WorkSample

About

Why I Built This

I'm an operations manager with a background in automotive and building products manufacturing. I've spent my career on the floor — running lines, training supervisors, and trying to figure out why good lean deployments succeed and why most of them don't.

Building tools is how I learn. I got tired of doing work sampling on a clipboard and losing the data, so I built this. It's free because it should be — the methodology belongs to everyone.

My real interest is in what happens between the training room and the floor. The moment a supervisor stops doing something because someone told them to and starts doing it because they understand why. That gap is where I spend most of my energy.

If this tool helps you see your floor differently — or if you have an idea for something that would make your deployments faster — I'd like to hear about it.

Connect with Eric on LinkedIn →

What tools would make your next deployment faster? That's the question I'm building toward.

Getting Started

How It Works

Work sampling is a statistical technique for understanding how operators actually spend their time on the floor. You observe at random moments, classify what you see, and build a picture from the data.

1Add your production lines in Setup. Pick which TIMWOODS waste types you want to track for this study.

2Go to Study. Name your study — name the investigation, not the date. "Assembly Baseline" is useful in six months. "6/14" is not.

3Hit Start. Walk the line at a natural pace. Glance at one operator. Tap what they are doing at that exact moment. Move on. Every 3–5 seconds per person.

4Hit Finish when you need to stop. Log conditions and what you saw. Your data is saved to this iteration.

5Come back later. Same study name, same line — your new observations add to the existing total. One study can span multiple walks, shifts, or days.

Statistics

Why 97 Observations?

Work sampling uses a statistical formula to estimate time proportions. At 95% confidence with ±10% precision, you need 97 observations. That number doesn't change based on how many operators you have — it's based on the precision you need.

You don't need 97 in one walk. Do 30 before a meeting, 40 after lunch, 27 the next morning. Each is its own iteration. The dashboard aggregates them all. The important thing is that your observations span real variation — different times of day, different conditions.

Below 50 observations is directional only. Don't make permanent changes from one 15-minute walk.

Technique

How to Observe

Don't announce what you're doing. You want to see natural behavior, not performed behavior. Walk at the pace you normally would on a Gemba walk.

Glance at one operator. What are they doing right now — not what they just did, not what they're about to do. Tap it. Move to the next person. One glance, one tap. You should be through 10 operators in under a minute.

The hardest part: resisting the urge to watch a full cycle. You're not doing a time study. One snapshot per operator per pass. That's the method.

Classification

The Categories

Value Add is always tracked. For waste, you choose which TIMWOODS types matter for the question you're asking. Don't track all eight — pick two or three that are relevant to what you're investigating.

✅ Value Add

The operator's hands are on the product in a way the customer pays for. Welding, assembling, cutting, fitting, forming. The test: would the customer pay specifically for this action? If yes — VA. If no or maybe — it isn't.

🚶 Motion

The operator is moving — walking between stations, reaching, repositioning. Not processing, not waiting. Moving. High motion points directly to workstation layout and material placement issues. Root cause is almost always fixable with 5S.

⏳ Waiting

The operator is stopped and idle — feet planted, nothing happening. But don't assume you know why. There are two completely different root causes that look identical from the outside:

Starved — upstream isn't feeding fast enough. The operator is waiting for the next unit to arrive. Fix is upstream: flow, pitch interval, feeder rate.

Blocked — downstream can't accept output. The operator is waiting because there's nowhere for the finished unit to go. Fix is downstream: clear the constraint, reduce WIP.

Work sampling shows you where the waiting is happening. You have to go stand there and watch to know which one it is.

Important: Rising Waiting is often a sign of progress. When you eliminate Motion, operators get more free time — it shows up as Waiting. When you stop Overproduction, operators aren't building ahead anymore — it shows up as Waiting. High Waiting in a system that used to run hot means you've made waste visible. Now you can line balance to takt and convert that Waiting into capacity.

⚠️ Defects

The operator is reworking, inspecting for defects, correcting an error, or scrapping. This is not value-adding — the customer doesn't pay for fixing your own mistakes. If you see high defect time, the root cause is upstream, not at this station.

📈 Overproduction

The operator is building more than demand requires, or more than SWIP standards allow. In a pull system, building ahead of the signal is waste. This is the worst TIMWOODS category because it creates inventory, hides other problems, and consumes capacity you need elsewhere.

🚚 Transportation

Materials, parts, or products are being moved — by the operator or by someone else. Moving things doesn't transform them. Every foot of conveyance is waste. Map the material flow and ask why it needs to move that far.

📦 Inventory

The operator is managing excess WIP, searching through stock, or working around pile-up. Inventory between operations signals a flow problem — either pitch is off, a feeder is running ahead of demand, or pull signals aren't working.

⚙️ Over-processing

More work than the product requires — extra finishes, re-checking already-checked work, unnecessary adjustments. Often invisible because it looks like work. Ask: does the standard require this step? If not, it's waste.

👤 Skills

A skilled operator is doing work below their capability — administrative tasks, cleanup, running to get materials. The waste is the gap between what they can do and what they're actually doing. Also captures time spent waiting for instructions or decisions that should be pre-determined.

Analysis

Reading the Data

VA% is your headline number. World-class assembly operations run 70–85% VA. Most lines I've observed run 35–55%. Every percentage point below your target is recoverable capacity — you don't need more people, you need less waste in the time you have.

Rising Waiting is often a sign of progress. Here's the sequence that works: eliminate Motion first — operators stop walking and get more free time, which shows up as Waiting. Stop Overproduction — operators aren't building ahead anymore, which also shows up as Waiting. Now Waiting is your dominant waste. That's not a failure — that's waste becoming visible. You've consolidated it into one category. Now line balance to takt and convert Waiting into capacity. VA goes up without adding a single person.

Don't act on a single iteration. One walk tells you the direction. Three walks across different conditions start to tell you the truth. Run before and after a change to measure impact.

Conditions matter. Tag every iteration. A short-staffed day with 40% VA and a normal day with 40% VA are two different problems. The data only makes sense in context.

The sequence that works: Eliminate Motion → stop Overproduction → Waiting rises and becomes visible → line balance to takt → VA improves. Never skip to the solution. The data earns the countermeasure.