit cycle.
Which KYC Exceptions Should Auditors Record?
Record each KYC exception under one of four categories: missing, expired, mismatched and process. A shared code list makes branch and product comparisons possible.
| Code | Exception | Example in a Loan File |
|---|---|---|
| M1 | OVD copy missing | Only the application form carries the address |
| M2 | PAN or Form 60 missing | PAN field blank in the loan system |
| M3 | Photograph or signature missing | Unsigned KYC form |
| M4 | CKYCR upload missing | No KYC identifier against the customer |
| E1 | OVD expired at onboarding | Driving licence past validity on sanction date |
| E2 | Periodic updation overdue | High-risk borrower, last update 30 months ago |
| E3 | Deemed OVD not followed up | Utility bill accepted, no current-address OVD after three months |
| X1 | Name mismatch | OVD, PAN and loan agreement spell the name differently |
| X2 | Date of birth mismatch | PAN and OVD show different years |
| X3 | Address mismatch | OVD address differs from the agreement and the field verification |
| P1 | Risk category not reviewed in six months | Category unchanged since onboarding |
| P2 | Masking failure | Full Aadhaar number visible on a stored copy |
Exception rate = files with at least one exception ÷ files sampled × 100. In an illustrative audit of 185 files with 14 exception files, the rate is 7.6%, which projects to about 151 files across a 2,000-file population.
How Should Each KYC Exception Category Be Closed?
Each category needs its own closure route, owner and deadline, or the same findings return next quarter. These illustrative routes double as a closure checklist:
- Missing (M1–M4): re-collect the document or upload the record to CKYCR; branch operations; 30 days.
- Expired (E1–E3): run periodic updation and collect a current OVD; credit operations; 60 days.
- Mismatched (X1–X3): verify with the borrower, then correct the loan system or treat as possible fraud; credit operations and fraud control; 30 days.
- Process (P1–P2): fix the control, then re-test a fresh sample; compliance; next audit cycle.
Where Can Auditors Verify KYC Details Online?
Three official portals cover most look-ups, and all were confirmed loading on 19 November 2026.
- Income Tax e-filing: Verify Your PAN: whether a PAN is active and matches the name and date of birth entered.
- Election Commission electoral search: voter details by Electors Photo Identity Card (EPIC) number.
- CKYCR portal of the Central Registry of Securitisation Asset Reconstruction and Security Interest of India (CERSAI): registry access for reporting entities to search, download and upload records.
Driving licence checks are covered in our guide to driving licence verification for lenders, and registry workflows in CKYC registry integration.
Are Closed Loan Files Still in Scope for a KYC Audit?
Yes. Under paragraph 47, identification records must be kept for at least five years after the business relationship ends, and transaction records for at least five years from the date of transaction.
How Does DocuEye Speed Up a KYC Audit?
DocuEye speeds up a KYC audit by turning scanned OVDs, PAN cards and KYC forms into structured fields that rules can compare with the loan record. For audit sampling, DocuEye brings artificial intelligence (AI) and optical character recognition (OCR) together with a logic-based extraction engine that covers 50+ document types.
Without it, an auditor opens each scan, skips handwritten KYC forms in older files and eye-checks every field against the loan system. Findings go into a spreadsheet typed by hand, and the same errors return each quarter.
With DocuEye, smart field detection extracts name, date of birth, number, address and expiry, while handwriting recognition and multi-language support bring older files back into the sample. Built-in validation rules flag mismatches and expiry, structured exportable output feeds the worksheet, and continuous learning on your templates keeps improving extraction.
Portal lookups, sampling decisions and the final opinion stay with the audit team.
What Should a KYC Audit Worksheet Include?
A KYC audit worksheet should include fourteen columns so every finding is traceable to a file, a rule and an owner. Copy this layout into your audit workbook.
- Loan account number
- Branch and sourcing channel
- Disbursement date
- Risk category and last review date
- OVD type and number (masked)
- OVD expiry date
- PAN or Form 60 status
- Name, date of birth and address match (yes or no for each)
- Last KYC update date and next due date
- CKYCR identifier and upload date
- Exception codes
- Evidence reference (page number in the scanned file)
- Owner and target closure date
- Closure status and re-test result
Feed open exceptions into post disbursement document tracking, correct loan management system (LMS) data errors through customer master data cleanup in LMS, and keep masked copies as described in Aadhaar masking compliance for lenders. Summaries belong in your RBI inspection readiness pack.
Want KYC exceptions found by your auditors before an inspector finds them?
Book a DocuEye Demo for KYC File Audits
Send us ten anonymised sample files, or visit Digiqt to schedule a working session.



