How Do You Run a Bag Sourcing Project From Brief to Shipment?

Sourcing projects slip for a reason that is invisible in most plans: the plan contains the work and not the decisions. Every gate has a duration that the production side controls and a preceding decision that the buyer controls, and the second is routinely longer than the first. A sampling window of 6-10 working days can sit behind three weeks of unanswered emails about artwork, and the plan will still show the sampling window as the long part.

The fix is to plan the decisions. Name the owner of each gate, name the input it needs, name what it produces, and set a date by which the decision has to be made for the following gate to start on time. That is a plan a buyer can run, and it is the difference between a project that moves and one that waits.

This piece sets out the nine gates, the owners, and the artefacts, with the latency traps marked. It assumes one style in one or two colourways at a first-order quantity, which is the shape most of these projects take; a multi-style launch is the same sequence run in parallel with a shared critical path.

Who Owns Each Gate and Why Does It Matter?

A gate with two owners has no owner. The pattern that works is one named person per gate, with the authority to say yes or no at that gate and the obligation to say it by a date. The buying entity owns the brief, the tech pack and the approvals. The sourcing desk owns the quotation and the coordination. The production side owns sampling and bulk. The buyer or an agent owns inspection. The forwarder and the broker own shipment and entry.

What Does the Brief Have to Contain?

Four constraints and a reason for each. The quantity, per style and per colourway. The date, with the reason the date matters — a campaign, a store opening, a season — because a date with a reason behind it can be traded and a date without one cannot. The market, because it sets the compliance list. And the cost envelope, because it decides which constructions are in scope before anyone draws them.

The fifth item is the one that saves the most time: what is fixed and what is open. A brief that says the dimensions are fixed and the material is open produces a focused quotation. A brief that leaves everything open produces three rounds of samples, and three rounds of samples is the most common way a project loses a month.

How Does the Tech Pack Gate Work?

The tech pack is the document the project is built on, and its gate is the point at which it is signed off with a revision number. Everything after that refers to the revision: the quotation, the sample, the approval, the order. A project that loses track of its revision number will eventually approve one revision and build another, which is an argument that nobody wins.

Dimensions and tolerance

Stated, with a tolerance on the critical ones

Materials

Fabric, lining, foam, webbing and hardware named

Artwork

Vector, placed, with the colour count and method

Construction

Seam method, stitch class, reinforcement

Compliance list

The regimes the market imposes

Packing specification

Carton, pack-out and marking

Revision number

On the document and on every later artefact

Sign-off

Both sides, dated

Change rule

What happens to the window when the revision changes

Reference sample intent

That a production-tooling sample will be retained

The change rule is the field that protects the schedule. A project that agrees in advance that a revision after sign-off restarts the sampling window makes better decisions, because the cost of a late idea is visible before the idea is adopted rather than after.

Where Does Decision Latency Hide?

In four places. Artwork, because a brand takes three weeks to release a vector file that existed all along. Material choice, because two people have an opinion and neither owns the decision. Compliance scope, because nobody wants to say which markets are in play. And approval, because a sample arrives and sits on a desk while the window it was made in runs out.

The second rule is the one that matters most and the one most projects resist. A committee that has to approve a sample will take three weeks; a named person with authority will take two days. Where a committee is unavoidable, give it a date and a default — the sample is approved if nobody objects by then — and the calendar survives.

How Does the Critical Path Differ From the Calendar?

The calendar is the list of dates. The critical path is the chain of gates that cannot move without moving the end date, and it is usually shorter than the calendar suggests: brief, tech pack, sampling, approval, bulk, inspection, shipment, entry. Several things run in parallel and quietly — compliance testing, freight booking, labelling artwork — and a project manager who knows which are parallel can recover a week without changing any duration.

What Reporting Does a Project Need?

One page, updated weekly, with four fields per gate: the planned date, the current forecast, the owner, and the one thing blocking it. Anything longer is read less often. A status report that lists the gates in date order and marks three of them amber is more useful than a deck, because the reader can act on it in a minute.

There is a third discipline, and it is the one that separates a managed project from a tracked one: every gate has a definition of done. Sampling is done when a production-tooling sample is approved in writing, not when a sample ships; bulk is done when the goods are packed and offered for inspection, not when the last unit is sewn. Without a definition of done, two parties can each believe a gate is complete while the calendar moves.

The second discipline is the exception. A gate that has moved once should be reported as a date change with a cause; a gate that has moved twice should be escalated to someone who can add resource or change scope. Escalation on the second move rather than the fourth is the single habit that most improves delivery performance on these projects.

How Should Risk Be Carried on the Plan?

As a register with an owner and a trigger, reviewed weekly, and short enough to be read. Ten entries is the working size: the risk, the likelihood, the impact, the mitigation, the owner and the trigger that says the risk has become an issue. The development calendar from tech pack to bulk note sets out the calendar view of the same sequence, and the risk register belongs beside it rather than inside it.

The triggers are what make a register useful. "Dye lot may be late" is a worry; "if the lab dip is not approved by Friday the bulk window moves by a week" is a plan. Rewrite every entry until it has a date in it and the register becomes an early-warning system rather than a list.

What Happens When a Gate Slips?

Three options, in order of preference: recover it inside the following gate by running activities in parallel, compress the following gate by narrowing scope, or move the end date and tell everyone early. The third is the one that projects do last and should do first, because a date moved early can be absorbed by the rest of the business and a date moved late cannot.

The mechanism for recovery is almost always the same: something on the parallel list gets pulled forward. Freight booking at sample approval rather than at the end of bulk, compliance testing started on the sample rather than on the production lot, entry documents drafted during production rather than after it. Each of those recovers days without changing a single production duration.

How Much Float Should a Plan Carry?

Enough to absorb one thing going wrong, and not so much that the float becomes the schedule. The working rule is a float of a few days at each of the three gates the buyer controls — tech pack sign-off, sample approval and inspection release — and no float at all in the production window, because a float inside a 35-50 day window is consumed rather than held. Where a buyer wants more safety than that, take it at the front by starting earlier rather than in the middle by padding a duration.

The reason float belongs at the buyer gates is that those are the gates with variance. Production windows are stable because they are scheduled; approvals are not, because they depend on people. Put the buffer where the variance is and the plan becomes more predictable without becoming longer.

How Is the Team Structured on a Small Programme?

On a small programme most of those roles are one person wearing two hats, which is fine as long as the hats are separated at the gates. The failure mode is not understaffing, it is a decision-maker who is also the specification owner and cannot approve their own work, and a project that waits three weeks for an approval nobody realised was theirs to give.

How Do You Run a Multi-Style Launch?

As one critical path with several styles hanging off it, rather than as several projects. The shared gates — tech pack discipline, material procurement, inspection booking, freight — are planned once and the styles are sequenced through them. The binding constraint is usually not the production capacity but the number of things the buyer has to approve in the same week, which is why a multi-style launch slips at the approval gate rather than on the line.

The sequencing rule is to stagger the sampling so the approvals arrive in a planned order rather than together, and to share materials across the styles wherever the design allows. Shared fabric and shared hardware reduce procurement risk, reduce the number of tests required, and make the whole launch easier to reorder later.

The last question most teams ask is how much of this is worth formalising on a small programme. The answer is: the gates, the owners and the revision number, and nothing else. A single page listing nine gates with a name and a date beside each captures almost all of the benefit, and a project management tool with no agreed gates captures almost none of it. Formality belongs in the decisions rather than in the software.

What Does a Well-Run Project Leave Behind?

Four artefacts the next project starts with: a signed-off tech pack at a known revision, a retained reference sample, a recorded risk register with what actually happened, and a realistic calendar with the real durations in it. A team that has those four runs the second project in half the meetings, because the decisions were made once and recorded.

Frequently Asked Questions

People Also Ask