SMED (Single-Minute Exchange of Die): Definition & 7 Steps

What is SMED (Single-Minute Exchange of Die)?

SMED, short for Single-Minute Exchange of Die, is a lean manufacturing method to reduce machine changeover time to less than ten minutes, which in Japanese is expressed as a single-digit number of minutes. The method was developed by Japanese industrial engineer Shigeo Shingo at Toyota and Mazda between 1950 and 1969, and it is the technical backbone of every modern pull system and small-batch production strategy.

Changeover time is measured from the last good part of the old product to the first good part of the new product at target speed and target quality. Everything in between counts, including the walk to the tool crib, the search for the torque wrench, the first-off inspection and any adjustments after the restart. This strict definition is what separates SMED from informal setup improvement exercises.

The economic logic is brutal. A press with a 90-minute changeover and 300 € per hour of production value loses 450 € in direct capacity per change. Cut the changeover to 9 minutes and the same machine runs smaller batches profitably, inventory drops, lead time shrinks and OEE rises. SMED does not just save time. It unlocks a completely different production economics.

Internal vs. external setup: the single most important distinction

Shingo's core insight was that setup work falls into two categories, and most plants mix them up constantly. Once you separate them, half of the improvement comes for free.

In an unimproved changeover, roughly 30 to 50 % of what operators do while the machine is stopped is actually external work that was simply postponed. That is pure, recoverable downtime.

The seven steps of a SMED project

Shingo's original method and its modern extensions boil down to seven concrete steps. They are sequential on paper and iterative in practice.

The benchmark from the original Toyota case is legendary: a stamping press changeover that took four hours in the early 1950s dropped to three minutes by 1969. Most of that came from steps 2 and 3, not from exotic hardware.

Beyond SMED: OTED and the modern tool landscape

Shingo himself later formulated One-Touch Exchange of Die (OTED) as the aspiration to cut changeover below 100 seconds, and No-Touch Exchange of Die (NTED) as fully automated tool changes. They are not marketing terms. CNC machining centres with automatic tool changers, IMM with quick-coupling systems and robotic die handling have made both realistic for a growing number of processes.

For most mid-sized manufacturers, OTED on bottleneck equipment and disciplined SMED on everything else is the economically rational target. Full NTED only pays off on machines with very high changeover frequency.

How a cloud MES makes SMED real

Every SMED project starts strong and then quietly regresses. The root cause is almost always the same: nobody measures changeover time continuously, so the discipline erodes within six months. This is where a Manufacturing Execution System turns a workshop exercise into a permanent capability.

At Meleghy Automotive, press-line changeovers are tracked per order and per article pair across six plants. The MES captures every cycle via SAP IDoc integration, so setup time is not a project metric, it is a shift-level KPI. That is how the plant locked in a 10 % reduction in downtime and a 7 % output improvement, and more importantly, held it.

SMED in the real world: what the numbers look like

The gap between a good and a mediocre SMED programme is usually an order of magnitude. Three representative anchor points from mid-size European manufacturers:

FAQ

Does the "single minute" in SMED really mean under 60 seconds?
No. "Single minute" refers to a single-digit number of minutes, meaning less than ten. Sub-60-second changeovers are covered by the follow-on concept OTED (One-Touch Exchange of Die). The distinction matters in audits and coaching because setting a 60-second target on a press that today takes two hours demoralizes the team on day one.

Where does SMED show up in OEE?
In the Availability factor. Changeover time is planned downtime in some OEE definitions and unplanned in others, but in every credible calculation it counts against the machine. A plant that cuts a 90-minute changeover to 9 minutes on a machine with five changeovers per day recovers 6.75 hours. On a 16-hour day, that is a direct 42 % Availability uplift on top of whatever else OEE work delivers.

Do we need SMED if our changeovers are already short?
The test is economic, not absolute. If a changeover is short enough that lot sizes can be reduced without hurting capacity, and inventory is already where you want it, further SMED has diminishing returns. For most mid-sized manufacturers the opposite is true: lot sizes are kept artificially large because nobody has ever properly attacked changeover time, and finished-goods inventory absorbs the penalty.

How long does a SMED project take?
A structured kaizen event on a single machine typically runs one week: two days of observation and video analysis, two days of implementation and trials, one day of standardization. A 30 to 50 % reduction in this first pass is normal. Getting below ten minutes usually requires a second pass with hardware investments (quick clamps, zero-point systems). The sustaining layer, and this is where an MES earns its keep, is permanent.

Is SMED compatible with regulated industries like pharma?
Yes, with added discipline. GMP environments impose cleaning, line clearance and documentation steps that cannot simply be parallelized away. The SMED logic still applies: separate what can be external, choreograph what must be internal, and digitize the documentation so it stops being the bottleneck. Klocke runs exactly this model on its blister and sachet lines with SYMESTIC, capturing every changeover against the production order while staying GMP-compliant.

Related: Lean Production · Kaizen · Muda · Takt Time · Heijunka · TPM · OEE

About the author

Christian Fieg

Head of Sales at SYMESTIC. Six Sigma Black Belt, 25+ years in automotive manufacturing. Led SMED programs on headliner, injection moulding and stamping lines at Johnson Controls across plants in China, Mexico, the US and Europe. Author of "OEE: Eine Zahl, viele Lügen". · LinkedIn