Second sourcing, done properly

Dual sourcing fails for a structural reason, not a lack of conviction: the cost is immediate and specific, the benefit is deferred and probabilistic. Programmes that survive are the ones that stop treating it as a project and make it a maintenance routine on a short, scored list.

Score criticality before you do anything

Dual-sourcing everything is its own form of waste — it doubles qualification work, halves your volume leverage on each source and multiplies the QA surface. The first job is to shorten the list. Score every raw material on four axes and only work on the top of the pile:

Substitutability. Could you reformulate around it in a month? Commodity carbonate, yes. A specific matting agent your customer approved, no.
Source concentration. One plant, one quarry, one country? A single site is a single point of failure regardless of how large its owner is.
Consequence. What does one week without it cost — a line, a customer, a contractual penalty?
Exposure. Is the origin sensitive to freight, energy, sanctions or a regulatory review already underway?

In most portfolios this produces a list of five to fifteen materials, not fifty. That list is workable.

What "qualified" has to mean

A name in a spreadsheet is not a second source. The test is whether you could switch on Monday without convening a meeting. That requires:

A production trial, not a lab plaque. Alternates most often fail on dispersion behaviour at plant shear, on dosing equipment, or on something mundane like bulk density in a hopper — none of which a lab disperser reveals.
Documented equivalence, and documented non-equivalence. Where the alternate is genuinely different, write down how. That note is what prevents a bad decision under time pressure.
Pre-agreed formulation adjustment. If the alternate needs 3% more loading or a different wetting agent, that change should be approved and sitting in the file, not invented during a shortage.
Commercial terms in principle — including what the price does if you arrive as an emergency rather than a planned customer.

If the switch requires a decision no one has pre-approved, you do not have a second source. You have a phone number.

The paperwork nobody plans for

This is where second-source projects die quietly. Before an alternate is usable you may need: an updated safety data sheet and classification, confirmation of REACH registration status covering your tonnage and use, food, feed or pharmaceutical grade declarations, and — the one that catches people — your own customers' approval, where their specification names the raw material or its origin.

That last point is worth checking early. In regulated chains, a raw material change can trigger a customer requalification that takes longer than your own. Discovering this during a shortage means the alternate you carefully qualified cannot legally be used.

What it costs, honestly

For a typical mineral or additive: a few days of laboratory work, one production trial including the off-spec material it generates, and the internal effort to document and file it. Call it a small number of thousands of euros and a fortnight of somebody's attention, per material.

Against that: one week of unplanned downtime on a single line, or one quarter of paying an allocation premium because you had nothing to point at in the negotiation. The arithmetic is not close for anything on a properly scored criticality list.

An annual cadence

Once, done thoroughly, then reviewed. A workable rhythm:

Re-score the criticality list annually — consumption and origins move.
Qualify two or three alternates a year rather than attempting a programme.
Re-confirm existing alternates every two years; producers change grades and close plants without telling their non-customers.
Run one live switch as an exercise, on a low-risk material, to find out whether the paperwork actually works.

The negotiation dividend

The under-appreciated return is commercial rather than operational. A buyer with a genuinely qualified alternate negotiates differently, and the other side knows it. In our experience that shows up in price and in priority long before it shows up in a disruption — which is what makes the programme pay for itself in years when nothing goes wrong.