How to Do a Time Study in Manufacturing

The short version

A time study in manufacturing means standing at one workstation, breaking the job into its repeating work elements, and timing each one across many cycles — then turning those raw seconds into a standard time you can plan and balance against. The stopwatch is the easy part. The hard part is doing it honestly: separating the work that repeats every cycle from the stuff that happens occasionally, watching a representative operator across enough cycles to trust the number, and taking the lowest repeatable time — the best pace a trained operator holds without rushing — as your standard. This guide walks the whole thing on a real line — how to break down elements, which operator to time, how many cycles to watch, why you use the lowest repeatable time (not an average, a rated guess, or the single fastest read), and the mistakes that produce a number nobody believes. You'll finish able to run a defensible study on one station this week.

What makes a time study trustworthy

A time study lives or dies on a few decisions you make before the stopwatch even matters. You're at the gemba — the production floor where the work actually happens — watch in hand, observing a physical, repeating cycle on a line that has a customer pace it must hold. A few things separate a number the floor trusts from one nobody believes:

A time study is one person, one station, many cycles — and an honest breakdown of the work.

• The work repeats, so the breakdown matters more than the timing. A station runs the same cycle hundreds of times a shift. If you slice it into clean, repeatable work elements with crisp start/stop points, your data is gold. If you time "the whole job" as one blob, you've learned almost nothing you can act on.
• Occasional work hides in the average. Replacing a tote, changing a reel, a quality check every tenth unit — these aren't part of every cycle, but they steal time. Lump them into the cycle and you inflate your standard; ignore them and you understate it. Manufacturing is where separating cyclic from out-of-cycle work is the difference between a real standard and a guess.
• Who you time matters as much as how you time. The operator you watch shapes the result. Time the fastest person and the standard becomes a stick to beat everyone with — and a pace nobody can hold; time the slowest and you bake in the worst case. The fix: watch a steady, experienced B / B+ operator working their normal way, and read the lowest repeatable time — the best pace that recurs without rushing, not a one-off fluke. That's an observed fact, not a judgment call.
• The output feeds everything downstream. Your standard time isn't a trophy — it's the input to line balancing, capacity planning, and comparing real cycle time against takt. Get it wrong and every plan built on it inherits the error.

So the goal of a time study isn't to catch people being slow. Borrowing the Toyota lens: efficiency is the consequence, not the goal. You study the work to make it lighter, smoother, and more standard — and the right output follows.

“

The one-line test for every element you time: does it start and stop at a point I could recognize without watching the clock? If you can't name the exact motion where one element ends and the next begins, your breakdown isn't ready — and neither is your data.

What a time study actually captures, step by step

A time study isn't one number; it's a small stack of them, each built on the last. Here's the sequence that holds up on the floor.

1Break the job into work elements: name where each one starts and stops

Before you time anything, watch the cycle a few times and write down its work elements — the small, distinct chunks of the job, each with a clear breakpoint ("reach for part" ends and "place in fixture" begins). Aim for elements of a few seconds each: small enough to spot patterns and improve, big enough to time reliably by hand. Mark which elements are cyclic (every unit) and which are occasional. This breakdown is the backbone of the whole study and the foundation of any future standard work.

2Time the elements across many cycles — cycle time: the observed reality, not the engineered guess

Pick who to watch first: a steady, experienced B / B+ operator working their normal way — not the fastest (you can't run a line on a sprint) and not the slowest (you won't set the business to the worst case). Then record each element's time over repeated cycles — using either continuous timing (let the watch run and read the running total at each breakpoint, then subtract to get each element) or snapback (reset the watch to zero at the end of each element). Continuous timing is less error-prone and lets you reconstruct everything later, which is why most studies favor it. You're capturing what the work actually takes on the floor — not the time the routing sheet claims.

3Watch enough cycles to trust the number

One cycle is an anecdote. The longer and more variable an element, the more cycles you need before the average means anything. A rough rule of thumb: time on the order of 10 cycles for a short, stable job and more for a longer or erratic one, then check whether your readings cluster tightly or scatter. Toss the obvious outliers — a dropped part, a phone interruption — but write down why you tossed each one. Those discarded readings often point straight at the next kaizen.

4Take the lowest repeatable time — the standard is what's provably achievable

For each element, your standard isn't the average and it isn't the single fastest read — it's the lowest repeatable time: the shortest time the operator hit consistently, with clean motions and no fumble. Lean uses this on purpose instead of the older pace-rating-and-allowances method. Rating ("they were going 110%, so multiply by 1.1…") is a subjective estimate that turns political fast; the lowest repeatable time is observed reality you can point to — it already happened, more than once, without anyone rushing. If an element's readings scatter (say it bounces between 12 and 16 seconds), the low end that repeats is the standard, and the scatter above it is the waste — your next kaizen, not a number to average away.

Breaks, personal time, and fatigue aren't padded into the element, either. They come out of available time when you set takt (available time ÷ demand) — so the work standard stays a clean measure of the task, and rest is handled at the schedule level.

How to run your first time study (step-by-step)

Pick one station or one operation — not a whole line — and work it end to end.

1. Prep and tell the operator. Never study someone in secret; it poisons trust and skews the very pace you're measuring. Explain you're studying the work, not the worker. Stand where you can see the hands without crowding.
2. Break down the elements first. Watch a few cycles and write the element list with crisp breakpoints, flagging cyclic vs. occasional. Lock this before you start timing.
3. Time multiple cycles. Run continuous timing across enough cycles to see the pattern. Capture occasional elements when they occur and note how often (e.g., once every 10 units).
4. Clean the data. For each element, set aside outliers (with a noted reason) and find the lowest time that repeats cleanly. Prorate occasional elements (a 30-second tote change every 10 units adds 3 seconds per cycle).
5. Take the lowest repeatable time. Sum the cyclic elements' lowest repeatable times plus the prorated occasional work — that's your standard time per unit. No pace rating, no allowance padding.
6. Validate against reality. Compare your standard to actual output over a known period and to takt. If they're wildly apart, re-observe — the gap is either bad data or a real problem worth a kaizen.

A focused single-station study like this is the raw material for a process capacity sheet and for balancing the line — and it's the unit you repeat, station by station.

A worked example: timing one assembly station

To make the math concrete, here's how a first study plays out. The numbers are an honest hypothetical — the moves are the ones you'd actually make.

Say you study a sub-assembly station with a steady B+ operator. You break the job into four cyclic elements plus one occasional one (a tote swap every 10 units) and time it across a dozen cycles. For each element you take the lowest repeatable time — the low end that recurred cleanly, not the average and not a single fluke read:

Add the cyclic lowest-repeatable times (7 + 12 + 8 + 5 = 32 seconds) and the prorated tote swap (3 seconds) for a standard of about 35 seconds per unit. No pace-rating fudge, no allowance padding — just the time the work provably takes when it runs clean, because each of those element times actually happened, more than once.

Read that study and the targets jump out, and they aren't "go faster." The fasten element bounces between 12 and 16 seconds — that scatter is the torque-tool variation, and it's a kaizen target because steadying it removes time and improves quality. (Taking the lowest repeatable time, instead of an average, is what makes that scatter visible rather than hiding it.) The 30-second tote swap is waste sitting inside the cycle; pull it out of cycle — feed totes from the side so the operator never stops — and you take 3 seconds off every single unit. The standard didn't tell you to push the person; it showed you exactly where the work is heavy.

Capture your own elements across cycles — the tool records each one, prorates the occasional work, and totals your standard:

A simple time-study health check

You don't re-run a full study every week. Before you trust a standard or hand it downstream, walk the station and answer five questions:

1. Breakpoints: Could a second observer recognize where each element starts and stops without you explaining it?
2. Cyclic vs. occasional: Are the every-tenth-unit tasks separated and prorated, not buried in the cycle average?
3. Sample size: Did you time enough cycles that the readings cluster, and did you note why you discarded any?
4. Operator: Did you time a steady B/B+ operator working normally — not the fastest or the slowest?
5. Lowest repeatable: Is the standard the lowest time that repeated cleanly — not an average, a single fluke-fast read, or a padded guess?

Record the raw element times across cycles and let the free time observation sheet tool total your elements and prorate the occasional work — no spreadsheet wrangling, no login.

Common manufacturing time-study mistakes

• Timing the whole job as one blob. A single 48-second reading tells you nothing you can fix. Break it into elements or you've wasted the study.
• Burying occasional work in the cycle. A tote change or quality check that happens every tenth unit must be prorated, not averaged into every cycle as if it were cyclic. It's the single most common way standards drift wrong.
• Timing the fastest operator. A standard built on a sprint is a pace nobody can hold, and the floor stops believing it. Watch a steady B/B+ operator and take the lowest repeatable time.
• Averaging instead of taking the lowest repeatable time. The average buries the clean cycles under the fumbled ones and hides the variation. The lowest repeatable time shows what the work takes when it goes right — and makes the scatter visible as your next kaizen.
• Studying in secret. Hidden timing destroys trust and changes the pace you're trying to measure. Tell people what you're doing and why — the work, not the worker.
• Letting the standard rot. A study reflects the method on the day you ran it. Change the layout, the tooling, or the part and the standard is stale — re-study after any real change.

Templates & tools

• Free time observation sheet tool — record each element's time across cycles, prorate occasional work, and total your standard time. No login required.
• Standardized Work Combination Table — turn your timed elements into a balanced, repeatable station sequence against takt.
• Standard work — the document your study feeds, capturing the agreed best-known method so the standard survives the next shift.

FAQ

What is a time study in manufacturing? A time study is the practice of observing one operation at the gemba, breaking the job into repeating work elements, and timing each element across many cycles to set a standard time per unit. You take the lowest repeatable time — the shortest time a steady operator hit consistently, without rushing — as the standard. The output feeds line balancing, capacity planning, and standard work.

How do you conduct a time study step by step? Tell the operator and explain you're studying the work; pick a steady B/B+ operator; break the job into elements with clear start/stop points; time enough cycles to see the pattern using continuous timing; clean the data and prorate occasional work; then take the lowest repeatable time per element and sum for the standard. Finally, validate against actual output and takt.

How many cycles should you time? It depends on how long and how variable the element is — short, stable jobs need fewer cycles; long or erratic ones need more. A practical approach is to time on the order of ten cycles, look at whether the readings cluster tightly, and keep observing if they scatter. Always note why you discard any outlier reading.

Which time do you use as the standard — the average, the fastest, or something else? Neither. You use the lowest repeatable time: the shortest time the operator hit consistently across your cycles, with clean motions and no fumble. The average buries good cycles under bad ones; the single fastest read is a fluke you can't reproduce. The lowest repeatable time is provably achievable — it already happened, more than once, without rushing — which is why Lean uses it instead of the older pace-rating-and-allowances method.

How is a time study different from takt time? A time study tells you how long the work actually takes (the cycle time and the standard). Takt time tells you how fast you need to go to meet customer demand. You compare the two: any station whose standard time is over takt caps the line and becomes your balancing and kaizen target.

Where do breaks and fatigue go, if not into the standard? They aren't padded into the element time. Breaks, personal time, and fatigue are accounted for in available time when you calculate takt (available time ÷ demand) — so the work standard stays a clean measure of the task itself, and rest is handled at the schedule level rather than by inflating every element.

Related concepts & guides

• Dictionary: Time Study · Time Observation Sheet · Cycle Time · Standard Work · Line Balancing · Takt Time · Process Capacity Sheet · Out-of-Cycle Work · Muda
• Guides: How to Use Takt Time in Manufacturing for the demand-pace your standard gets compared against, and How to Run a Value Stream Map in Insurance for seeing process time across a whole flow.

Sources

• Lean Enterprise Institute, Lean Lexicon: Standardized Work
• The Productivity Press Development Team, Standard Work for the Shopfloor (Productivity Press)