How Do You Set Up a Pilot Bag Order That Proves the Specification?

Most first orders are treated as small versions of the order that will follow, and that is the mistake at the centre of most disappointing first programmes. A first order is the only run that will be made without any history behind it, so it is the only run that can answer the questions the later ones depend on. Whether the construction survives a line, whether the colour holds across a dye lot, whether the pack-out survives the lane, whether the market wants the colourway — none of those can be answered more economically than by making the thing once, properly, and writing down what happened.

The second reason a pilot is worth designing is that the learnings have to transfer. A pilot run on a simplified material set, or on a sample-room process, or at a quantity the repeat will not use, produces answers that do not survive contact with the repeat. The discipline is to make the pilot as close to the intended programme as the minimum allows, to measure it properly, and to accept that it costs more per unit than a larger run will because it is buying information.

This piece sets out what a pilot has to prove, how to design it so the answers transfer, what to measure, and how to end it with a decision rather than with an impression. It assumes a first programme with a production base the buyer has not run with before, which is the case where a pilot pays for itself.

What Is a Pilot Order For?

Four questions, and nothing else. Can the article be made to the specification on a production line rather than by hand? Does it arrive in the condition it left, through the actual pack-out and the actual lane? Does the market want it, at the price and in the colourways chosen? And can the run be repeated, with the same result, at the same cost? Those four are worth more than a small saving on the first run.

How Does a Pilot Differ From a Sample?

A sample proves that the specification can be made once; a pilot proves that it can be made repeatedly, shipped and sold. The sample is made with attention, often by the most skilled people available, and it is the best the article will ever look. The pilot is made on the line, by the people who will make it next time, in the quantity the line will see next time, and it is the honest version.

That distinction is why a pilot cannot be replaced by a good sample. Buyers who approve on the strength of a beautiful unit and then skip the pilot discover the line-learning curve in bulk, where it costs a re-work rather than a note. The sample answers "is this the article"; the pilot answers "is this the programme".

How Should the Pilot Be Specified?

Identically to the intended repeat. Same tech pack, same materials, same hardware, same construction, same pack-out. The temptation is to simplify the pilot to get it made faster or at a lower unit cost, and the effect is that every answer the pilot produces is about a different article. A pilot with a substituted fabric answers nothing about the fabric the programme will use.

Specification

The full production specification, at a revision number

Quantity

500 pcs per style and colourway, the standard minimum

Sample route

A production-tooling sample, approved in writing

First-piece check

Against the reference, before the run continues

Inspection

ISO 2859-1, General Inspection Level II, AQL 2.5 major

Pack-out

The pack-out the repeat will use, tested on the actual lane

Measurement plan

What will be counted, by whom, and when

Review date

A dated session to read the results

Decision rule

Scale, adjust or stop, with the criteria written

Artefacts

Reference sample, parameters, defect history, sell-through

The review date and the decision rule are the two fields most often left out, and they are the reason pilots drift. Without a date, the results are read when someone has time, which is after the next season has been planned. Without a rule, the reading becomes a discussion about whether the results are good enough, and the discussion ends wherever the loudest voice sits.

One more field belongs in the pilot specification because it is the one that decides whether the evidence is admissible: what the repeat will be measured against. The pilot produces the reference sample, and the repeat is compared to it. If the pilot is not specified to produce a unit worth retaining — made on production tooling, at production settings, from a recorded material lot — then the repeat has nothing to be compared with.

How Much Should a Pilot Cost?

More per unit than the repeat, and that is the correct answer rather than a problem. The first run carries the pattern work, the tooling where there is any, the setup time and a higher defect rate while the line learns the article. A buyer who negotiates the pilot down to the repeat price is usually buying a pilot made on a simplified process, which is the one thing a pilot must not be.

What should not be paid for twice is the specification work. The tech pack written for the pilot is the tech pack for the programme, at a revision number, and the second run starts from it. That is where the pilot earns its premium back: the repeat does not re-develop anything.

One cost on a pilot is worth paying deliberately, and it is the independent check. A line-side count is useful for running the line; a check made by someone who did not make the article is useful for deciding whether the specification is right. The first tells the supervisor what to correct today, and the second tells the buyer what to change before the next run. A pilot measured only from the inside has to be taken on trust, and trust is not what a pilot is for.

What Does the Pilot Do to the Calendar?

It adds nothing to the production windows and it adds a review period at the end. Sampling is 6-10 working days and bulk is 35-50 days whether the run is a pilot or a repeat; what a pilot adds is the period in which the results are read and the next run is specified. That period is usually a few weeks of selling, and it has to be on the calendar from the start.

The consequence for a seasonal programme is that a pilot has to be timed so its review lands before the scale-up must be ordered. A pilot that lands in the selling season produces a decision too late to act on, which is the most common way a pilot wastes its own cost. Put the review date and the scale-up order date on the same calendar at the beginning.

What Should Be Measured During the Run?

The second of those is worth more than the third. A pre-shipment count says whether the lot passed; an in-process count says why. A pilot that produces only the first is a pilot that has to be run again to learn the same thing, and running a pilot twice is the most expensive way to learn anything.

The fifth measurement deserves a note about timing, because it is the one that is usually left too late to use. Sell-through per colourway needs a stated period on the calendar before the goods land, and it needs the denominator: how many units of each colourway were actually available to sell. A rate computed against the forecast rather than against what reached the shelf will flatter the colourway that sold out first and condemn the one that arrived last.

How Should the Pilot Be Packed and Shipped?

In the pack-out the repeat will use, on the lane the repeat will use, with a condition check at the far end. Arrival condition is one of the four questions a pilot answers, and it is the only one that cannot be answered any other way. A pilot shipped by a faster route than the repeat, or packed more carefully for the occasion, answers nothing about arrival.

The check itself is simple: count the affected cartons before unpacking, photograph them, then count the affected units and note where they sat in the carton. That last detail is the useful one, because damage concentrated at the base or at the edges points at the carton and the pallet rather than at the bag.

How Does the Inspection Plan Work on a Pilot?

The same way as on any run: ISO 2859-1 at General Inspection Level II, AQL 2.5 for major defects and 4.0 for minor, critical at zero. On a 500-piece lot that draws code letter H and a sample of 50 units, accepted on three major defects and rejected on four. What is different on a pilot is the class list, which should be written with the pilot questions in mind rather than copied from a previous programme.

What Are the Common Ways a Pilot Goes Wrong?

Five. It is specified differently from the repeat, so nothing transfers. It is measured only at the end, so the pattern is never seen. It produces no artefacts, so the reorder starts from nothing. It is read without a date or a rule, so it becomes a discussion. And it is treated as the final answer rather than as the first data point, so a mediocre result is either over-corrected or ignored.

The fifth is the subtle one. A pilot is a sample of one run; a defect count from one run has wide confidence limits and a sell-through figure from eight weeks is not a forecast. Read the pilot for direction and for artefacts, and treat the second run as the confirmation rather than assuming the first settled anything.

A sixth way a pilot goes wrong is quieter than the other five: the specification is changed during the run. A substitution agreed on the phone, a hardware change made because the specified part was late, a pack-out altered to fit a carton that was available. Each is reasonable at the time and each invalidates a different answer. The control is the same one that governs bulk — any change to a specified item is written, dated, and either approved or reverted.

What Does the Pilot Mean for the Repeat?

Everything, if the artefacts exist. A repeat specified from the pilot tech pack, made to the recorded parameters, compared against the retained reference sample, packed in the tested pack-out, with the colourway split informed by actual sell-through, is a different proposition from a second first order. The windows are the same — 6-10 working days for sampling where anything changed, 35-50 days for bulk — but the probability of hitting them is not.

Where the pilot showed a problem, the repeat is where it gets fixed, and the fix belongs in the specification rather than in an instruction. A shade that drifted gets a tolerance and a dip route. A carton that crushed gets a board specification. A panel that scuffed gets interleaving. Each of those is a line in a document, and each is the reason the second run behaves differently from the first.

When Should a Pilot Be Skipped?

When the construction is known to the line, the materials are unchanged and the market question has already been answered by a previous run in the same range. Adding a colourway to a running programme is not a pilot; it is a reorder with a new dye lot. What that needs is a dip and a first-piece check, not a full pilot cycle.

The other case is where the buyer already has the evidence from a local run: the same specification proved on a domestic line, now being moved offshore. Even then, a small pilot earns its keep, because the question is no longer whether the article works but whether the new line can make it, which is a different question with a different answer.

How Should a Pilot End?

With a written decision, on the date that was fixed at the start, against the rule that was fixed at the start. Scale means the repeat is specified and the parameters are recorded. Adjust means the changes are listed, costed against the calendar, and applied at a new revision. Stop means the reasons are recorded so the next idea does not repeat them.

The discipline that makes all three of those outcomes work is the artefacts. Whatever the decision, the programme keeps the tech pack at its revision, the reference sample, the parameter and material-lot record, and the defect history. A pilot that ends in a stop decision and leaves those behind has still paid for itself, because the next programme starts from them.

Frequently Asked Questions

People Also Ask