Improving DIFOT-Q in a Growing Garment Manufacturer

How customer complaints, 100,000 emails and an unstable production plan exposed a delivery-control problem in a growing garment business.

Business situation

The company was a growing garment manufacturer in Asia, exporting mainly to Australia and Europe. It employed around 350 people and produced relatively small batches of higher-quality apparel, where customers expected the correct garments, in the required quantity, on the agreed date and to the expected quality standard. Many managers had grown with the business and, while the company was smaller, experienced people had been able to keep a great deal under control through judgement, memory, relationships and personal follow-up. As order volumes, styles, suppliers and customer requirements increased, more information had to move between more people and the informal methods that had supported earlier growth were becoming increasingly difficult to sustain.

The delivery problem appeared while we were examining something else. An AI-assisted project had been reviewing approximately 100,000 emails covering a nine-month period in order to extract useful operational information from a very large volume of poorly structured correspondence. A repeated pattern appeared in both customer and internal emails: delivery dates were being missed, some shipments were late or short, customers were asking for explanations, merchandisers were chasing internally and managers were repeatedly becoming involved in delivery problems. On the factory floor there was plenty of activity, with production lines running, supervisors working, suppliers being chased and shipments being arranged, yet the customer experience remained inconsistent. The next stage was therefore to trace those delivery problems backwards through the operating process.

Comparing the production plans

The spreadsheet-based production plan became an important source of evidence because it connected customer delivery commitments with what the factory expected to manufacture. Production plans inevitably move in garment manufacturing because fabric can arrive late, trims may be missing, customers change requirements, quality problems arise and supplier dates change. Some movement is therefore normal, and the issue was to determine whether the amount and pattern of movement showed routine adjustment or instability in the operating system.

AI was used to compare two production plans several months apart. Twenty-one styles were examined and 20 had been rescheduled, giving plan churn of 95 per cent. Ten styles had been pulled forward by an average of 10 days, six had been pushed later by an average of 15 days, four other rescheduled styles were not cleanly classified in the summary and one style had no confirmed date. At that level of movement, the production plan was providing weak stability as a management control, while every unplanned change also created additional work as people revised schedules, communicated changes and dealt with the effects elsewhere in the process.

Finding Results

                   Styles reviewed                                           21

                   Styles resheduled                                       20

                   Plan Churn                                                   95%

                   Styles pulled forward                                 10

                   Average movement forward                     10 days

                   Styles pushed together                                6

                   Average movement later                           15 days

                   Other rescheduled styles                             4

                   Style with no confirmed date                       1

The styles that had moved later were consistent with the customer complaints and therefore did not materially change our understanding of the problem. The more interesting evidence came from the ten styles that had moved forward. Bringing production forward can reflect good customer service, so the direction of movement alone could not establish whether the change was beneficial. The pattern needed to be examined in the context of what was happening elsewhere in the production plan.

What the forward movements showed

When an order could not proceed because materials, customer approvals, supplier performance, quality or another constraint prevented production, another available order could be brought forward to occupy the line. There was a sound operational reason for doing this because idle sewing lines are expensive, labour has to be used and supervisors need work available for their teams. Filling an occasional production gap can therefore make good economic sense. Across the plan, however, repeated reshuffling created a different operating pattern in which production sequence was increasingly influenced by what the factory was able to make at that moment rather than consistently by the sequence required to protect customer commitments.

The plan was absorbing disruptions by rearranging production. This protected utilisation in the short term, while the resulting churn transferred consequences into other orders, customer commitments and later parts of the schedule. That helped explain why a factory that appeared busy and well occupied could still disappoint customers. Production activity itself was not giving management enough information about whether the right orders were being made in the right sequence to achieve the required delivery outcome.

Connecting the functions

Delivery in full, on time and to quality — DIFOT-Q — depended on much more than sewing-line performance. Merchandising, procurement, planning, production, quality, warehouse and logistics all contributed to the customer outcome, with Finance also providing information and control over parts of the process. The review found that too much operating knowledge remained distributed between individual managers, email threads, spreadsheets and informal conversations, so different people could hold different parts of the current position. A merchandiser might understand a customer change, procurement might know that a material date had moved, while production might see only that another style had become available for manufacture.

The business lacked one trusted operational view linking the customer-required date, current production position, material readiness, reason for any movement, effect on the customer commitment and responsibility for the next action. Where information is fragmented in this way, production priorities can increasingly reflect urgency, personal knowledge and whoever is pressing hardest at the time. Experienced managers can compensate for that weakness for a considerable period, although the amount they have to remember, reconcile and chase rises as the business becomes larger and more complex. The organisation had reached the point where the management system needed to carry more of that load.

Management response

The production-plan analysis changed the management discussion around delivery. Instead of treating each late shipment as an individual problem requiring explanation afterwards, senior management could examine the movement of the plan itself and ask what had caused a rescheduling decision, which customer commitment was affected and whether the movement protected delivery or filled an immediate production gap. DIFOT-Q consequently became a broader senior-management issue rather than a production statistic, while accountability also required attention because several functions influenced the customer outcome without sufficiently clear end-to-end ownership of the delivery promise.

The CEO recognised that the existing management controls had not developed at the same rate as the business. The company had grown in size and complexity while coordination still depended heavily on experienced people, spreadsheets, email and personal intervention. The review also brought production-management capability into the discussion because the existing team had grown with the company while the next stage required stronger planning discipline, additional production-management experience and management systems capable of supporting a larger operation.

Outcome

The immediate value of the project was a clearer diagnosis of what management needed to address. AI analysis of the email traffic had identified the repeated customer-delivery problem across a body of correspondence too large for normal manual review, while comparison of the production plans quantified the instability: 20 of 21 styles had moved, with substantial movement in both directions. Examining why so many orders were being brought forward showed how disruption elsewhere in the system was being absorbed by repeated resequencing of production.

Management could now see that late and short shipments were connected to a wider control problem. Production utilisation had become a powerful influence on daily sequencing, while customer commitments lacked equally strong end-to-end control across the organisation. The CEO accepted that the systems and production capability which had supported the company’s earlier growth were no longer sufficient for its current scale, DIFOT-Q became part of the senior-management agenda, accountability for delivery required strengthening, and production-plan movement could now be treated as an early indicator of operating instability rather than merely another revised schedule.

For a growing manufacturer, the distinction matters because high utilisation remains economically necessary, but production capacity only earns its value when it is converted into reliable customer delivery. As complexity grows, achieving that consistently requires clear ownership of the customer promise, common operational information and a production plan stable enough to function as a management control.