Retail and Commercial Banking Operations Case Studies
Finance5 Case StudiesUpdated Sep 2026
Retail and commercial banking operations sit where customer activity meets the ledger. Deposits, wires, loan draws, and card settlements each leave records in several places, and those records must agree before anyone relies on them. The cases below come from branch networks, business lending units, and payment operations at fictional institutions. Each begins with a problem that reached management: a variance, a missed forecast, an alert, a credit request, or an audit report. Read them by asking what the records show, what exposure sits behind them, what the controls left as evidence, and where the decision threshold falls.
Case Study 1
Deposit ledger variance that blocked the monthly close
Marlow Ferry Savings Bank, a regional retail and small business bank with forty branches and a single payments operations center.
The situation
Marlow Ferry Savings Bank runs forty branches and a central operations center that posts about 2,300 deposit transactions each business day. At the close of the period, the personal checking subledger did not agree with its general ledger control account. The net difference was 12,460 dollars, small against total deposits but large enough to block sign off. Three records were in play: the core system's daily balance report, the general ledger detail, and the statement file that feeds customer statements. A junior analyst had marked the reconciliation complete three days earlier using a summary tie-out that compared totals only.
The challenge
The variance was not a single error. It combined timing items, a duplicate posting, and at least one reversal booked to the wrong period, and a modest net figure hid all of them. The controller wanted three answers before signing: how large the gross exposure was, whether any customer balance or statement had been misstated, and whether a plug entry could be justified. The team did not want to clear the difference with a plug entry until the gross items were understood.
Analysis
The first move was to stop reasoning from the net figure. The team rebuilt the reconciliation at item level, then grouped every open entry by cause. Timing items came out first, including deposits in transit and a weekend settlement file. What remained was 41 items with gross debits of 86,300 dollars and gross credits of 73,840, so the 12,460 difference was only the visible residue of much larger movements. Exposure followed the same logic. An overstated customer balance invites complaints and reimbursement, while an understated one can flow into fee and interest calculations that are hard to unwind after statements go out.
Teams grounded in Retail and Commercial Banking Operations treat a completed reconciliation as a claim that needs evidence, not a status flag. The summary tie-out proved only that two totals matched at one moment, and it left no record of which items had been examined. The control that should have caught this is a preparer and reviewer sign off attached to the item listing, with each reconciling item showing its age and owner. The team also set thresholds for the rebuilt process, with an unexplained item above 2,500 dollars or one older than two business days going straight to the controller.
- Close dayNet variance of 12,460 dollars blocks sign off.
- Review day oneItem detail pulled from core, ledger, and statement files.
- Review day twoDuplicate posting and two period errors documented with source references.
- Review day threePlug entry rejected; each item cleared to a named owner.
- Review day fourReviewer retested tie-out; closure evidence filed with controller.
Actions taken
- Rebuild the reconciliation at item level, matching each deposit subledger entry to a general ledger posting or a documented timing item. Assign every open item an owner and a clearing date.
- Reverse the duplicate posting through the approved correction route, with the original and reversal both retained, and confirm that no customer statement was produced from the erroneous balance before the correction.
- Set a written escalation rule: any unexplained item above 2,500 dollars, or any reconciling item older than two business days, goes to the controller the same day. Record the rule in the close procedure.
- Change reviewer sign off so that the tie-out cannot be approved without the item listing attached. The file should show who prepared it, who reviewed it, and the evidence each reviewer examined.
Results
The rebuilt reconciliation closed to a residual of 85 dollars, a documented timing item tied to a weekend settlement file. The duplicate posting was reversed through the approved correction route, and both entries remain in the audit trail with the approver's name. The controller signed the close after reading the item listing rather than the summary. Over the next quarter, two more items crossed the 2,500 dollar line and were escalated the day they were found. The open item count fell from 61 to 9 by the end of that quarter.
Questions for discussion
- Which of the three records would you trust first when they disagree, and what would make you change that order?
- What evidence should a reviewer see before accepting a tie-out, beyond the matching totals?
- How would you set the escalation threshold for a bank whose largest deposit items vary widely by branch?
Case Study 2
Forecast drift in commercial net interest income
Ostrander Commercial Bank, a business lender with about 1,200 commercial clients, a construction and hospitality loan book, and a treasury team that owns the funding plan.
The situation
Ostrander Commercial Bank built its annual plan on steady growth in net interest income, driven by a commercial loan book that was expanding each quarter. Two quarters in, income was running 6.8 percent behind plan. Most of the gap came from commercial deposits leaving faster than modeled and from funding costs that rose ahead of loan repricing. The treasury team had assumed a deposit runoff of 4 percent a year. Hospitality balances were running closer to 9 percent, while construction borrowers drew down committed lines more slowly than the model expected. The chief financial officer asked for a revised outlook before the next board meeting and a recommendation on whether to cut expenses or change deposit pricing.
The challenge
The gap had several causes, and each pointed toward a different response. Repricing deposits would protect balances but compress margin. Cutting expenses would protect earnings but could slow the lending pipeline the strategy depends on. The team also had to separate timing from structure, since a temporary gap supports a short-term fix while a permanent one changes the target the board should accept. Every option was defensible alone, so the recommendation had to show which assumptions would change the answer.
Analysis
The team started with the deposit ledger rather than the plan. Account-level balances over eight quarters showed uneven runoff: hospitality operating accounts moved most, while construction reserves held steady. From that base the team built three scenarios. The base case held current runoff for two quarters, then returned to the original curve. The stress case kept hospitality runoff near 9 percent for the full year. The recovery case assumed outflows stabilized within two quarters. Sensitivity mattered most. Each additional percentage point of commercial runoff removed roughly 1,150,000 dollars of annual net interest income, which made the deposit assumption the largest single driver of the forecast.
Good forecasting work tests the assumptions that drive the result before it tests the result. Here the runoff curve had never been compared with actual outflows by segment, and the model had no documented trigger for moving from one scenario to another. The fix is a written threshold: if hospitality runoff exceeds 7 percent for two consecutive reporting periods, the stress case becomes the planning case and the pricing committee meets within a week. Each assumption should rest on a reproducible extract from the deposit ledger, with segment definitions recorded beside it.
Actions taken
- Rebuild the deposit runoff assumption by segment and by balance band, using at least eight quarters of account-level movement, and document where the new curve departs from the original plan.
- Set scenario triggers in writing: a stated threshold for hospitality runoff, a stated threshold for construction draw rates, and the named owner who switches the planning case when either is crossed.
- Run the expense plan through a sensitivity test that shows which hiring and technology commitments can be paused without touching loan origination capacity. Keep the pipeline staffing list out of the cut list.
- Present the board with a range rather than a single revised target, showing the base, stress, and recovery outcomes with the decision points that would move the company from one to another.
Results
The revised outlook placed full-year net interest income about 4.1 percent below the original target with a bounded stress range. Hospitality runoff held near 7.6 percent for the next two reporting periods, which crossed the trigger and moved the planning case to stress. The pricing committee repriced the top tier of commercial deposits and accepted a narrower margin, while construction lending kept its prior pace. Technology projects were trimmed by a little under 900,000 dollars, and origination staffing was left untouched. The board accepted the range, and the treasury team now reports trigger status in every monthly pack.
Questions for discussion
- Which assumption in the original plan deserved a test before the first quarter closed, and what evidence would have exposed it?
- At what point would you stop treating a deposit outflow as timing and start treating it as structural?
- Should a forecast owner be allowed to change the planning case alone, or should the trigger require committee approval?
Case Study 3
Cash deposits and wires that outran a client profile
Thornwick Community Trust, a state chartered bank serving small businesses in three metro areas, with a financial crimes unit of nine analysts.
The situation
Thornwick Community Trust had banked a landscaping and equipment rental company for six years, with modest activity. In one period, cash deposits rose to 38,500 dollars in a single week. The money was split across eleven branches within three days, with each deposit kept under 3,000 dollars. Outgoing wires followed to two counterparties that the bank had never seen in this relationship, both in a free trade zone. The monitoring system generated two alerts on the same account, one for deposit structuring and one for velocity. The relationship manager's note called the cash seasonal, and annual revenue on file supported higher volume, though not this much cash.
The challenge
The pattern was unusual enough to escalate, but the harder question was what it meant. The team had to decide whether the client's business explained the activity, whether the file contained support for that explanation, whether the controls had worked as designed, and whether the facts met the bank's standard for filing a report. An error in either direction carries cost. A missed report creates regulatory exposure, while an unjustified one damages a legitimate customer and invites scrutiny of the bank's judgment.
Analysis
The file was rebuilt from the account's history: opening documents, the expected activity profile set at onboarding, and a year of cash transactions. Average weekly cash deposits over that year had been about 2,100 dollars, so the recent week ran at roughly eighteen times that level. The wires totaled 214,000 dollars across two counterparties registered in the month before the first transfer. Exposure ran three ways: illicit proceeds could move through the account, the client could be defrauded by the counterparties, and the bank's handling of the alert would be examined later. Deposits sat just under the level that draws routine attention.
A good file separates an alert from a case. An alert asks for a look, while a case needs documented evidence that the activity has an explanation or does not. Here the relationship manager's seasonal note had been accepted without invoices, delivery records, or contracts, and the cash procedure did not require them for a spike of this size. When support is missing and the counterparties cannot be verified, the matter goes to the compliance officer with the full timeline. Control evidence matters at each step, since a file closed on a verbal explanation is weak protection for the bank.
- Detect Velocity and structuring alerts fire on the same relationship.
- Document Gather invoices, contracts, and the expected activity profile on file.
- Decide Compliance officer weighs explanation against the evidence available.
- Escalate or close Report within the required window, or record why no report was filed.
Actions taken
- Pull the full relationship history, including account opening file, prior cash volumes, and the expected activity profile, and compare each week against it so the deviation is measured rather than described.
- Request invoices, delivery tickets, and counterparty contracts from the client, and record the request date, the response, and any gap. Do not accept a verbal explanation as the file's only support.
- Put the two counterparties through a screening and ownership check, and confirm whether either appeared in prior alerts across the bank's own portfolio before the wires are released again.
- Write the decision memo with the reasoning, the evidence weighed, and the reason for the chosen outcome, then have a second officer review it before closing, whichever direction the decision takes.
Results
The review found a partial explanation. The client's landscaping business did buy equipment seasonally, and two invoices supported roughly a third of the cash. The remaining deposits had no support, and one of the wire counterparties did not answer the verification request. The compliance officer filed the required report within the window and restricted outgoing wires on the account pending further review. The relationship manager's file now records that the seasonal note was accepted without support. The cash procedure also requires documents for any spike above the client's profile before activity is cleared.
Questions for discussion
- Which document would you have requested first at intake, and why does its absence matter more than a verbal explanation?
- How should a relationship manager's explanation be weighed when the client has not yet produced support?
- What would have to be true for a pattern like this to be closed without a report?
Case Study 4
Expansion loan request for a packaging manufacturer
Varnell Folding Carton Company, a family-owned manufacturer with two plants, about 310 employees, and sales concentrated in two customers.
The situation
Varnell Folding Carton Company asked for a 7,500,000 dollar seven-year term loan to finance a third production line. Revenue had grown steadily over three years, but two customers accounted for about 58 percent of sales. The owners offered the existing equipment and the new line as collateral, plus a personal guarantee from the majority owner. Management accounts showed EBITDA of 3,100,000 dollars in the last full year, slightly above the year before because of a large one-time order. The new line was projected to add about 1,400,000 dollars of annual EBITDA once it reached target volume, roughly eighteen months after startup.
The challenge
The committee had to judge whether the projections were credible enough to lend against, and whether the company could survive a major customer cutting orders. The statements were reviewed rather than audited, and the line economics rested on a single sales assumption supplied by the owners. Approval would also set a precedent for how the bank handled customer concentration across its manufacturing borrowers.
Analysis
The starting point was debt service. At the proposed rate, annual payments on 7,500,000 dollars over seven years come to about 1,390,000 dollars, which puts coverage on last year's EBITDA at roughly 2.2 times. That looks comfortable until the concentration is considered. In the downside case the largest customer cuts orders by 40 percent in year one, removing about 900,000 dollars of contribution and dropping coverage to about 1.6 times. A second stress delays the new line by two quarters and pushes coverage toward 1.3 times, close to the floor the bank applies to this profile.
Approval conditions belong to the credit itself. Owner-prepared projections need independent support before they carry weight, so the committee should require a tie-out of the last three years of management accounts to tax returns and bank statements, with any gap above a set materiality level explained in writing. The facility has to fit the company's cash cycle, and the owner's guarantee should be sized to his verified liquidity, not his stated net worth. Set the thresholds before the meeting: 1.5 times minimum downside coverage, a loan-to-value ceiling on the appraisal, and a concentration review above half of sales.
Actions taken
- Obtain the last three years of tax returns and bank statements and tie them to the management accounts, documenting every difference above a set materiality level before the credit memo is drafted.
- Rebuild the new-line projection from capacity, price, and customer order data, and run a case where the top customer reduces orders by 40 percent in the first year.
- Set the approval conditions in writing: a minimum downside coverage of 1.5 times, a loan-to-value ceiling of 70 percent on the equipment appraisal, and a customer concentration covenant reviewed at 50 percent.
- Size the personal guarantee to the owner's verified liquidity, using statements he provides and a bank review of them, and state in the memo what the guarantee covers if the line is not repaid.
Results
The committee approved seventy percent of the request, or 5,250,000 dollars, with the third line funded in two draws. The tie-out found one year of management accounts overstated EBITDA by about 240,000 dollars through inventory timing, slightly reducing base case coverage. The equipment appraisal supported the smaller facility, and the guarantee was sized to the owner's reviewed liquidity. Under the downside case the request sat in the secured but thin quadrant, justifying tighter terms over a decline. Two quarters later the largest customer cut orders, and the concentration review became a planned conversation.
Questions for discussion
- What evidence would you need before accepting an owner-prepared projection for a new production line?
- Is a phased facility a sound compromise, or does it simply move the repayment risk into the company's planning?
- How would you set a concentration trigger when the largest customer is also the most stable one?
Case Study 5
Payments operations audit findings and closure tests
Calloway Ridge Bank, a regional bank with forty-two branches and a central payments operations unit serving them.
The situation
Internal audit reviewed the payments operations unit for the prior four quarters. The unit handles incoming wires, outgoing ACH batches, and card chargebacks, with about 46,000 outgoing payments in a typical quarter. The sample covered 1,800 items. Testing found that some wire releases were approved by the same person who keyed them, that chargeback reserves were not recalculated when case outcomes changed, and that user access reviews had been signed without checking terminated employees against active accounts. The unit manager disputed one finding, but the draft remediation plan had no named owners and no deadlines.
The challenge
The audit committee needed a ranked list it could act on, not a long catalog of observations. Some findings were design gaps, others were execution failures, and a few overlapped. The team had to decide which items were material to the financial statements or to customer harm, which were housekeeping, and what evidence would show that each fix worked. Closure could not depend on someone reporting that the work was done.
Analysis
Ranking started with monetary exposure, adjusted for likelihood and customer impact. Self-approved wire releases came first. Nine sampled releases lacked independent approval, totaling 412,000 dollars, and a single unauthorized wire could exceed the unit's daily limit. Chargeback reserves came second: thirty-seven changed case outcomes had left reserves unadjusted, a net understatement of 218,000 dollars that flowed straight into the quarterly results. Stale access ranked third because fourteen terminated employees still held active rights, a control weakness with no sampled loss yet. Documentation gaps and an outdated procedure came last, important for evidence but small in exposure.
Closure tests should be written before the fix is designed. For the wire finding, closure means a system report showing zero self-approved releases across a full quarter, plus a rerun of the sampled batches with keyer and approver fields checked. For reserves, the test is a monthly recalculation tied to case status, with any variance above a set threshold explained before the close. Owners should be named by role and person, and the executive sponsor signs closure, not the team that performed the fix. Until the test passes, the finding stays open, whatever the status report says.
Actions taken
- Dana, the wire operations supervisor, enforces system-level segregation so that the person who keys a wire cannot release it, and exports a release report with keyer and approver fields for each batch.
- Ravi, the finance controller, owns the reserve fix: recalculate chargeback reserves monthly from case status, post true-ups through the close, and reconcile the reserve ledger to the case register.
- Marisol, the access administrator, removes terminated users within one business day of notice and runs a full entitlement comparison against the HR leaver list before each quarterly review is signed.
- Internal audit reruns the matching sample after each fix, and the executive sponsor, Tomas, the head of payments operations, signs closure only when the test passes.
Results
Wire segregation went live in the second month after the report, and the first quarterly release report showed no self-approved batches. The reserve recalculation exposed a further 61,000 dollars of understatement on its first run, which was booked through the close. Stale entitlements were removed, and the rerun of the access review found no terminated users with active rights. The procedure document was rewritten to show the new approval step, which resolved the manager's objection. One documentation item stayed open two periods longer, by agreement, and closed once its test passed.
Questions for discussion
- Which finding would you have ranked differently, and what exposure figure would change your order?
- What should an executive sponsor require before signing closure on a finding that was fixed but not yet tested?
- When is it reasonable for the team that performs a fix to also keep ownership of the finding?
Learning Resources
Finance CASE STUDIES
China, India, United States, Indonesia, Pakistan, Nigeria, Brazil, Bangladesh, Russia, Japan, Mexico, Philippines, Vietnam, Egypt, Turkey, Germany, United Kingdom, France, Italy, Canada, South Korea, Australia, Spain.