The cash path behind the balance

In September 2022, treasury identified a week in which several large customers' expected payments were clustered shortly before our own facility obligations. The monthly forecast remained comfortably positive. The daily view was less reassuring: a modest delay in two collection groups would leave little room for advances already discussed with sellers. Our commercial director argued that we should not pay additional fees for a problem that had not occurred. His concern was reasonable because the annual expense plan was already tight.

I asked treasury to compare three choices: retain ¥1.8 billion of additional undrawn capacity for four months, defer a defined group of proposed advances, or accept the narrower runway with a documented escalation plan. We used observed payment timing rather than assuming every customer would pay late. We also checked that the retained facility could support the assets likely to need funding. Nominal capacity with the wrong conditions would have given us reassurance without solving the practical problem.

I recommended keeping the capacity. The additional commitment fees were approximately ¥1.8 million over the four-month period. We deferred a small finance reporting enhancement to accommodate that cost and explained the trade-off to the board. One collection group subsequently arrived three working days later than expected, and we drew ¥620 million before releasing it after the receipts were identified. This was a timing event, not evidence of a new credit loss.

The unused portion of the facility still cost money, so I do not describe the whole decision as a saving. It bought a bounded ability to meet commitments while information caught up. In hindsight, I would have brought the seller-advance schedule into the discussion earlier. Commercial colleagues could have helped shape the options before the issue appeared to be a choice between their growth plan and treasury's caution.

During a 2023 acquisition review, the proposed receivables perimeter was ¥8.4 billion. The seller's summary records reconciled at aggregate level, and our relationship team saw an opportunity to enter a promising supplier segment. A transaction sample told a more complicated story. Several customer groups used deductions that were recorded as adjustments only after settlement. The files did not consistently show who had accepted those deductions or how unresolved differences would be handled after transfer.

The acquisition sponsor wanted to complete the original perimeter and repair the records during integration. I understood the argument: a narrower purchase would reduce expected income and might make some seller relationships less attractive. I nevertheless asked for a transaction-by-transaction explanation of the balances we would need to fund immediately. Our legal and credit teams conducted their own reviews; my responsibility was to test the funding and reporting assumptions, not to substitute for their conclusions.

I recommended a ¥5.9 billion initial perimeter with a separately negotiated route to consider additional assets once their records met the agreed requirements. The approved business case removed roughly ¥42 million of projected first-year contribution and retained an integration team for longer than the original plan. The seller accepted the smaller transaction after difficult negotiations. We did not count excluded balances as future purchases in the approved forecast.

Nine months after completion, the acquired book's collections were being reconciled through our normal process, but not all excluded relationships had returned. That lost opportunity belongs in the outcome. The decision gave us a perimeter we could fund and report with understood evidence; it did not prove the rejected assets were inherently poor. My correction for a future acquisition would be to sample deduction practices before commercial discussions harden around a headline portfolio size.

Our 2024 month-end review found 146 receipts older than five working days awaiting complete allocation information. Some were small. A few involved large remittances spanning many invoices. Operations proposed a broader tolerance for automatic allocation, expecting it to reduce the queue and release experienced staff for new business. The controller worried that this would obscure disputed deductions. Both teams were trying to solve a genuine workload problem, but they were using the same queue count to describe different forms of uncertainty.

I asked them to divide the items by the evidence missing: payment reference, agreed deduction, seller instruction or a mismatch requiring investigation. We then followed 30 items through the proposed treatment. Several would have disappeared from the exception report while leaving a balance whose meaning was no clearer. I declined the broad tolerance change and approved a more limited matching rule for records with exact invoice references. The remaining categories retained named owners and explicit follow-up dates.

We assigned two experienced staff to the difficult cases for twelve weeks, delaying a planned transfer of those employees to onboarding work. The queue older than five days fell from 146 items to 61 over the review period. Incoming volumes and the mix of receipts also changed, so the reduction cannot be attributed entirely to the new process. More importantly, the remaining 61 had an identified reason and next action; they were not written off or treated as resolved to improve the chart.

The relationship team still experienced slower onboarding during the temporary reassignment. I included that cost in the review rather than treating control work as free capacity. I would now establish the incoming-volume measure at the start of a trial, because without it a declining queue can look more conclusive than the evidence warrants.