Vasudha Distributors: Reading a Cluster of Control Failures at One Branch

Finding-by-finding mapping

Finding	Failed control objective	Read-across
34 payments in the ₹4.70–4.99 lakh band	Expenditure authorisation; limit design	Structuring to stay below the approval step
Two vendors sharing a bank account, created by the same person	Master-file change control; segregation of vendor creation from payment	Possible fictitious vendor or redirected payment
11 credit notes, all to 3 customers, all written off in 30 days	Segregation of authorisation, custody and recording	Classic lapping: receipts diverted, later masked by a credit note
Opens post, prepares collection sheet, posts receipts	Custody + recording + reconciliation in one pair of hands	No independent check exists between customer and ledger
Shrinkage 0.4% → 1.1% (≈ ₹45 lakh on branch sales)	Physical safeguard and dispatch controls	Possible collusion with dispatch or a weak control environment

Why this is not error. Random clerical error does not cluster: 34 payment amounts in a 29-paise-wide band 11 months into the year is a distribution no random process produces. Two "unrelated" vendors sharing a bank account is not a coincidence. Credit notes approved by the person who also controls receipts and writes them off within 30 days describe the standard concealment mechanism for diverted collections, and the ₹92 lakh is 1.4% of revenue at one branch.

Escalate first: the credit notes. It is live (receipts continue to arrive), the amount is already quantified, and the mechanism is testable within days by matching each credit note to whether the same customer paid later in cash. The account should be frozen and the three customers confirmed independently before the trail goes cold.

Why lowering the threshold fails. The 34 payments show that whoever can create a vendor, release a payment and conceal a receipt adapts to whatever limit exists; the limit simply moves the structuring band. A ₹2.5 lakh threshold doubles the approval workload across 14 branches, adds no independent check, and leaves the actual gap - no segregation between vendor creation, payment and reconciliation - untouched. Approvals are a detective device against error; they are weak against collusion.

Preventive controls (design):

1. Vendor master: creation and bank-detail changes moved to a central shared-service team; any bank change verified by an independent call-back to a previously recorded number and approved by two officers.
2. Segregation at branch: post opened under rotation or centrally scanned; the cashier may not post receipts; bank reconciliation performed by someone outside the branch.
3. Credit notes: reason codes, approval by regional finance head above a low value, monthly trend review by customer and by issuer.
4. Payments: vendor-level monthly cumulative caps; second signature for any new payee's first three payments.
5. Whistleblower channel with confidential escalation outside the branch line, plus a mandatory-rotation and leave policy so concealment cannot survive absence.

Detective controls and analytics:

1. Threshold-band test - histogram of payment amounts; flag any unexplained spike just below any approval limit.
2. Duplicate-entity screening - match vendor master on bank account, GSTIN, address, phone and creation user; run monthly.
3. Lapping test - match credit notes to subsequent re-receipt of cash from the same customer, and test whether DSO rises only at branches with high credit-note volumes.

Add quarterly surprise branch audits selected on risk score, and report a monthly control dashboard to the audit committee.

Next ten days

1. Preserve evidence before acting: secure the vendor master, payment files, credit-note register, email and the branch's system logs; take an image rather than allowing live edits.
2. Access and role: suspend the accountant's system access and move approval rights, on HR and legal advice, without destroying the ability to investigate.
3. Independent investigation: appoint a forensic reviewer reporting to the audit committee, not to the CFO, with scope covering the branch, the three customers, the 11 vendors and the related-party connection.
4. Audit committee, day 10: report the findings, the related-party relationship (the finance director must recuse himself from the discussion and from any decision on his nephew), the investigation scope, and the interim controls put in place.
5. Statutory auditors must be told before they opine. Concealment until after the audit is the aggravating factor, not the fix. Consider the company's reporting obligations for fraud by employees or officers, and legal/insurer notification.
6. Do not let the CFO's five-week timetable set scope. Closing the matter to meet an audit date is exactly the pressure that converts a control failure into a governance failure.

On the ₹18 lakh offer. Refuse it as a settlement. Restitution is mitigation, not absolution: it does not establish what was taken, it may be an attempt to stop the investigation, and accepting it privately could itself be treated as compounding or concealment. Record the offer, and let any recovery be negotiated by counsel after the quantum is established.

Limits of any system. Internal control gives reasonable, not absolute, assurance. Collusion between two or more people defeats segregation of duties; management override defeats approvals; cost-benefit limits how many controls a ₹640 crore distributor can sustain. That is why the whistleblower channel, rotation, surprise audit and an audit committee that asks its own questions carry more weight than another layer of signatures.