Method

DFSS — design for Six Sigma

In French: DFSS — conception pour Six Sigma

DFSS applies the Six Sigma logic to design, when there is no process yet to improve. It aims to create a product or service that is robust from the outset, rather than correcting afterwards.

Why an approach distinct from DMAIC

DMAIC starts from an existing process whose defects are measured. With nothing existing, its first three steps have no object: one does not measure what is not yet produced, and one does not analyse the causes of a defect that has not occurred.

DFSS therefore replaces the analysis of an existing process with the translation of a need into measurable characteristics, then with the anticipated verification that the design will hold them under the real conditions of use and manufacture.

The stakes are economic: a change costs little at the concept stage, a lot at the industrialisation stage, and a great deal after launch. DFSS shifts the effort upstream, where decisions are still reversible.

The steps, and their variants

Several sequences coexist — the most cited chains define, measure, analyse, design, verify; another, older one, identify, design, optimise, validate. They say the same thing in a similar order, and the choice between them is a matter of organisation more than of method.

What counts is the common logic: understand the need with the voice-of-the-customer tools, translate it into measurable technical characteristics with tolerances, explore the concepts and choose on criteria rather than by habit, optimise the parameters so that performance withstands variations, verify on a prototype or pilot before the industrial commitment.

Robustness, the heart of the approach

A robust design gives a stable result despite the variations it does not control: spread of the material, workshop temperature, tool wear, the customer’s actual use.

The idea is counter-intuitive: rather than tightening the tolerances of everything that comes in — which is expensive and degrades over time — one looks for the settings that make the output insensitive to those variations. A design of experiments is the natural tool for that search.

It is also what distinguishes a robust design from a high-performance one: the second gives an excellent result under laboratory conditions, the first a good result everywhere.

The common mistakes

Translating the customer need into specifications without going back to the customer. The slippage happens naturally: the team interprets, then interprets its interpretation, and the product ends up answering a third-hand rewording that nobody ever validated.

Launching a DFSS on a project whose concept is already fixed. The approach is worth the choices it informs; applied to a frozen design, it is reduced to documentary window-dressing.

On the shop floor — Medical devices

A locking mechanism held its specifications in the laboratory and failed in use. Characterising the real variation — force applied, grip angle, wearing of gloves — led to changing the geometry rather than tightening the manufacturing tolerances. The production cost went down.

The other tools illustrated by this trade: healthcare and life sciences.

Frequently asked questions

When should DFSS be used rather than DMAIC?

When there is nothing to improve: new product or service, process to be created, or existing process so degraded that a redesign costs less than an improvement. The practical marker: if the historical data does not exist, DMAIC has no grip.

Does DFSS apply to services?

Yes, and the approach changes little there: understand what the customer values, translate it into measurable characteristics — lead time, first-contact resolution rate, clarity of a piece of information — and test on a pilot before deploying. Verification on a restricted scope is even easier there than in industry.

What is the link with FMEA?

Design FMEA is one of the central tools of DFSS: it lists the possible failure modes before they exist, and prioritises the points to address. It comes in early, on the concepts, where its process version comes in on an existing process.

Do you have to be a Black Belt to lead a DFSS?

Not necessarily, but the approach is more demanding than a standard DMAIC: it draws on listening to the customer, robust design and designs of experiments. Organisations generally entrust it to experienced profiles, in support of a design team.