Count the linen the customer can actually use

In 2022 I supported a commercial planning assumption of 24 tonnes per day based on the washer-stage capacity available for a proposed mix of hospital bed linen and staff garments. Site leaders challenged the figure because drying and finishing were already creating queues. We carried out a defined operating study using the same dry-weight basis, time window and product mix at each stage. It showed 14 tonnes through drying and 12 through finishing under those conditions. The washer figure did not describe the completed service we could promise.

The commercial team wanted to accept the proposed additional work before a competitor did. I withdrew my support for that volume assumption and asked engineering, quality and operations to show alternatives. We considered phasing the customer start and adding downstream resources rather than purchasing more washing equipment. Quality retained the relevant acceptance requirements; the proposal did not shorten a process or release held material to make the capacity account look better. The comparison also retained downtime and the limits of the measured conditions rather than treating a theoretical rate as a daily certainty.

We approved ₹1.8 crore for downstream equipment and work organisation, with a ten-week implementation and review period. The new volume remained conditional on demonstrated capability after the change. I did not publish a new achieved rate before that evidence existed. Delaying acceptance risked losing business, and the investment was a real commitment rather than a claimed saving. The board paper showed those disadvantages alongside the problem with accepting an order the existing flow could not complete.

The case changed our capacity review. A commercial proposal now identifies the constraining stage, the measurement basis and the evidence needed before an expanded service begins. A useful number must describe the work that can reach the agreed completion point, not merely the work that can enter the process. I had allowed the faster-stage figure to stand for the whole system. Correcting that assumption made the immediate opportunity smaller, but gave the company a more defensible choice about investment and timing.

A 2024 order called for 160 standard sheets and 80 long sheets, a total of 240 pieces. The packed consignment contained 200 standard and 40 long sheets. The total matched the order, but forty of the required long sheets were missing and forty standard sheets were surplus. Our aggregate verification had treated the two item types as interchangeable. The customer-service team identified the discrepancy before the consignment was accepted as meeting the requirement. I had supported the simplified count check because it appeared to speed dispatch.

We corrected the consignment and reviewed the picking and verification process with packing staff. Some supervisors preferred adding a final recount because that was familiar and quick to explain. I asked them to follow where the size distinction had been lost. We separated the item identifiers and packing locations, retained the ordered mix on the work instruction and checked each required category before the total. Quality and customer acceptance conditions remained in force; an accurate count by itself did not establish the condition or suitability of the linen.

The repair cost ₹22 lakh over seven weeks, including changes to layout, work instructions and supervised practice. Packing took longer during the transition. We kept that consequence visible instead of claiming that improved accuracy immediately increased productivity. The customer received the corrected mix, but we did not extrapolate one repaired order into a universal error rate or a clinical benefit. We also retained a route for discrepancies that could not be resolved by swapping a nearby item.

The operating lesson was that a total can pass while the service fails its actual requirement. We now ask whether the verification preserves the categories the customer needs, not simply whether the arithmetic adds up. The same thinking informs garment and site-specific instructions without assuming that all orders should share one standard mix. A count is useful when it describes the right things. It becomes misleading when the process quietly allows an extra item in one category to stand in for a missing item in another.

In 2025 a new-contract proposal described 12,000 serviceable linen pieces at our sites as an available pool. A customer-service review found that 4,500 belonged to existing customers under segregated-stock terms. They were physically present and serviceable, but they were not company-owned stock available for the new commitment. The company-controlled pool was 7,500 pieces. I had accepted the aggregate inventory view without asking whether custody and allocation authority had been kept separate.

The sales team argued that the pieces moved through the same facilities and that replenishment could follow the contract start. I rejected that approach because it assumed rights we had not established. Operations, finance and customer-service colleagues reconciled the ownership records and relevant terms with appropriate advice. We retained the customer stock in its agreed allocation and corrected the proposal. This was a decision about the fictional contracts and ownership records, not a general legal opinion about every laundry arrangement or a claim that customer property could never be handled under a different authorised agreement.

We deferred the new contract and approved 3,000 additional company-owned pieces at a cost of ₹18 lakh, with a six-week sourcing and acceptance plan. Their arrival would bring the company pool to 10,500 pieces before considering the needs of the actual contract. Quality acceptance and the final operating allocation still had to be completed; we did not treat a purchase order as ready-to-use stock. The delay could cost the company the opportunity, and the purchase was additional expenditure rather than a recovery from existing customers.

The repair established a planning view that distinguishes location, condition, ownership and permitted allocation. A move between sites should not change those attributes silently. I also changed the question I ask before approving a stock-backed promise:what makes these particular pieces available for this particular customer? The answer needs more than a count and a warehouse location. Keeping the ownership condition visible narrowed the immediate offer, but avoided asking operations to fulfil it through an unauthorised assumption.