ATM Reconciliation Example: Finding a $500 Variance

ATM Reconciliation Example: Finding a $500 Variance

A useful ATM reconciliation example starts with a familiar morning problem: the cash count from a terminal does not agree with the amount the host, switch, or cash ledger says should be in the machine. The difference may be only $500, but the operational question is larger. Was cash physically dispensed without a matching transaction? Was a valid customer transaction reversed? Or did the operation compare files built to different business-day cutoffs?

Reconciliation is the control process that turns those questions into an evidence trail. Done well, it confirms that physical cash movement, ATM transaction records, network settlement, and accounting entries describe the same activity. Done poorly, it can create avoidable claims exposure, inaccurate cash positions, and repeated field dispatches for issues that began in a file-timing mismatch.

The operating picture behind an ATM reconciliation example

Consider a financial institution that uses an armored carrier for cash replenishment and a managed services provider for first-line maintenance. The ATM has four cassettes, each loaded with $20 notes. At the start of the business day, the terminal contains $40,000.

During the day, the ATM switch records 1,225 approved withdrawals totaling $24,500. It also records 14 approved cash withdrawals that were later reversed because the dispenser reported a partial-dispense or no-dispense condition. Those reversals total $500. The electronic journal, or EJ, shows the same 1,225 initial approvals, but also includes dispenser status records that must be read alongside the financial messages.

The basic expected-cash calculation is:

Opening cash + cash loaded – net dispensed cash = expected closing cash

In this case, no replenishment occurred during the day. Net dispensed cash is $24,000, calculated as $24,500 in approved withdrawals less $500 in reversals. Expected closing cash is therefore $16,000.

When the cash-in-transit crew services the ATM the next morning, however, it counts $15,500. The terminal is short by $500 against the expected balance.

That result does not yet prove a loss. It establishes a variance that must be classified and supported. The next step is to determine whether the $500 is tied to a specific transaction, a dispenser event, a settlement timing issue, or a counting and documentation error.

Reconciling the four records that matter

An effective investigation compares records with different purposes. No one source is sufficient on its own.

The cash ledger or cash forecast file establishes what should be physically present based on the opening balance, documented loads, documented removals, and accepted transaction activity. The ATM switch log identifies authorization decisions, transaction amounts, reversals, and financial completion messages. The EJ provides device-level evidence, including cassette movement, reject-bin activity, shutter events, and dispenser error codes. Settlement and network reports confirm what was posted or presented for clearing, which may follow a different timing cycle than the terminal business day.

For the $500 variance, the operations analyst first verifies that all four sources use the same terminal ID, currency, and business-date convention. This sounds basic, but terminal identifiers can change after a processor migration, and files may use local terminal time, Eastern Time, or UTC. A transaction occurring at 11:58 p.m. local time can easily land in different reporting days across the ATM application, switch, and card-network settlement system.

The analyst then isolates the 14 reversed transactions totaling $500. Thirteen show a dispenser error in the EJ and a corresponding reversal in the switch. One $500 transaction shows an approved withdrawal message and an EJ record showing notes presented, but no clear confirmation that the customer removed the notes. There is no reversal in the switch.

That transaction is the leading candidate. If five $20 notes were presented and not retracted, physical cash should be $500 lower than the ledger calculation that assumed a reversal. If the notes were retracted, the reject-bin count and cassette totals should reflect that outcome. The service record shows no unusual reject-bin volume, strengthening the case that the customer received the cash.

Why settlement alone cannot close the case

A common mistake is to treat settled transactions as the final truth for a cash variance. Settlement confirms the financial side of an interchange or on-us transaction, but it does not independently confirm note movement. A customer may have received cash while a reversal was sent in error, or may have been charged correctly despite a device condition that later requires a dispute adjustment.

Likewise, a cash difference does not always map cleanly to a single customer transaction. A cassette may have been loaded with the wrong denomination, a pack may have been miscounted, or a service technician may have documented a residual cash removal incorrectly. The evidence has to be reconciled in sequence rather than forced to fit the first plausible explanation.

The investigation and correction path

For this example, the analyst creates a case record that preserves the original counts and source files. Adjusting a ledger before recording the variance can erase the audit trail and make repeat-event analysis harder.

The review proceeds as follows:

  1. Confirm the opening balance against the prior service receipt and verify that the $40,000 load consisted of 2,000 $20 notes.
  2. Match every withdrawal, reversal, and cash-related EJ event to the terminal business day, including transactions near the cutoff.
  3. Compare the counted cassette residuals with expected note counts, then verify reject-bin and purge-bin contents where applicable.
  4. Review the questionable $500 transaction’s sequence numbers, dispenser diagnostics, image or video availability under policy, and any customer dispute or complaint.
  5. Determine whether the financial record requires correction, then route the adjustment through the institution’s established settlement, general-ledger, and dispute procedures.

In the scenario, the switch has retained the original $500 approval without a reversal, while the local reconciliation worksheet incorrectly netted the event as reversed after reading a generic dispenser status entry. The physical count is therefore correct. The accounting worksheet, not the terminal, created the variance.

The correction is to restore the $500 to net dispensed cash, reducing expected closing cash from $16,000 to $15,500. The case closes with no physical shortage. More importantly, the operations team identifies a control gap: the reconciliation logic treated a device alert as equivalent to a completed financial reversal.

What this example reveals about control design

The most durable ATM reconciliation controls distinguish between a device event and a financial event. A dispenser fault, retract, partial dispense, reversal request, reversal response, and settlement adjustment are related events, but they are not interchangeable. The system logic and exception workflow should preserve that distinction.

Exception thresholds also need context. A $20 variance may warrant an automated review at a high-volume retail ATM, while a $500 discrepancy at a low-volume branch terminal may justify immediate review. The right threshold depends on denomination mix, transaction volume, cash servicing frequency, contract responsibilities, and the cost of a field visit.

Ownership must be explicit across the chain. The financial institution may own the customer adjustment decision, the processor may control switch records, the managed service provider may access EJ diagnostics, and the armored carrier may hold the cash count documentation. A reconciliation process that depends on email requests between those parties will be slow precisely when evidence is most time-sensitive.

Automation can reduce manual matching work, particularly when transaction identifiers and terminal sequence numbers are consistently retained. But automation does not remove the need for operational judgment. A rule that automatically labels every reversal message as resolved can hide a repeated dispenser issue or a message-ordering defect.

Practical checks before calling an ATM short

Before recording a terminal as cash short or cash over, confirm the business-day cutoff, denomination configuration, documented cash adds and pulls, reject-bin treatment, and whether incomplete transactions have reached their final reversal status. These checks are particularly valuable after software releases, processor changes, daylight-saving transitions, or a shift in cash-in-transit schedules.

Trend the exceptions as well. Repeated differences at the same terminal may point to worn dispenser components, a cassette sensor issue, a recurring configuration error, or a servicing process that does not capture residual cash consistently. One resolved variance matters less than a pattern that reveals a control or equipment problem.

The practical value of reconciliation is not simply that balances eventually match. It is that each mismatch produces enough reliable evidence to protect customers, account for cash, and direct field attention toward the problem that actually occurred.

ATM Reconciliation Example: Finding a $500 Variance

Cash Recyclers Versus Smart Safes Compared