When Does Dual Sourcing Make Sense in Apparel Production?

Somebody senior asks why the programme depends on one factory, and within a week there is a plan to split it. The plan often skips two questions: whether a second factory addresses the thing that would actually stop supply, and what the same garment made in two places will look like on the same shop floor.

The question that is actually being asked

"Should we single-source or dual-source?" is the wrong shape, because it treats supplier count as the variable. The useful version is: which failure would stop supply, and would a second factory address it? Sometimes the answer is obviously yes. Often the thing that would stop supply is a mill, a nominated component or a specialist process that both factories would share, in which case a second supplier costs money and changes nothing.

Working out which of those you are in is the subject of supplier risk management, and it is worth doing before this decision rather than after. This guide assumes that analysis has happened and asks the narrower question: given that a second production source would genuinely help, is splitting the programme the right way to get one?

When one supplier is the right answer

Single sourcing has a bad reputation it does not always deserve. There are programmes where concentrating is the better decision and duplicating is simply an expensive way to make the product worse.

• The volume is small. Split across two factories, each half may fall below a workable minimum, and both suppliers lose interest in a programme that no longer justifies attention. Where that boundary sits is a function of the product, see minimum order quantities.
• Development or setup cost is high. Patterns, fit sessions, decoration setup and material development are paid twice, and the second payment buys no additional product.
• The process is genuinely specialised. Where few suppliers can do the work properly, the realistic alternative is often worse rather than equivalent.
• Consistency is the product. A corporate uniform replenished over years, or a colour-critical range, can be damaged more by visible variation than by the risk the split was meant to reduce.
• The programme is short-lived. A single season or a one-off event may end before the duplicated development has paid for itself.
• Concentrated volume buys something. Better pricing, better scheduling, better attention. Splitting it halves the leverage at both suppliers.

When splitting earns its cost

The cases where a second production source pays are reasonably specific, and they are worth naming so the decision is made on one of them rather than on general anxiety.

• Volume genuinely exceeds one site. The clearest case, and the least controversial, it is a capacity decision as much as a risk one. What changes as volume grows is in scaling production.
• A fixed, unmovable date. An event, a launch or a seasonal window where being late is indistinguishable from not delivering shifts the balance sharply toward having a running alternative.
• The product is commercially critical, whatever its volume. A core style that carries a range is more exposed than its unit count suggests.
• The dependency is genuinely at the factory. Where the risk is the site, its capacity, its stability, its concentration of your volume, a second site addresses it. Where the risk is upstream, it does not.
• The product tolerates variation. Some garments can absorb small differences between makers without a customer noticing. Some cannot.

Active split, or standby alternative

These are different arrangements with different costs, and conflating them is why some dual-sourcing plans deliver neither resilience nor savings.

The consistency bill

This is the part that gets underestimated, because it does not appear until the two deliveries are next to each other. Two competent factories working from the same specification produce two garments that are the same in every respect somebody wrote down, and different in several nobody did.

None of that is an argument against splitting. It is an argument for pricing it honestly: a split programme needs a stronger specification, more approvals, more physical references and more inspection than a single-source one, and those costs land whether or not anybody budgeted for them. The drift controls themselves are the same ones that hold a product stable across repeat orders, covered in reorder consistency, and colour has its own mechanics in dye-lot shade variation.

The duplication nobody budgets

The obvious cost of a second source is the development. The less obvious cost is that everything downstream of development also happens twice, permanently.

• Approvals. Lab dips, strike-offs, sew-outs, fit samples and pre-production samples at both, for every style and every change.
• References. Sealed samples and colour standards produced, distributed and kept current in two places.
• Inspection. Two production windows to cover rather than one.
• Compliance evidence. Testing and documentation assessed for both routes, since a report on one supplier's material does not automatically cover another's.
• Change control. Every specification revision has to reach both suppliers, at the same revision, and be confirmed by both. A split programme running two revisions is worse than a single-source one.
• Relationship time. Two sets of reviews, two sets of corrective actions, two commercial negotiations.

That is why splitting a programme is a decision about the whole operating model rather than a sourcing tweak. A team that is already stretched managing one supplier well will manage two badly, and two badly managed suppliers are less reliable than one good one.

How much volume to each

There is a strong temptation to reach for a split, eighty-twenty, seventy-thirty, half and half. None of those is a rule, and none of them is derived from anything. The right allocation depends on what the split is for.

• If the purpose is capacity, the allocation follows what each site can actually run well, not a ratio.
• If the purpose is keeping an alternative live, the second supplier needs enough real work to stay familiar with the product, which is a question about the product's complexity, not a percentage.
• If the purpose is leverage, splitting reduces it at both suppliers, so the calculation is about what concentrated volume was buying in the first place.
• Minimums bind the answer. A split that takes either supplier below a workable quantity is not a split, it is two uneconomic orders.

Making two sources produce one product

If the decision is to split, the controls are not optional extras. They are the thing that makes the split work rather than producing two similar programmes.

The last one is a control often missing and the one that catches what the others do not. Each supplier can pass inspection against the specification while the two of them drift apart from each other, and only a direct comparison shows it.

The check before splitting

That last question is worth asking properly rather than rhetorically. For a large share of programmes, the honest answer is that a genuinely qualified standby supplier, one that has made the product, holds the current specification and has been checked recently for availability, delivers most of the protection an active split would, without the permanent consistency and duplication bill.