A changeover is more than the minutes when a machine is stopped. It includes the preparation and coordination that determine how quickly a line can move from its last good part on one run to its first good part on the next. Without a clear method, teams can focus on speeding up the stopped portion while overlooking work that could happen beforehand.
SMED in manufacturing is a focused setup-reduction method. It separates tasks requiring stopped equipment from work that can happen while it runs. Teams then convert or streamline tasks where safe and practical. Its name describes a single-digit-minute aspiration, not a guaranteed result for every machine or process.
For discrete manufacturers, the value is in making setup work visible and repeatable, not chasing a universal stopwatch target. The method starts by looking closely at what operators actually do and why each step needs to happen when it does.
What SMED in Manufacturing Means
SMED, or Single-Minute Exchange of Die, is a structured method for reducing time between manufacturing runs. Despite the name, it is not limited to die changes. The method applies to equipment and workstations where setup interrupts production. Teams examine the work inside a changeover and identify which activities require stopped equipment. They then improve how remaining tasks are prepared and performed. The University of Texas Rio Grande Valley’s Manufacturing Assistance Center describes SMED as a way to convert a process from running the current product to running the next one.
The central distinction is between internal and external setup work. Internal tasks can only be completed while the machine or process is stopped. External tasks can be done while it is still running. For example, a team might examine whether tools, materials, or the next job’s setup information can be prepared in advance, rather than gathering them after production has stopped. Those are possibilities to investigate, not automatic classifications: the actual task, equipment, safety requirements, and operating procedure determine whether work can be moved outside the stopped window.
| Setup type | Machine state required | SMED question |
|---|---|---|
| Internal | Equipment is stopped | Can the task be simplified or safely prepared earlier? |
| External | Equipment can keep running | Can this task be completed before the changeover begins? |
SMED is not simply a demand to make operators move faster. Teams study the sequence, separate internal from external work, convert feasible tasks to external preparation, and simplify the work that must remain internal. The aim is to reduce avoidable time in the stopped-equipment portion of a changeover while keeping the process controlled. If a task cannot safely or reliably be done while the machine runs, it should not be reclassified just to improve a time measure.
The name “single-minute” refers to an aspiration: bringing a changeover into single-digit minutes, or under ten minutes. It is a goal associated with the method, not a universal guarantee or a suitable promise for every machine, product mix, or plant. A team should first define what counts as the changeover, observe its actual work, and identify a realistic improvement opportunity for the process it is studying. Progress can still matter even when a setup does not reach the single-digit target.
The distinction between internal and external work gives teams a practical way to analyze a stoppage. It avoids treating all setup activity as one undifferentiated block. It also makes preparation visible. Work that can happen beforehand still needs an owner, the right materials, and a clear standard. Otherwise, it can become a new source of delay or confusion. The sequence and constraints vary by process, so teams should observe the work rather than assume a task is transferable.
This makes SMED more specific than general advice to reduce setup delays. JobPack’s general changeover reduction guide covers the broader topic; this article focuses on applying SMED’s internal/external distinction and improvement logic in a manufacturing setting. For the method’s core definitions, see the Lean Enterprise Institute’s SMED glossary.
How to Separate Internal and External Setup Work
Classify each setup element by the machine state it requires, not by where it usually appears on a checklist. Internal work requires stopped equipment; external work can happen while it runs. Confirm the distinction at the machine because requirements and safety controls vary by process.
- Observe a normal changeover. Watch from the end of one run through the start of acceptable production on the next. Record the sequence, who performs each action, tools or materials used, and when the machine runs or stops. Operators can identify workarounds a written procedure may miss. Follow site policy and get permission before recording; notes are an alternative.
- Break the sequence into clear elements. Replace broad labels such as “change tooling” with observable actions: retrieving a tool, removing a component, adjusting, or checking output. Keep order and handoffs visible. This helps expose waiting, searching, travel, and repeated handling.
- Classify each element by its requirement. Ask whether the machine must stop for the action, or whether it can happen before shutdown while the prior run continues. Mark work internal only when the machine state is genuinely required. Checking tools, confirming parts, and staging materials may be external if the work area and controls allow it. Usual timing alone does not make a task internal.
- Challenge internal tasks cautiously. Ask what makes each action machine-dependent: access, temperature, energy isolation, guarding, sequence, or a quality check. Consider whether a safe fixture, advance preparation, or revised sequence could remove that dependency. Involve operators, maintenance, quality, and safety personnel as needed. Do not move work outside the stopped window if it exposes people to moving equipment, bypasses safeguards, or weakens inspection or product quality. Validate methods against site procedures before a trial.
- Convert feasible tasks, then simplify the rest. Test proposed external work under controlled conditions and check for new risks, errors, or extra work elsewhere. Keep tasks that still require stopped equipment internal. Simplify those by reducing unnecessary adjustments and motion, making tools and materials easier to reach, and clarifying responsibilities. Parallel work is useful only when roles and safe access are clear.
- Document and verify the revised method. Update the sequence, checklist, and training materials so the distinction is repeatable across shifts. Observe the revised changeover and confirm correct equipment preparation, effective safety controls, and acceptable first output. If quality or safety suffers, stop and restore the approved method while the team reviews it.
A Practical SMED Implementation for a Pilot Machine
A useful pilot turns SMED from a workshop idea into a repeatable way to change over one machine. Start by choosing a setup that occurs often, causes a meaningful production problem, and is practical to study. A dramatic but rare changeover may be less informative than a routine one that repeatedly disrupts the schedule. Check that the team can observe the work and test changes without compromising safety or customer requirements.
Bring the people who perform and support the changeover into the work from the beginning. Operators can explain why a step is done a certain way, which tools or materials are often missing, and where instructions differ between shifts. Include maintenance, quality, or material-handling staff when their work affects the sequence. Employee engagement was identified as an important lesson in a NIST manufacturing improvement case, reinforcing the value of involving the people closest to the process: NIST’s Modelcraft case study.
Before changing anything, agree on what counts as the setup and record the current method. Observe multiple representative runs when possible, and document the ordered work elements, who performs each one, when it starts and ends, and whether the machine must be stopped. Use the same start and finish conditions each time, and note delays, adjustments, missing items, and quality checks rather than folding them into a single unexplained duration. This baseline helps the team distinguish a genuine process change from normal variation.
Next, review each element with the operators. Classify it as internal if it requires stopped equipment, or external if it can be done while the machine is running. Challenge assumptions, but do not reclassify a task until the team confirms that it can be completed safely and correctly outside the stopped window. Prioritize a small number of feasible changes: stage verified tools and materials in advance, clarify responsibilities, or simplify a recurring adjustment. Test changes deliberately and record what changed, instead of altering several parts of the method at once and losing the ability to learn what helped.
For each trial, confirm that the revised sequence meets safety procedures, produces an acceptable first part, and does not create extra work elsewhere. If a change fails a check, revise it or return to the previous method while the team investigates. Once a sequence works reliably, update the setup instructions or checklist with the actual order, required materials, checks, and role assignments. Train affected shifts and review subsequent changeovers to confirm the method is repeatable, not just fast when the pilot team is present.
This machine-level experiment is more structured than a general list of changeover tips: it establishes a baseline, tests specific SMED changes, and verifies the new standard in production. For broader ideas that are not specific to a pilot method, see JobPack’s general changeover reduction guide.
The SMED opportunity becomes clearer when setup tasks are measured alongside machine time and schedule assumptions. Request a demo to see how JobPack supports production scheduling and shop-floor visibility.
How Should Manufacturers Measure Changeover Performance?
Start with a boundary everyone applies the same way: from the last good part of the outgoing run to the first good part of the next run. The new part must meet quality requirements. This interval includes setup work, adjustments, and delays before acceptable production resumes. NIST uses this last-good-piece to first-good-piece definition for quick changeover measurement: NIST’s quick changeover guidance.
Before a SMED trial, document the current process using that definition. Keep the measure tied to the same machine and changeover context, such as the same product pair, tooling, and operating conditions. If those factors vary, record them rather than treating every event as directly comparable. This helps distinguish a change in the method from a change in what was being produced.
Record elapsed time and what happened during it
Capture the total elapsed interval, then note the major tasks and delays. For each activity, record its start and finish where practical, who or what was involved, and whether it required stopped equipment. Separate hands-on work from waiting, searching, transport, adjustment, and troubleshooting.
Keep planned setup separate from unrelated downtime, such as a material shortage or an unplanned equipment fault. Record the reason and set a clear rule for whether an event is included in the changeover interval or reported separately. Apply that rule before and after the trial. A downtime log can help teams track setup-related downtime without blending distinct problems into one number.
Compare like with like, including startup quality
Observe repeated changeovers, not just one favorable run. Compare the same machine and, where possible, the same product transition and shift conditions. Review both elapsed time and the spread between runs: a process that is fast once but difficult to repeat may need clearer standard work or better preparation. Operators can also identify missing steps or unusual conditions that a timestamp alone will not explain.
Time is not the only acceptance check. Confirm that the first good part meets specifications and that the new sequence preserves safe work practices. Include defects, re-adjustments, or a delayed restart in the review instead of treating a quick restart as success by itself. This guards against trading apparent speed for unstable startup or risk.
After a change, compare trial observations with the baseline under similar conditions, then examine which work and wait causes changed. Treat the result as evidence for the next decision, not a promise that every machine will achieve the same reduction. Setup losses can affect availability and how changeover affects OEE, but keep this measure focused on the changeover itself. For planning work around expected setup duration, see how to schedule jobs around setup times.
Connect SMED Gains to Scheduling and Capacity
A shorter setup matters to the production plan only when the schedule reflects what the team can now repeat. If the system still reserves the old setup duration, planners may continue to treat usable machine time as unavailable. If the revised estimate is too optimistic, the schedule can instead stack work more tightly than operators can deliver. Use observed, repeatable results to update the planning assumption, and keep any uncertainty visible while the new method is being established.
In a finite-capacity schedule, setup time consumes capacity at a specific machine or work center. Accurate estimates help planners compare workload with available capacity. They can then identify due-date risks and decide whether to resequence or move work. A setup reduction validated on one machine and product family should not automatically apply to every changeover. Keep assumptions specific to the measured conditions.
Sequencing also affects how setup work fits between jobs. Grouping similar jobs may avoid some changes, but could delay another order or require a larger batch. Reliable setup assumptions help the team compare tradeoffs: keep the sequence, arrange work differently, or test a smaller lot. Each option must fit delivery and capacity needs. The right choice depends on demand, due dates, routing, and time at constrained resources, not a universal batch-size rule.
SMED can make more frequent product changes practical to consider, but it does not by itself establish that smaller batches are economical or feasible. Before changing lot sizes, check whether the measured setup method holds across the relevant operators, shifts, materials, and product variants. Account for remaining internal work, external preparation, startup checks, and any time needed to confirm the first good part. If results vary, schedule with a conservative assumption and gather more observations rather than treating the best run as the norm.
When a method has been validated, update the setup-time data used for planning and note its scope and review date. Then use the revised assumption in a controlled schedule review. Compare planned and actual setup duration, capacity loading, and delivery performance over comparable work. If actual setup time drifts, revisit the standard before relying on it for tighter sequences or smaller lots. This makes the scheduling decision traceable to shop-floor evidence instead of an aspirational target.
For a practical walkthrough of the planning inputs, see how to schedule jobs around setup times. Teams evaluating capacity-aware planning can also review production scheduling with setup times, while keeping the measured SMED method and its assumptions current.
How Do You Sustain SMED Improvements?
A faster changeover is not sustained by a one-time workshop. The revised method has to remain clear, practical, and repeatable when the usual operators are unavailable, the product mix changes, or production pressure rises. Treat the improved setup as controlled standard work, then check whether actual practice still matches it.
Make the method visible and owned
Document the agreed sequence in a concise work instruction or checklist. Identify which tasks happen before the machine stops and which require stopped equipment. List the tools or materials that must be ready, along with the checks that confirm readiness for the next good part. Keep instructions specific while preserving approved safety and quality requirements. UTRGV’s SMED guidance also describes checklists and tool function checks: SMED setup guidance.
Name an owner for keeping the standard current, such as a supervisor or continuous-improvement lead, and involve the operators who perform the changeover. The owner can review exceptions, collect suggestions, and coordinate approval when a step, fixture, or sequence needs to change. A checklist should support the work, not become a box-ticking substitute for observing it.
Train across shifts and verify the actual method
Train each shift on the revised sequence and explain why internal work must be done with equipment stopped while external preparation can happen while it runs. Have operators demonstrate the steps, clarify handoffs, and surface differences between the documented method and real conditions. Repeat observations after the change is introduced and at routine review points. Compare the same start and end points and note task delays, missing materials, adjustments, or quality concerns.
When performance drifts, investigate before resetting the target. The product, tooling, staffing, machine condition, or sequence may have changed, or the old setup-time assumption may no longer fit the work. Observe the current process again, confirm which tasks remain necessary and safe, then revise the standard and retrain affected shifts. This keeps improvements grounded in current shop-floor conditions rather than relying on an outdated best-case run.
Use production data as evidence, not an automatic SMED diagnosis
Machine-state and event records can show whether equipment is running, idle, or in an alarm condition. User-defined activity codes can record operator-entered reasons for downtime. JobPack’s machine-monitoring tools provide this kind of visibility to monitor machine downtime and utilization. Shop-floor records can also capture actual shop-floor activity through automated machine capture or manual operator input. These records can help flag patterns for review, but they do not automatically identify SMED tasks or analyze the method. The team still needs to observe and validate the setup.
After a revised setup time is demonstrated consistently and quality requirements are met, production schedulers can review the setup assumptions used in capacity plans. Production scheduling with setup times can reflect approved planning assumptions, while operators and process owners remain responsible for maintaining and verifying the standard work.
See how production scheduling, machine monitoring, and shop-floor data collection may support a more visible manufacturing operation. Request a demo.
Looking to connect scheduling and shop-floor information? Request a demo.
Frequently Asked Questions
What are the three stages of SMED?
The stages are to separate internal setup work from external work, convert as many internal tasks as practical to external tasks, then streamline the work that remains. Internal tasks require the machine to be stopped; external tasks can be completed while it is running. Any change must remain safe and protect process quality.
What are some examples of SMED practices?
A team might stage the next job’s tools and materials before stopping the machine, verify tools in advance, and use a checklist so required items are ready. It can also review whether a setup task can safely happen while production continues, or whether a repeatable locating method can reduce adjustments. UTRGV’s manufacturing training describes checklists, tool checks, advance preparation, and intermediary jigs as setup-reduction practices (UTRGV training).
Is SMED a Lean method or a Six Sigma tool?
SMED is generally treated as a Lean setup-reduction method. It focuses on observing a changeover, distinguishing internal from external work, and simplifying the process. A manufacturer can use it alongside Six Sigma or other improvement approaches, but SMED itself is not a Six Sigma tool.
Is SMED part of TPM?
SMED is not a required component of Total Productive Maintenance (TPM). They address different but related concerns: SMED improves the changeover process, while TPM focuses on equipment effectiveness and maintenance. A plant can coordinate both, for example by including safe equipment-condition checks in standardized setup work without confusing maintenance tasks with setup-reduction tasks.
JobPack connects production scheduling, machine monitoring, and shop-floor data collection for manufacturers that want clearer operating information. If your team is assessing how those capabilities fit its planning and visibility needs, Request a demo.
