Dual-Sourcing Precision Parts Across India and Vietnam: A Buyer's Operating Guide

A buyer with a qualified Vietnam supplier for a family of aluminum brackets reads about a fire at a factory down the road from their supplier's gate. Not their supplier. Close enough that it easily could have been. That is the moment the real question stops being abstract: not "is Vietnam a good place to manufacture," but what happens to this order if a bad week hits this specific shop, a fire, a flood, a labor dispute, a capacity crunch it cannot absorb.

A second supplier down the same road does not answer that question. Same grid, same port, same regional labor pool, same monsoon season, the disruption that takes out one can take out both. That is what actually pushes a buyer toward dual-sourcing: not swapping Vietnam for India, but qualifying a second source whose risk exposure does not overlap with the first, for the same part family. Done properly, it means splitting volume on a defined allocation, running parallel qualification instead of a full restart, deciding whether tooling gets duplicated or transferred, verifying two quality systems against the same standard and agreeing in advance what triggers a shift in volume. It is not a hypothetical at any scale either: Apple has openly split iPhone and iPad/Apple Watch assembly between India and Vietnam since 2022 rather than betting on one country, proof that running parallel manufacturing across two countries is an established, workable decision, not a fringe idea.

Two sources have to be independent not just far apart

The reason to run two sources is to stop one disruption from taking out both. That only works if the two sources are not exposed to the same risk and distance alone does not guarantee that.

The clearest lesson on this comes from outside precision manufacturing entirely. In 2011, severe flooding hit Thailand's industrial zones, which accounted for a large share of the world's hard-drive production. Four of the world's five largest hard-drive makers had factories or key suppliers concentrated in the same flooded region. Buyers who thought they had a second source often found their "second" supplier or a shared upstream supplier, sat in the same flood zone. MIT Sloan Management Review's Yossi Sheffi used this case to make a point that still holds: dual sourcing only reduces risk when the two sources do not share the same exposure. Two suppliers in the same industrial cluster or on the same power grid are not a real hedge, even with different addresses.

For precision parts specifically, that shared exposure is easy to miss upstream: a shared foundry or forging source, plating or heat-treatment house or raw-material mill can sit behind two factories that otherwise look fully independent. The practical check: different power grids and port dependencies, no shared upstream supplier at any of those steps and different weather and seismic exposure. None of this is about which factory makes a better part, it is about whether one event can plausibly hit both at once. Different countries also bring their own customs and transit dependencies, which belong in the backup lead-time calculation rather than being assumed away by the border itself.

Splitting the volume: four models not one default

Once a second source is qualified, the next decision is how much of the part family goes to each one. There is no single correct split. Four models cover most real situations.

A volume-weighted split is often a practical starting model precisely because it does not require guessing a final number up front. A new source, however well it performs in qualification, has no delivery history yet. Weighting it lighter at first and increasing its share as real orders get scored is the low-risk way to test it without betting the primary supplier's business on day one.

Parallel qualification not a lighter copy

The biggest operational mistake in dual-sourcing is treating the second source's qualification as a faster, lighter copy of the first. It should run to the same standard, in parallel, not diluted.

The automotive industry's Production Part Approval Process (PPAP), maintained by AIAG, is a useful structural reference even outside automotive, because it already solves "how do I prove a specific site can make this specific part." Its required submission level is customer-specific, so a buyer should define deliberately what evidence each manufacturing site must provide, rather than letting the second source default to something lighter simply because it came second. The precision turned parts qualification guide covers PPAP mechanics in depth.

Running qualification in parallel means the same document set, first-article standard and sign-off criteria apply to both sites, submitted and reviewed independently. Each site retains its own approved reference sample and production records against its own tooling, per the buyer's control plan, not a shared file borrowed from the first site. If the buyer's approved-supplier checklist exists, it runs twice, on the same schedule, with no shortcut for the second entry.

Tooling - duplicate it or plan a transfer

Tooling is where dual-sourcing plans most often stay vague and vagueness here is expensive later. There are two real options and the mistake is not choosing between them before a disruption forces the question.

Duplicate tooling means both sites get their own set, built to the same drawing. It costs more up front and it removes the tool-transfer step entirely, but the second site still needs a validated production process, first article and whatever customer approvals apply before it can actually take production. Duplicated tooling shortens the path to activation. It does not skip it.

Single tooling with a transfer plan means one master tool exists, normally at the primary source, with a documented plan to move it if activated. This is cheaper but slower, since a move involves shipping, reinstallation and a requalification run. Before relying on a transfer plan, confirm who legally owns the tooling, where it is physically held, what documentation and spare components exist and whether the supply agreement actually gives the buyer the right to transfer or duplicate it. A transfer plan is only as real as that paperwork.

For a risk-tier priority part family, duplicated tooling is usually worth the extra cost. For lower-criticality parts, a transfer plan reviewed at least annually is a reasonable middle ground.

Tooling readiness is not the same as capacity readiness. A second source can be fully qualified and still unable to absorb an emergency allocation, if its bottleneck operation is already running near full utilization, a heat-treatment or plating subcontractor it depends on has a long queue or only one machine on its floor can hold the critical tolerance. Before counting on a second source in a crisis, verify its available capacity at the bottleneck operation and a realistic ramp-up time, not just whether it can make the part at all.

Two QMS landscapes one standard to check against

A common assumption is that qualifying suppliers in two countries means learning two different quality systems. For the certifications that matter most here, that is not quite right.

IATF 16949, the automotive standard and ISO 9001, the general one, are internationally applicable standards, not different country editions of the same idea. Certification is performed by independent, accredited bodies under the applicable arrangement, not one single global authority, so the buyer's job is not translating between two versions of a standard. It is independently verifying each site: certificate scope actually covers the process and product being sourced, the certified site is the one making the parts and the certification body's own accreditation can be checked.

What actually differs between the two sites is everything the standard itself does not cover: the buyer's own purchasing controls have to extend to each supplier individually. ISO 9001 requires a certified organization to control and evaluate its externally provided processes and products, which means the approved-supplier file needs two independent, current audit trails, not one shared file with two names on it. And a shared certificate framework does not make two factories equivalent: the buyer still has to compare actual process capability, equipment, inspection controls and production capacity. A buyer who treats the second source's file as a lighter copy of the first has quietly reintroduced the single-source risk they were trying to remove.

A real second source passes three tests

Qualified means the site has demonstrated it can make the part to the drawing and standard, the PPAP-or-equivalent discipline covered above. Ready means the tooling, fixtures, programs, material and inspection method actually exist at that site and it has real spare capacity to take volume rather than a capable machine that is already booked solid. Activatable means the trigger, decision owner and initial volume split are written down before a disruption forces the question. A source can pass the first test and still fail the other two. Qualified is not the same as usable.

Signs a part family is not ready to be dual-sourced yet

A quick check before setting up a second source, mapped to the three tests above.

• Nobody has confirmed whether the Vietnam supplier's upstream inputs (raw material, plating, heat treatment) overlap with the India candidate under evaluation.
• The drawing and tolerance package has never been sent to two shops at once and compared for how each one reads it.
• There is no current, complete tooling record for the part, which makes a transfer-based plan unworkable if it is ever needed.
• Nobody has written down what event would trigger shifting volume, so a disruption forces the decision under pressure instead of by an agreed plan.
• The part has never had its criticality ranked against the buyer's other SKUs.
• The approved-supplier file has one active audit trail for this part family, not two.

If most of these are true, this is the groundwork to fix first, not a reason to drop the idea.

Setting the activation trigger before you need it

The hardest part of dual-sourcing is not qualifying the second source. It is deciding, mid-disruption, how much volume moves and how fast. Buyers who wait to decide in the moment tend to either overreact, shifting everything and starving a supplier they still need next year or freeze while the primary source's problem gets worse.

Toyota's post-2011 changes are a useful reference for the shape of a good answer, not because a small buyer needs Toyota's scale, but because the structure translates. After the 2011 earthquake exposed how much of its production depended on suppliers it could not see past the first tier, Toyota built ongoing visibility into which suppliers and parts were at risk early in a developing crisis and required extra buffer inventory for its most critical components.

A much smaller buyer can copy the structure without the scale: write down, before a disruption happens, the specific conditions that count as a trigger for each dual-sourced part family (a lead-time overrun past an agreed threshold, a confirmed quality escape, a declared force majeure, not a vague feeling that something is off), who has authority to activate the shift and what share of volume moves at first versus what ramps in later. None of those numbers are universal. They belong in the buyer's own supply agreement, set for that specific part family.

None of this works if the two quotes are not on the same basis. Normalize the commercial terms before treating the sources as interchangeable: Incoterm, freight, duties, packaging, MOQ, payment terms, lead time and currency. A second source that is technically ready but commercially incomparable is hard to activate quickly, because nobody actually knows what shifting volume there costs.

What this does not mean

This is not a case for moving away from a Vietnam supplier that is performing well or a claim that India is the right second source for every buyer. It is also not a reason to treat a second source as backup-only, with no real orders. A source that exists only on paper is not actually ready when it is needed and the volume-weighted model above exists precisely so the second source gets real orders instead of a qualification file that quietly goes stale.

Where this leaves you

Dual-sourcing across two countries is an operational project, not a comparative judgment. Pick an allocation model on purpose, qualify the second source at full strength, make it ready, not just qualified and write the activation trigger down before a disruption forces the decision.

For the current single-source qualification process, the five-check remote qualification guide covers the baseline.

Related reading

• Sourcing Injection Molded Components
• Sourcing Forgings from India
• Sourcing Investment Castings
• Sourcing CNC Machined Components Buyer Guide
• Rajkot - India’s One of the Cluster for Casting, Forging and Machining Ecosystem