Knitted Bag Second Source Strategy: How One Buyer Qualified a Second Floor on One Specification Without Losing the Product

The Programme That Prompted It: 40,000 Pieces, One Floor

The programme was 40,000 knitted bags a year across two styles and six colourways, running on one production floor, and it had run cleanly for two seasons before the problem appeared.

Nothing about the arrangement looked risky. Deliveries had arrived on the quoted dates, release had passed at AQL 2.5 each time, and the buyer had no complaint about the product. What the arrangement lacked was not quality but optionality: with everything on one floor, a booking conflict, a shutdown, a compliance finding or a price increase at that floor became a problem at the buyer with no alternative available. That is the risk a second source addresses, and it is worth being clear that it is a concentration risk rather than a quality risk.

The volume also mattered to the shape of the answer. At 40,000 pieces a year the buyer could split the programme meaningfully: 70% to the incumbent and 30% to a second floor gives 28,000 and 12,000, and both are large enough to be worth a floor's attention. At a 100 pcs per style and colourway minimum, the same split would be available to a far smaller programme too, which is the part most buyers underestimate. The constraint on dual sourcing is rarely the minimum; it is the qualification work.

What Actually Went Wrong, and What It Cost

The failure was not dramatic: the incumbent floor's August to November peak window was fully committed, the buyer's programme was scheduled three weeks later than quoted, and the season opened short.

That is worth dwelling on, because it is the most common shape of a concentration failure. Nothing was defective, nobody behaved badly, and no contract was breached. A 30-day bulk block that was promised for a given window could not be given, because the floor had sold the capacity to someone else and the buyer had no alternative. The cost was three weeks of a selling season and a partial air freight shipment at roughly five days of transit against about 21 by sea, on a bulky knit charged by volumetric weight, which is the most expensive way to recover a schedule.

What the buyer took from it was specific rather than general. The problem was not that the floor was unreliable; it was that the buyer had no way to test a claim about a booked window, and no alternative if the claim turned out to be optimistic. A second source addresses the second half directly and the first half indirectly, because a floor that knows a buyer can move 30% of a programme answers booking questions differently from one that knows they cannot.

Qualifying a Second Floor on One Specification

Qualification took one sample round of 7 working days and a trial order of 500 pieces, and it worked because the buyer had recorded figures rather than an impression to hand over.

This is the step that decides whether a second source is possible at all, and it depends entirely on what the first programme left behind. The buyer had a signed approval sheet with twelve recorded figures, a stitch file whose parameters were known, and a written specification. All three went to the second floor, and the result was a sample that could be compared against the incumbent's rather than merely assessed. Without them, qualification becomes a fresh development round, which is a much larger commitment and the reason most buyers never start.

Qualification sample

7 working days against the existing technical pack

Trial order

500 pieces, above the 100 pcs per colourway minimum

Documents provided

Approval sheet, stitch file parameters, written specification

Criteria matched

Gauge, GSM within about 5%, Delta E, dimensions, hardware cycles

Release standard

AQL 2.5 under ISO 2859-1, one defect list for both floors

Steady-state split

About 70% incumbent, about 30% second floor

Review point

After two runs at each floor, then annually

The Gauge and Shade Matching Problem

The one place qualification nearly failed was shade: both floors ran the same gauge at E12 and both hit the GSM target, but the two shells differed by a Delta E of about 1.2 against a target of 1.0.

That gap is small enough to be invisible to most people and large enough to be visible on a shelf where both versions sit side by side, which is exactly what happens when two floors supply the same style into one retail network. The causes are ordinary: a different yarn batch, a different dye house, a slightly different relaxation time after knitting, or a different machine setting at the same nominal gauge. None of it is a defect, and all of it produces two products that a customer can tell apart.

The fix the buyer arrived at is worth stating because it costs almost nothing. Both floors were given the same physical reference panel rather than the same Pantone code or the same dyelot number, and the shade tolerance was restated as a Delta E of 1.0 or tighter measured against that panel under a named illuminant. A physical reference removes the argument about whose reading is correct, and it is the single most useful object in a dual-source programme. A second useful measure is to run both floors on the same yarn batch where the volume allows, which removes the largest source of variation at once.

Splitting Volume: Why 70/30 and Not 50/50

The split settled at about 70/30 rather than 50/50, because a second source has to be kept qualified and 30% is roughly the smallest share that does that.

The reasoning is about capability retention rather than price. A floor running 12,000 pieces a year on a style keeps the stitch file loaded, keeps the operators familiar with the construction and keeps the yarn and trim in its ordering cycle. A floor running 2,000 does none of those things, and by the time the buyer needs it in an emergency it is effectively a new qualification again. That is the difference between a second source and a spare name on a list, and it is why an even split is not automatically more secure. A floor running 12,000 pieces keeps a full 30-day bulk cycle in its planning rather than an occasional one. Both floors shipped FOB Xiamen, so the freight and document side of the split stayed simple.

Keeping the incumbent at 70% matters for a second reason: it preserves the relationship that produces the best terms. Price, payment flexibility and booking priority at the main floor are all functions of volume, and cutting that volume in half to buy insurance tends to worsen all three. The buyer's conclusion after two seasons was that the second floor's real value was not the 30% it produced but the fact that it existed, and that the insurance was worth paying for at the smallest share that kept it real.

Holding One Defect List Across Two Floors

The buyer wrote one defect list and issued it to both floors, which is what made the two comparable: under ISO 2859-1 at AQL 2.5, the sample sizes are the same for the same lot size, so the results can be set side by side.

A lot of 281 to 500 pieces draws 50 with an acceptance number of 3, and a lot of 1,201 to 3,200 draws 125 with an acceptance number of 7. Those figures are identical at either floor, which means a defect rate can be compared directly. What cannot be compared is a pass statement issued against two different lists, which is what happens when each floor is allowed to define its own criteria. One list, held by the buyer and attached to both contracts, turns two pass results into two comparable numbers.

The comparison then becomes useful in a way that a single-source programme never is. After four runs the buyer could see that one floor had consistently lower dimensional variation and the other consistently better shade consistency, and could allocate the styles that cared about each accordingly. That is information a single source cannot produce, and it is arguably worth more than the insurance value of the arrangement.

What Dual Sourcing Actually Costs

The cost of a second source is one additional sample round, one trial order, one extra qualification audit, and a small amount of ongoing administration, and the largest of those is the first.

In money the qualification is modest: a sample round of 7 working days, a 500-piece trial order, and a compliance check on the second floor covering the same ground as the first, typically BSCI or a Sedex SMETA audit and the restricted substance position under REACH Annex XVII. In time it is about one season. What the buyer found more expensive than expected was the administration: two sets of booking conversations, two specification issue points, and two invoice and payment cycles. Keeping one specification and one defect list is what kept that manageable.

There is also a unit-cost effect worth being honest about. Splitting a programme means each floor runs a smaller quantity, and setup time is spread over fewer pieces at each. On this programme the effect was small, because a 100 pcs per colourway minimum means setup was already being amortised across colourways rather than across one large run. On a single-colourway programme of the same volume the effect would be larger and worth pricing before committing.

Keeping the Stitch File Portable

The arrangement only worked because the buyer owned the stitch file, and the clause that established that had been in the first contract for a reason that had nothing to do with dual sourcing at the time.

This is the connection worth drawing. A clause stating that the stitch file and any paid tooling belong to the buyer, with a handover obligation in a usable format, is usually justified as design protection. Its second function is that it makes a second source possible at all. Without it, the buyer would have been asking the incumbent floor to release something it had no obligation to release, which is a conversation that rarely ends well and often ends the relationship.

The practical version of portability is five parameters: gauge, loop length, yarn count, panel dimensions and the structure map. Those plus the specification were enough for the second floor to knit the same product inside one sample round. Buyers who want optionality should ask for those five at the first contract rather than at the point they want to move, and should confirm they are included in whatever file is handed over.

When a Second Source Is Not Worth It

Dual sourcing is not always worth it, and the two cases where it is not are a genuinely bespoke construction and a programme small enough that qualification exceeds the value at stake.

A bespoke construction is one whose stitch file depends on a specific machine class or a specific setup that only one floor has. In knit that is less common than buyers assume, because gauge ranges are widely distributed, but it happens with specialised jacquard or unusual gauge combinations. The test is simple: if the five parameters cannot be written down and handed over, the product is not portable and a second source is a second development. The second case is arithmetic. Qualification costs a sample round, a trial order and a season of attention; on a programme of a few hundred pieces a year, that exceeds the exposure.

For programmes in between, there is a middle option that is often overlooked: qualify a second floor without splitting volume at all. Run the trial order, hold the qualification, and place no regular volume until it is needed. It costs nearly as much and delivers less, because an unused qualification decays, but it is a reasonable first step for a buyer who wants to know the option exists before committing to two supply relationships. The buyer in this case went straight to a 30% split and would do the same again.

Frequently Asked Questions

People Also Ask

Optionality Is Bought Before It Is Needed

A second source is not a quality fix; it is insurance against concentration, and it is bought with a specification, a file and one season of attention. The two things that keep the cost down are the recorded figures from the first programme and the clause that gives the buyer the stitch file. Both are available at the first contract for nothing.

For the qualification steps in detail, see knitted bag supplier onboarding, and for the pack that both floors receive, see knitted bag RFQ template. The framework sits in the knit bag sourcing guide.