RCU Document Screening Software: Check Every File, Not a Sample

RCU document screening software checks every document in a loan file for tampering and mismatches. A lender’s Risk Containment Unit is then not limited to the files it can sample.

An RCU Manager at an NBFC usually sees only a share of logins, and the rest move on trust. That gap is where forged payslips and altered bank statements reach disbursal.

Sampling made sense when each check needed a person and a field visit. It costs far more now, because advances drive India’s fraud losses and forged income papers are cheap to produce. KlearStack screens each file against your credit policy and sends exceptions to your team, while your credit officer keeps the final decision.

TL;DR

1. RCU means Risk Containment Unit; many Indian lenders use it alongside the term FCU.
2. Most RCU teams sample files, so unsampled files move ahead with no forensic check.
3. Advances fraud reached Rs 40,774 crore in FY26, across 8,640 reported cases.
4. A high hit rate proves the sample was well aimed, not that the rest of the book is clean.
5. The Unsampled 100 Test measures what sampling missed, using files already disbursed.
6. Software should screen every file; samplers and agencies should spend their time on the exceptions.
7. No tool should approve or reject a loan; the officer decides and the check stays logged.

Run your unsampled files through a free screening test

What Is RCU in Loans, and What Does It Actually Check?

RCU in loans means Risk Containment Unit, the team that screens loan documents for fraud before and after disbursal. Many lenders use FCU, or Fraud Control Unit, for the same work. A Head of Credit Operations relies on it to catch document manipulation that a credit analyst would not stop to test.

Lender job descriptions show how the work runs. Aditya Birla Capital roles track sampling and screening, Hunter checks, hit rate and fraud saved. Tata Capital describes checking loan documents through empanelled agencies to stop fraud at entry.

Those checks sit on top of a loan file that already holds KYC, income proof and property papers. Our guide to loan document compliance covers the full file. This page covers the screening step inside it.

Why Sampling Leaves the Biggest Gap in the Loan Book

RBI data for FY26 shows 8,640 advances fraud cases worth Rs 40,774 crore, up from 4,105 cases two years earlier. The amount includes older cases re-reported that year, so read the rise with care. The direction is still clear: advances fraud is where the money goes.

📊 Rs 40,774 crore of advances fraud in FY26

Card and digital fraud was Rs 29 crore in the same year. For an RCU Manager, the exposure sits in loan files, not in payment channels.

Source: RBI annual report data via Outlook Business

Audit leaders see the same risk. In an IIA survey, 65% named fabricated invoices or financial documents a top AI-enabled fraud threat.

A Rs 12.40 crore Versova case now under investigation alleges forged ITRs, bank statements and sale agreements in home loans. Altered pay stubs follow the same route into a file.

What we see in document-heavy lending teams is that samplers rarely miss what they open. The loss sits in the files nobody opened, because a tired sampler and a clean-looking PDF are a poor match for a volume target.

A High Hit Rate Does Not Mean the Book Is Clean

The assumption is that a rising hit rate shows the RCU is working. The reality is that hit rate only measures the files you chose to sample. It says nothing about the files you did not.

Think about what each outcome tells a Chief Risk Officer:

So a team can improve every number on its dashboard while the unsampled book gets worse. Screening each file changes the question to “which files failed a check, and who reviewed them?” A reviewed file is not a checked file, and a checked file proves nothing unless the check was logged.

That logged trail matters when a case turns up later. RBI’s fraud Directions require staff accountability to be examined in all fraud cases, so the record protects the people who signed. Our document forgery detection guide shows what those checks look for.

The Unsampled 100 Test: Find What Sampling Missed

Here is a test an RCU Manager can run without buying anything. It uses the files your process has already passed.

1. Pull 100 files disbursed 6 to 12 months ago that were never sampled.
2. Pick the same product and channel you sample most often.
3. Run each file through document-level checks for edits, mismatches and duplicates.
4. Send only the flagged files to your samplers for a normal review.
5. Count confirmed issues, then compare that rate with your sampled hit rate.

If the unsampled rate is close to your sampled rate, your sample is not finding the risk. It is only showing you a slice of it. The same reasoning sits behind continuous auditing, which moves audit teams from samples to the full population.

Here is an illustrative calculation, so use your own numbers. A lender with 10,000 monthly logins, 20% sampling and a 3% confirmed rate on the unsampled book has 240 problem files passing each month. If half are worth Rs 5 lakh each, that is Rs 6 crore a month of exposure with no check at all.

What Should RCU Screening Software Check on Every File?

Good RCU software checks completeness, consistency and tampering on each document, then shows the exact field that failed. These are the checks a Credit Operations Manager should test on real files.

• Completeness: every document your policy requires is present, valid and readable.
• Income agreement: payslip income is compared with bank credits and the application.
• Tampering: statements, payslips and bills are tested for edits. Document tampering detection explains the page-level view.
• Duplicates: a file already seen elsewhere in your book is flagged, even if renamed.
• Policy rules: your own thresholds run on every file, and the rule is visible on each flag.

KlearStack’s Consumer Loan KYC Fraud AI Agent runs more than 10 forensic checks per document and compares income across the file. For common signals, see how to spot a fake bank statement and what salary slip verification should catch.

See which of your unsampled files fail a check

Will Software Replace RCU Agencies? No, It Moves Their Time

Software does not replace RCU samplers or agencies, because some checks need a person, a phone call or a site visit. It moves their time from routine reading to the files that deserve attention.

KlearStack takes the drudgery; your team keeps the judgment and the final decision, and spends its time on the work that needs it. Concurrent auditors can use the same log, as our bank audit software guide explains.

The split usually looks like this for an RCU vendor manager:

The pattern across audit cycles is that an agency report answers “what did you find?” It rarely answers “what did you check?” Auditors ask the second question, so keep a record that answers it, whichever tool or agency you use.

A check layer is the wrong buy in some cases. It does not replace identity verification, bureau checks or Hunter-style database matching, so keep those. If you handle a few hundred files a month, focused manual sampling may be enough.

How to Start: One Product, Closed Files

Start with one product and one month of closed files, so no live decision is at risk. A Head of Credit Operations can then decide on evidence before any integration work begins.

1. Choose one product where forged income papers worry you most.
2. Share your document list and credit rules, with masked files if your policy needs it.
3. Review the flags with your RCU team, and mark which ones were real.
4. Tune the rules, then run new logins for that product alongside your existing sampling.
5. Expand only when your team trusts the flags.

KlearStack has processed 150M+ documents in total, supports 500+ document types and offers on-premise deployment for lenders who keep files inside their network. It reaches 95%+ straight-through processing within 90 days on suitable document flows.

Treat your own pilot as the proof, not these numbers. Older scanned files may need clean-up first, which our lending document guide covers.

Conclusion

Sampling tells you about the files you opened. Rs 40,774 crore of advances fraud in FY26 shows why the unopened files matter more. When one forged file reaches disbursal, the cost is the write-off, the provision and a staff accountability review for the person who signed.

Screen every file, send the exceptions to your people, and keep the record of each check.

Test KlearStack on files your sampling never opened

FAQ