Peak-season DPI timing works best when the booking follows the line instead of the calendar. This seventh part of the series recommends reserving a target week early, then triggering the visit when about one third of the order is finished and machines still run. That production-share trigger keeps the inspector inside the correction window even when start dates slip by a week. Fair sampling from partial output and attention to process discipline complete the method, drawing on acceptance sampling logic and ISO 9000 process thinking.
Key takeaways
- Book a target week plus a production-share trigger, such as one third finished with lines running, instead of betting everything on a single calendar date.
- A fair mid-run sample must spread across sizes, colors, and lines, applying the representative-sampling idea behind acceptance sampling to partial output[1].
- Mid-run visits should observe process discipline as well as product results, because the ISO 9000 family treats consistent processes as the foundation of consistent output[2].
- Daily output figures with photos in the week before the visit prevent wasted trips to factories that are not ready.
- If the production state has passed, re-scope the visit honestly rather than filing a final check under a DPI title.
Every peak-season DPI that fails was booked against a calendar date instead of a production state. The factory promises the 15th, the buyer books the 15th, and on the 14th the factory reports that only one tenth of the order is finished or that packing has already started. The inspector then arrives to find either too little output to judge or sealed cartons that no longer allow correction.
Experienced buyers book against output share instead. The instruction says the visit triggers when roughly one third of units are finished while lines still run, and the booking message names a target week rather than a fixed day. This small wording change absorbs the schedule slips that define peak season, because the factory calls the visit when the line state is right rather than defending an optimistic date.
This seventh part of the timing series shows how to run that output-based booking method. It covers setting the trigger, confirming readiness with daily figures, and sampling fairly from partial output. Acceptance sampling supplies the fairness logic for partial-lot decisions, and ISO 8000-family thinking on process control explains why mid-run visits also observe how work is done, not only what was produced.
Why calendar dates collapse in peak season
Peak production schedules are written under pressure and revised under greater pressure. Material deliveries slip, workers rotate between orders, and urgent repeat orders jump the queue. A fixed inspection date agreed three weeks ahead therefore meets the planned line state only by luck. Buyers who insist on that date either inspect too early, when output reflects setup trials, or too late, when corrections require reopening cartons.
Output-share triggers survive this churn because they describe a line state rather than a day. When roughly one third of units are finished, enough steady-run output exists to judge while most of the order can still benefit from corrections. The factory understands this language immediately, since supervisors already track daily output against the plan. Planning a visit around during production inspection in China works the same way: the service is defined by what the inspector can still influence, not by which weekday the visit falls on.
The booking message should therefore carry both elements: a target week that reserves capacity, and a trigger sentence that activates it. A practical wording states the visit will take place in the named week once finished output passes an agreed quantity with lines still running. Both sides then share one observable condition instead of trading optimistic promises.
Setting a trigger the factory can actually report
A good trigger has three qualities: observable, countable, and tied to running lines. Observable means the factory can photograph it, countable means both sides agree the number, and tied to running lines means packing has not yet absorbed the output. A trigger such as one third of units finished with at least two lines running meets all three, while a vague phrase such as mid-production invites dispute about what mid means.
Confirmation runs on short daily updates in the final week. The factory reports finished quantity per size and color, cartons sealed so far, and whether the relevant lines run that day, ideally with two or three line photos. These messages take minutes and prevent the classic failure where an inspector travels to find an idle workshop. When figures stall two days in a row, the buyer moves the target week rather than forcing a visit with nothing to check.
Edge cases need honest re-scoping. If output rockets past the trigger because the factory added night shifts, the visit should move earlier within the same week or convert to a final check with clear limits. If output stalls below one tenth, the visit should wait, because judging setup trials as steady production misleads everyone. The discipline to reschedule is part of the method, not a failure of it.
Sampling fairly from half-finished output
Partial output tempts selective sampling: the factory offers a basket of neat pieces near the packing table while the line produces a wider reality. A fair mid-run sample resists that temptation by spreading pulls across the output available so far, covering each size, each color, and each active line. The underlying idea matches acceptance sampling theory, which studies how to draw defensible conclusions about a larger quantity from a examined part[1].
Stratification matters more than raw sample count at this stage. Twenty pieces drawn evenly across four lines reveal drift that forty pieces from a single line conceal. The inspector should record which line and hour each piece came from, so a defect cluster can be traced to its source period. Measurement results then read as a process picture rather than a pass-or-fail lottery.
Reference discipline completes fairness. Every pull is compared against the same approved sample, specification sheet, and color standard, with lighting and measuring tools noted in the report. When the reference itself is ambiguous, the inspector flags the ambiguity instead of inventing a verdict. Clear references turn mid-run numbers into decisions the factory can act on the same day.
Watching the process, not only the product
A DPI visit sees something a final check never sees: work being done. Operators cut, sew, assemble, and test while the inspector watches, which exposes the causes behind the numbers. The ISO 9000 family of thinking treats this process view as central, framing quality as the result of controlled processes, clear responsibilities, and continual improvement rather than last-minute sorting[2].
Practical process checks fit inside a one-day visit without turning it into an audit. The inspector notes whether written work instructions sit at the stations, whether measuring instruments carry identification, whether first-piece approval is recorded at shift start, and whether rejected pieces are physically segregated. Each gap found here explains a class of defects downstream. A missing segregation area, for instance, predicts mixed batches; unidentified measuring tapes predict measurement drift.
Findings should separate product results from process observations in the report. Product results say what share of examined pieces drifted and where; process notes say which workstation habit likely produced the drift. The factory can then fix both the visible defects and the habit that creates them, which is the only way the next third of the order runs cleaner than the first.
Peak season rewards buyers who book early and trigger on output. Reserve the week, confirm with daily figures, sample across lines and sizes, and watch how the work is done. That combination keeps the mid-run visit inside the window where corrections still change the outcome.
Sources
- Wikipedia: Acceptance sampling —— Overview of acceptance sampling including AQL concepts used in product inspection.
- Wikipedia: ISO 9000 —— Overview of the ISO 9000 family of quality management standards.
Frequently asked questions
How do I pin down the visit date when the factory keeps moving its schedule?
Name a production-share trigger in the booking, for example when about one third of units are finished and lines still run. Ask the factory for a short daily output figure with photos starting one week before the target. When output reaches the trigger, both sides fix the visit within two days.
Can I use standard AQL tables directly on half-finished production?
Representative samples must come from current output across sizes, colors, and lines, which follows the same principle that acceptance sampling theory describes for drawing fair conclusions from a part of a lot. A basket of hand-picked good pieces proves nothing and hides drift. The sampling habit matters more than the exact table when only part of the order exists.
What does ISO 9000 add to a mid-production visit?
ISO 9000 describes process control, documented procedures, and continual improvement rather than a product pass mark. An inspector can observe whether operators follow written steps, whether measuring tools are identified, and whether nonconforming pieces are segregated. Those observations explain why defects repeat, which a pure product check cannot.
The factory finished early. Is my scheduled DPI still valid?
Keep it as a DPI only when lines still run and unfinished output remains. If packing has started, re-scope honestly as a final inspection with narrower claims. A mislabeled report corrupts trend data and gives false confidence about process stability.