You've done the work. Clipboard in hand, stopwatch running, you've timed 30 cycles of the process. You average the results: 4.2 minutes per unit. Confident, you plug that number into your capacity plan.
And then reality punches you in the face.
The Problem with Averages
Traditional time studies give you a single number — the average. But processes don't run on averages. They run on variability. That 4.2-minute average might mask a range of 2.1 to 8.7 minutes. And that range? It's where your bottlenecks, overtime costs, and missed deadlines are hiding.
Here's what a stopwatch can't tell you:
- How often the process deviates from the average (and by how much)
- What causes the variation — is it the operator, the material, the machine, or the time of day?
- How variability compounds through multiple connected steps
- What happens when two variable processes interact at a shared resource
Why This Matters
Imagine two processes, both averaging 5 minutes per cycle. Process A is rock-steady: every cycle takes between 4.5 and 5.5 minutes. Process B swings wildly: some cycles take 2 minutes, others take 12.
If you staff and schedule based on the average alone, Process A runs smoothly. Process B creates chaos — queues build up after long cycles, then resources sit idle after short ones. Same average, completely different operational reality.
What to Do Instead
Time studies aren't useless — they're just incomplete. Here's how to get the full picture:
1. Record every observation, not just the average. Keep the raw data. You need the distribution, not just the mean.
2. Calculate standard deviation. This tells you how spread out your times are. A high standard deviation means your average is a poor predictor of any single cycle.
3. Look for patterns in the variation. Does the process slow down after lunch? On Mondays? With certain materials? Stratify your data to find the signal in the noise.
4. Use simulation to test the impact of variability. Plug your actual distribution — not just the average — into a model. Watch what happens when variable processes interact with each other over hundreds or thousands of cycles. The results will surprise you.
5. Measure what the stopwatch misses. Setup times, changeover delays, rework loops, waiting for approvals — these often dwarf the "process time" you're measuring.
The Takeaway
A time study gives you a snapshot. Your process is a movie. If you're making decisions based on averages alone, you're optimizing for a world that doesn't exist.
The real insight isn't how long a process takes on average. It's how it behaves when everything — people, machines, materials, and Murphy's Law — interacts over time.
Stop guessing. Start simulating.