Industrial engineering software has come a long way. Line balancing tools, standard work libraries and OEE dashboards all look modern. The question that decides whether they help is older: where did the numbers come from? If the answer is a stopwatch and a clipboard, the software computes precisely from a handful of observations.
How stopwatch numbers end up in software
A traditional time study times a limited number of cycles, usually on one shift and often on one product variant. The observer adjusts the raw times with a rating factor and allowances, and someone enters the result once. From then on, the software treats that standard time as fact. Every downstream calculation, from balance to capacity to labor planning, inherits it.
Nothing is wrong with the method. The trouble is age and coverage. The line changes, and the number stays put.
Why the age of the number matters downstream
One stale standard time rarely stays in one place. Line balancing uses it to decide where work should go. Capacity planning uses it to decide how much a line can make. Standard work sheets use it to set the method’s expected time. OEE performance calculations compare actual output against ideal cycle time. When the input sits off by a few seconds at a few stations, every one of those outputs shifts. The software shows no error, because the arithmetic is correct. Only the input is wrong.
Five questions to test your data
Ask these about the standard times behind your tools. Each comes with what a good answer looks like.
When did someone last measure it? A good answer: within the last few months, or after the last process change.
How many cycles stand behind it? A good answer: enough to show the spread, not only an average.
Which shifts and variants did the study cover? A good answer: all the ones you run. If the study covered only first shift and your highest-volume variant, you have a day-shift number, not a line number.
Can anyone see the raw observations behind it? A good answer: yes, with notes or footage that show what happened in each cycle.
How long does a refresh take? A good answer: hours or days. If refreshing means scheduling another study, the number will go stale.
Three signs a standard time has gone stale
You do not need an audit to suspect it. Watch for these signs:
- A station runs over its standard on most shifts, and everyone has stopped treating it as news.
- Supervisors keep their own adjusted number, on a whiteboard or in a notebook, because the official one no longer matches the floor.
- Your balance chart and your walk of the line disagree.
- Each sign means the standard stopped tracking the station. Software built on that standard keeps producing confident answers that nobody trusts.
Where stopwatch studies still earn their place
Studies still do work nothing else can. When a process does not exist yet, you have no cycles to record. Method studies, where an IE watches to redesign how a task gets done, need a person at the station. Low-volume work may not produce enough cycles for analysis. Use studies where they fit, and use recorded data where the line already runs.
What recorded cycle data changes
In our system, cameras capture every cycle at the stations they cover, so the observation count goes from dozens to everything. Outliers come with footage, so an odd number arrives with an explanation. You can refresh standards as the line changes, because the data keeps arriving.
Our AI time studies break each cycle into work, wait, pick, walk, stand and unknown time. The same data can feed line balancing, standard work and OEE views, so your tools agree with each other. We integrate with MES systems and PLCs, so cycle data sits alongside the data you already collect. How it works explains the capture process in more detail.
Recorded data covers the stations the cameras cover, so decide where to start. Begin where the standard times matter most: your bottleneck and the stations next to it. Hand tracking misses losses in other places too. Short stops rarely get written down, because nobody pauses a busy station to log a 20-second interruption. Any number that depends on someone writing it down carries the same blind spot.
What it means for your IE team’s week
Collecting data takes IE time that could go to analysis. When cameras collect it, that time goes to the analysis itself: working through findings, covering more lines, and following each change to see whether it held. The team recovers capacity for improvement work.
A note for the floor
Stopwatch studies can feel like someone is watching the person at the station, and cameras can feel like more of the same. Say plainly what the data is for. In our system, it describes stations and processes, not individuals. It stays in the plant behind your firewall. We collect no biometric data and use no facial recognition. Our security and privacy page covers the details, and your team should hear them before any install. Where a labor agreement covers standard times, changes follow that agreement. Invite team leads to look at the footage with you, so the first time anyone sees the data, it comes with an explanation and not a verdict.
A 30-day check
Pick five stations on one line. Compare each station’s standard time with the recorded cycles from the last 30 days. List the gaps: which standards sit well off the real cycle, and on which shifts. Bring the list to a CI huddle.
Some gaps will show a process that drifted. Others will show a standard that was never right. Both give you something to fix. Repeat the comparison every quarter so the gaps stay small.
Compare your standards with recorded cycles
At Invisible AI we built our vision execution system to give IE teams a current view of every cycle at the stations our cameras cover, with footage behind it. If you want to compare your current standard times with recorded cycles on one of your own lines, we would be glad to set that up with you. Request a demo.