The changeover that was supposed to take 20 minutes
The production line stopped at 10:43. The changeover was planned. The standard time was 20 minutes. Materials were ready. The schedule looked achievable. In theory, this was routine. The operator logged the stop at 11:05, because that's when he finally had a moment to walk to the terminal. When the machine finally restarted, the clock read 12:57. The system recorded a 112-minute changeover. The standard is 20.
The discussion started the next morning. What happened? Nobody was quite sure. Someone said there had been a material issue. Someone else thought the tooling setup took longer than expected. Someone mentioned that the same thing happened last Thursday, but nobody had written it down. The shift supervisor said it felt like about an hour. The operator thought it was closer to 90 minutes. The production report said 112.
Three people. Three versions. One number. And strangely enough, the only number generated automatically was the one nobody fully trusted.
This is not a breakdown story
When manufacturers talk about lost production time, the conversation usually goes to breakdowns. The dramatic stops. The alarm that wakes up the maintenance manager at 2am. The machine that goes down for six hours and everybody remembers.
But breakdowns are rarely where most of the production time disappears.
The real losses in most production environments are quieter. They hide in the gap between how long something was planned to take and how long it actually took, between planned time and actual time. In changeovers that run 30 minutes over standard. In starts that never quite reach full speed. In the 7-minute stop that the operator doesn't bother logging because it feels too short to matter… until it happens ten times in a shift. These are the losses that don't trigger alarms. They don't appear in maintenance reports. They accumulate invisibly, shift after shift, week after week, month after month.
Where did the capacity go?
Almost every plant manager has experienced it. Production seems busy. Machines are running. People are working hard. Yet output consistently falls short of expectations. On paper, the schedule should be achievable. In reality, something keeps stealing time. Not enough to trigger concern during the day. Just enough to show up in month‑end reports, delivery delays, overtime costs, or missed production targets.
The result is a frustrating question: where did the capacity go?
The recording problem
The answer is often hidden inside the way production events are recorded. In many factories, a changeover starts when someone decides it starts and ends when someone decides it ends. That sounds obvious. But think about what it means in practice. The operator who logs the stop is often the same person managing the changeover itself, handling materials, adjusting settings, waiting for a quality sign-off. Logging is rarely the highest priority. Getting production running again is. So the timestamp gets entered when there's a moment to enter it, which is rarely the moment the machine actually stopped.
The result is changeover data that is systematically unreliable. Not because operators are careless, but because the recording process depends on human memory and available attention, both of which are in short supply during a changeover.
And unreliable data has a specific consequence: you cannot tell whether your changeovers are getting better or worse. You cannot identify which products, which machines, or which shifts have the longest overruns. You cannot make a targeted improvement because you do not have a trusted baseline to improve against.
You have a number. You just cannot be sure what it means.
The difference between memory and measurement
Consider this question:
if you ran the same changeover on the same machine on a Monday morning and a Friday night, would you expect the same result?
Most production managers, if they are honest, would say no. They have a sense that certain shifts run tighter than others, that certain operators are faster because they simply have more experience, that certain product combinations create more complexity. But a sense is not data.
The manufacturers that consistently improve changeover performance tend to make one critical shift: they stopped relying on estimates and started measuring from the machine itself. From the moment the machine signal changed state, the instant production stopped and the instant it restarted, captured automatically and recorded without anyone needing to do anything.
No memory.
No estimates.
No reconstruction after the fact.
Just facts.
That single shift, from reported time to measured time, changes the nature of the conversation entirely. Instead of debating what happened during last Thursday's changeover, the discussion moves to what the data shows across the last 90 days. Which changeovers consistently overrun. By how much. And whether the pattern points to a process issue, a training issue, or something upstream in planning and scheduling. Now the discussion moves away from opinions and toward patterns.
What accurate changeover data actually enables
Better changeover measurement is not the end goal. It is the starting point for a set of questions that previously had no reliable answers.
Accurate production data makes it possible to answer questions that otherwise remain guesses:
are our changeover standards still realistic, or were they set when the product mix was simpler? Is the overrun concentrated on specific machines, or is it spread evenly? Does it correlate with certain materials, certain shift patterns, or certain time pressures from the schedule? When we made a process change six months ago, did it actually reduce changeover time or did it just feel like it did?
These are not complicated questions. They are the questions every production team is already asking in some form, usually in morning meetings with incomplete information and competing recollections. Accurate, automatically captured production data does not make the questions easier. It makes the answers possible. It creates visibility into the losses that were previously invisible.
The gap between planned and actual changeover time is one of the most consistent sources of hidden capacity loss in manufacturing. It is also one of the most recoverable, but only once you can see it clearly, and see it consistently.
That is exactly what a Manufacturing Execution System is built to do. An MES connected directly to your machines captures the moment production stops and the moment it restarts. Automatically, without operator input, and without depending on anyone's memory of what happened during the changeover. That timestamp is the foundation of everything else: the benchmark, the trend analysis, the shift comparison, the improvement target.
Not an estimate. A measurement.
