How Do You Qualify a Second Source Through Sample Approval?

The case for a second source is obvious and the case against it is less discussed. Concentration risk is real: a single production base is one fire, one regulatory event or one capacity squeeze away from a missed season. But a second source is not free. It doubles specification maintenance, doubles the sample approval cycle, and introduces the hardest problem in the category, which is making two production bases produce the same bag to the same standard.

Most second-source projects fail on that third point rather than on the first two. Buyers qualify a second source carefully, place a split order, and discover that the two runs differ in shade, hand feel or construction, because the specification was never made portable in the first place. This piece is organised around preventing that outcome: four gates, in order, with the fourth one being the gate most programmes skip.

Why Qualify a Second Source at All?

Three reasons are legitimate. Capacity: the programme is growing beyond what one base can absorb at peak. Resilience: the buyer cannot accept a single point of failure on a seasonal launch. Leverage: a buyer with one source negotiates from weakness. A fourth reason is sometimes offered — price competition between two sources — and it is the weakest of the four, because running the same specification through two bases costs more than the price difference usually returns.

Name the reason before starting, because it determines the design of the arrangement. A resilience second source can run at low volume and be activated only when needed. A capacity second source has to run continuously or it will not be ready when it is wanted. A leverage second source is a commercial decision that should be taken openly rather than disguised as a sourcing one.

What Does "Second Source" Mean: A Second Line or a Second Company?

Two different instruments with very different costs. A second line at the same site shares management, quality systems, material purchasing and often tooling, and it removes a capacity constraint without introducing a specification problem. A second company removes an entity-level risk and introduces every specification problem there is. Choosing between them is the first and most consequential decision in the project.

The test is what risk is being hedged. Where the risk is peak-season capacity, a second line is usually the better answer and costs a fraction of a second company. Where the risk is the entity itself, a second line hedges nothing. Buyers frequently build a second company to solve a capacity problem, and the arrangement then fails on cost and on consistency while solving a risk that was never the binding one.

A question worth answering before any of the four gates: which styles move. Programmes that try to dual-source everything end up with two half-learned production bases and a specification nobody fully owns. The better arrangement is to select the one or two styles with the highest volume or the highest seasonality risk and qualify those first, leaving the long tail single-sourced until the arrangement has been exercised and the maintenance cost is understood.

How Do You Keep the Specification Identical Across Two Sources?

By making it portable, which is a different thing from making it complete. A specification can be complete for one production base and useless to a second, because it relies on shared assumptions: a material known by a local trade name, a construction known by a pattern that exists physically in one place, a shade known by a physical swatch. Portability means every one of those is expressed so that a party who has never seen the first run can reproduce it.

The practical checklist is short and worth running before any second-source work begins: every material identified by a measurable specification rather than a trade name, every colour by a reference with a tolerance, every construction by a drawing with dimensions, every hardware item by brand and size, and every process parameter recorded as a value with a range. Where a field cannot be expressed that way, that field is where the two runs will diverge.

What Has to Happen at Sample Approval?

The second source approves against the retained reference unit from the first, not against the original sample and not against a photograph. That is the single most important rule in the whole project. A reference unit is a physical object both parties hold; a photograph is an impression; the original sample was made under conditions the second source cannot see. Approval against the reference is the only route that produces a comparable product.

Then run the second sample cycle on production tooling at the second source. A sample made in a sample room at the second source tells a buyer nothing about a run made on its line, and the second source has no reason to volunteer the distinction. Ask directly, in writing, how the sample will be routed, exactly as a buyer would on a pilot at a first source.

What Volume Split Makes Sense?

Asymmetric, and stable. A 70-30 or 80-20 split with a defined review keeps one source primary and the other genuinely warm, whereas a 50-50 split halves the learning at both and produces two mediocre runs instead of one good one and one adequate one. The secondary volume should be large enough to keep the specification exercised and small enough that a problem there does not become a programme problem.

The split should also be reviewed rather than fixed. Where the second source outperforms the first over four quarters, moving the split is reasonable and is good governance; moving it on the basis of one lot is not. The rule that works is a review at the quarterly meeting with the same data the scorecard uses, and a change to the split only on a trend rather than on a single result.

How Do You Run the Qualification Without Doubling the Work?

By reusing every artefact. The tech pack, the inspection plan, the defect class list, the carton specification and the document list should all be transferred unchanged, and any change made for the second source should be made to the shared document rather than to a copy. Two versions of a specification is the single most expensive mistake available in this project, and it is made by default unless somebody forbids it.

The second saving is in sampling. The second source does not need a full development cycle; it needs a confirmation cycle. Where the specification is genuinely portable, the approval should take one round rather than three, because there is nothing to develop — only something to match. If approval at the second source takes as long as it did at the first, the specification was not portable and that is the problem to fix.

What Should Be Shared With the Second Source, and What Should Not?

Shared: the full technical specification, the inspection plan, the reference unit, and the quality expectations. Not shared: the first source pricing, the first source identity where confidentiality applies, and anything the buyer holds under a restriction from the first source. The technical package travels; the commercial package does not.

The confidentiality point is frequently mishandled in both directions. Buyers sometimes withhold the full specification to protect the first source, which guarantees the second source cannot match the product. Others share pricing, which damages the first relationship and usually produces a second source that prices against the first rather than against the work. The technical package travels, the commercial package stays.

How Do You Compare Two Sources Honestly?

With the same scorecard, the same weighting and the same data definitions, entered by the same person. A comparison built on two different instruments is not a comparison. Where a second source is being assessed against a first, the most common distortion is that the first is scored on two years of history and the second on two lots; the fix is to compare like periods, which usually means starting the second scorecard from zero and waiting four quarters.

The second distortion is novelty. A new source performs well in its first two quarters, because everybody is paying attention. The honest comparison is at quarter four, once the attention has moved elsewhere, and a buyer who decides at quarter one is measuring enthusiasm rather than performance.

One cost is consistently underestimated: the inspection regime. Two sources mean two inspection bookings, two defect dictionaries in use and two sets of photographs to compare, and on a small programme that overhead is real. It is manageable, but it should be counted before the arrangement starts rather than discovered in the first quarter, because an inspection regime that is quietly halved to save money removes the only mechanism that would have caught a divergence.

What Governance Changes Once There Are Two Sources?

Three things. The specification becomes a controlled document with a version number, because two parties are building from it and an unversioned change will be applied at one and not the other. The scorecard doubles, with a shared definition set. And the review cadence tightens, because a divergence between two sources is only visible when both are looked at in the same meeting.

One further change is worth making explicitly: name which source is primary for each style. Where both sources can make everything, both will, and the specification drifts at both. Where each style has a named primary and a named secondary, the secondary sees occasional volume and the primary holds the standard, which is the arrangement that keeps two bases consistent over years.

One governance detail is worth fixing at the outset and rarely is: who owns the specification. With two sources and one document, the document needs a single named owner on the buyer side, because a specification that two people can edit will be edited by both and the versions will diverge within a quarter. The owner does not have to make every change; they have to record every change. That distinction is small in principle and decisive in practice.

The commercial frame does not change with a second source. MOQ 500 pcs per style and colourway applies at each source; sampling runs 6-10 working days from a signed-off tech pack; bulk production runs 35-50 days from written sample approval; inspection sits at AQL 2.5 under ISO 2859-1 General Inspection Level II; terms are T/T 30/70 quoted FOB Xiamen. What a second source adds is resilience, and what it costs is specification discipline — which most programmes need anyway.

There is also a question of what the first source is told. Concealing a second source until it is activated tends to fail, because the first source notices the volume pattern long before the buyer says anything and fills the silence with its own explanation. The better approach is to state the reason plainly — capacity, resilience, or both — and to keep the primary volume visible. A first source that understands why a second exists usually competes on service; one that discovers it later usually competes on price and then leaves.

A final note on timing. Qualifying a second source during a peak season is the worst possible moment, because nobody has attention to spare and the qualification will be done badly. The right moment is a quiet quarter, with a real but small order attached, so that the arrangement has been exercised at least once before it is needed. A second source qualified in a hurry and never used is not resilience; it is a file entry.

Frequently Asked Questions

People Also Ask