How Do You Manage Inspection on a Remote Production Programme?

The instinct when production is far away is to compensate with travel, and travel is the least efficient instrument available. A visit sees one day of one run. A cadence sees every week of every run. What makes a remote programme predictable is not physical presence but the same three things that make a co-located one predictable: a known specification, a known schedule and a report that arrives whether or not anything has gone wrong.

The second instinct is to compensate with messaging, which is worse, because messages produce activity without producing records. A programme run on messages has no artefact to compare a reorder against and no evidence when something is disputed. A programme run on a weekly report with photographs has both, and it costs less time than the messaging does.

This piece sets out the four gates, the report, the photographs, and the escalation path, with the failure modes marked. It is written for a buyer who cannot be on the line and does not need to be, which is the normal condition of a programme rather than a compromise.

What Are the Four Gates and What Happens at Each?

Why Is the First-Piece Check Worth More Than the Final One?

Because it is the only gate at which a problem can still be corrected inside the window. A pre-shipment inspection can accept or reject and nothing else; a first-piece check can change the run. On a 500-piece order the first-piece check happens when a handful of units exist, which is the moment a pattern problem, a placement problem or a colour problem is still inexpensive to fix.

The check has to be against the reference rather than against a description. Dimensions against the tech pack, artwork placement against the marked drawing, colour against the retained swatch, construction against the named stitch and seam method. A first-piece check against a conversation produces a first-piece approval that means nothing by unit 400.

What Does an In-Process Inspection Find That a Final One Does Not?

The defect pattern rather than the defect count. A pre-shipment sample says that three units in fifty have a fault; an in-process visit says that the fault appears on units made on the second shift, or after a material change, or when a particular operator is on the station. The second statement is actionable and the first is not.

Date and run reference

So the report can be matched to a lot later

Units produced and units checked

A rate, not just a count

Defects by class

Critical, major and minor, each counted

Photographs

Taken to a stated standard, with a scale

Parameter record

The weld, stitch or press setting in use

Material lot

Which fabric and trim lots are on the line

Deviation from schedule

Units against plan, in units and in days

Corrective action

What was changed, by whom, and when

Re-check result

Whether the correction worked

Open items

What is still unresolved, with a date

Two of those carry most of the value. The parameter record is what makes a reorder consistent: if the setting that produced an acceptable run is written down, the next run can be made to it rather than rediscovered. The material lot is what makes a compliance file defensible: a declaration attached to a lot rather than to a material.

What Should the Weekly Report Contain?

The fourth field is the one that makes the report useful rather than decorative. A status update that does not ask for anything produces no decisions, and a remote programme dies of undecided questions rather than of poor workmanship. Every report should end with the decisions it needs, and every decision should have a date by which a silence becomes an approval or a rejection.

How Should Photographic Evidence Be Taken?

To a standard, because photographs are the substitute for being there and a bad photograph is worse than none. Four rules: a consistent distance and angle for the standard views, a scale in the frame for anything dimensional, a colour reference in the frame for anything about colour, and a caption that says what the photograph is of and when it was taken.

The colour reference deserves a note. A photograph of a bag on a bench tells a buyer nothing reliable about the shade, because the camera adjusts. A photograph with a retained swatch or a grey card in the frame tells the buyer something. Where colour is the acceptance criterion, it should be judged against a physical dip sent by courier rather than against an image.

How Is a Reference Sample Used Day to Day?

As the physical answer to every ambiguous question. A drawing can be read two ways and a sample cannot, so the reference settles the questions that drawings generate: how stiff the interlining feels, how the panel sits, how the print looks at the stated size. Both sides hold one, both are dated, and both are the same unit rather than two units from the same run.

The second use is the reorder, and it is the reason the sample is worth the courier cost. A second run compared against a retained physical unit is a comparison; a second run compared against a description is an argument. Programmes that hold the sample have short reorder conversations, and programmes that do not have long ones about whether anything actually changed.

How Does Escalation Work Across Distance?

What Does the Time Difference Do to the Cadence?

It turns a question into a day, which is the central fact of a remote programme. A question asked at the wrong hour waits a full day for an answer, and a run that needs three answers loses three days. The mitigation is not working odd hours; it is asking all the questions at once, in one message, at the start of the working day at the production end.

The second mitigation, and the one that removes the most latency over a programme, is the standing decision. Where a class of question recurs — a thread shade, a stitch density, a minor placement tolerance — agree the answer once, in writing, and let the line proceed without asking. A programme that converts recurring questions into standing decisions removes most of the latency that distance creates.

How Much Does a Visit Actually Help?

Less than expected on an established programme and more than expected on a new one. A visit to a line that is already running a known construction confirms what the reports say and adds little. A visit before the first run of a new construction is genuinely useful, because it is the only way to see whether the equipment, the material handling and the skill exist for that article.

The efficient pattern is therefore a visit at the start of a relationship and at the start of each new construction, with the routine runs managed on gates and reports. That concentrates travel where it changes decisions and removes it where it only reassures, which is the honest description of most visits to a line that is doing what it was told.

What Does the Reporting Cost in Time?

What Should the Buyer Hold on Their Side?

Four artefacts: the signed-off tech pack at a revision number, the retained reference sample, the inspection reports with their defect counts, and the parameter and material-lot records. Those four let a buyer answer a customer question, settle a dispute and run a reorder without asking anyone for anything.

The fifth artefact is a calendar with the real durations in it, updated after each run rather than carried forward from the plan. Programmes improve mainly by learning what their own windows actually are, and that knowledge lives in a document rather than in a memory. The development calendar from tech pack to bulk note covers the calendar view of the same sequence.

One more field belongs in the buyer file, because it is the one that turns a programme into an asset rather than a series of orders: a costed bill of materials at the revision that was built. Knowing what the article is made of, from whom and at what cost, is what makes the next negotiation short and the next substitution visible. Without it every reorder starts from a quotation rather than from a known position, and a buyer cannot tell whether a price move is a market move or a material one.

How Does This Change for a Reorder?

It gets shorter and it should not get looser. The gates stay, but the first-piece check is against the retained reference sample rather than against the drawing, and the in-process inspection can be lighter where the defect history is clean. What should never be dropped is the pre-shipment inspection, because a reorder is exactly where drift appears and the sample is where it is caught.

The third discipline is the defect history. Keep the counts by class, run by run, and the pattern becomes visible: a station that produces the same defect, a lot that produces a shade drift, a setting that produces a weak seam. That history is what turns inspection from a gate into a control, and it is the artefact a buyer most regrets not having kept.

The second reorder discipline is the parameter and lot record. A second run made to the recorded settings, from an approved lot, is a repeat of a known result. A second run made to a remembered setting, from a fresh lot, is a new experiment dressed as a repeat, and it is where most reorder disappointment comes from.

A further element deserves a place in the file because it is rarely written down and frequently decisive: the decision rights of the inspector. An inspector who can stop a run has a different effect from one who can only report, and the difference should be explicit in the inspection brief. Where a critical defect class has been agreed at zero acceptance, the authority to stop follows from it; where it has not, the run continues and the problem is counted later.

What Is the Failure Mode of a Remote Programme?

Not distance and not workmanship: undecided questions. A remote programme accumulates small open items — a trim shade, a label position, a tolerance interpretation — and each one waits for a day, and the run continues past the point at which it could have been corrected. By pre-shipment inspection the accumulation is a lot that fails, and the failure looks sudden.

The remedy for this failure mode is the same one that governs the rest of the programme: a dated decision on every open item, with a default. An open item with no answer by its date is answered by the standing decision, or the run stops and says so. Either is better than the drift that produces a rejected lot three weeks after the question was asked, and neither requires anyone to work outside their own hours.

Frequently Asked Questions

People Also Ask