Takt & Cycle Time Analyser

What this tool does

CM8-286 does two jobs that belong together. It works out your takt time from the production time you genuinely have available and the demand your customer actually takes, then analyses the cycle times you measure with a stopwatch, step by step, against it.

Record one row per timed observation — one step, one cycle, with the conditions it ran under. The tool groups the readings by step, reports the count, minimum, mean, median, maximum, standard deviation and coefficient of variation for each, marks the steps running over takt, names the bottleneck and the least repeatable step, and prints a study. Everything happens inside this single file: no account, no upload, no network request of any kind.

What takt time is

Takt is the rhythm the customer sets — the interval at which one unit must be completed to meet demand exactly. No faster, no slower.

Two inputs decide everything, and both are routinely got wrong. Available time must exclude breaks and planned stops. It is not shift length. Subtract the breaks, the team meeting, the planned clean-down, the planned maintenance and the changeovers you schedule; what is left is genuinely available to produce. Do not subtract breakdowns or waiting — those are losses you are trying to remove, and building them into takt writes them into the plan for ever.

Demand must be the real customer rate — what the customer actually takes, averaged over a sensible period, not the schedule you wish for. If it swings by season, recompute takt for each season.

Takt, cycle time and lead time

Takt time comes from outside the factory: demand divided into available time. Nothing you do at the machine changes it — only the customer's rate and your available hours.

Cycle time comes from inside the factory: how long one step actually takes, timed from a fixed start point to a fixed end point. Every step has its own, the longest sets the pace of the line, and it is the only one of the three you change directly.

Lead time is what the customer experiences: elapsed time from the order or the raw material arriving to the finished unit being handed over. It includes every queue, batch wait, transport and inspection, so it is usually many times the sum of the cycle times — shaving seconds off a cycle barely moves it, removing a queue moves it a lot.

Running a proper time study

A stopwatch study is easy to do and easy to do badly. The rules that matter:

• Fix the start and end points first. Pick something you can see or hear — "hand touches the blank" to "part released into the chute" — and use exactly those two points every cycle.
• Time the same operator over several consecutive cycles. Mixing operators in one set of readings measures the difference between people, not the variation in the method. Time others as their own rows.
• Take enough repeats. Three cycles show a tendency; ten begin to describe a process.
• Record the conditions. A new operator, a material problem, a machine issue or a first-off cycle all move the clock, and a study that does not note which was which cannot explain its own spread later.
• Never use the fastest cycle as the standard. The quickest reading is the one where everything went right at once; setting it as the expectation guarantees the standard is missed most days.
• Time the work, not the person, and say so beforehand.

Why the median and the spread matter

The mean is the figure everyone reaches for and the one a noisy process distorts fastest: a single interrupted reading in a set of five can make a capable step look incapable. The median — the middle reading — ignores how extreme the outliers are. Read the two together: a mean well above the median means a few slow cycles are pulling it, usually interruption or rework rather than a slow method.

The spread matters more than either. A step averaging 50 seconds and running between 48 and 52 is one you can write a standard for. A step averaging 50 seconds and running between 30 and 80 is not one process at all — it is two or three different things under one name, and no amount of balancing will make it behave.

The coefficient of variation

Standard deviation measures spread in seconds, which makes the same 5-second spread look identical on a 20-second step and a 200-second one. The coefficient of variation removes the scale:

The standard deviation used is the sample standard deviation, dividing by n − 1, the right choice when estimating a process from a handful of observations. A step needs at least two non-excluded readings before a CV exists at all.

As a working rule, a CV above roughly 15 % means the process is not repeatable enough to standardise yet. The threshold is a convention rather than a law — set your own in the settings — but the conclusion it points to is honest: fix the variation before you balance the line. Balancing assumes each step takes a knowable time; if a step swings by a third from cycle to cycle, a balance built on its mean is wrong most of the day. Chasing the causes of the spread — waiting for parts, hunting for a tool, an inconsistent fixture — beats chasing the mean.

Abnormal cycles, and excluding them honestly

Some cycles are not the process: a machine faults mid-cycle, a part arrives damaged. Including them makes the step look worse than it is; quietly dropping them from the notebook makes the study a work of advocacy. So exclude them explicitly — tick the abnormal box, record the condition and write what happened. The reading stays in the register and is counted, but is left out of every mean, median, standard deviation and CV, and the study table shows each step's excluded count beside the readings used.

Be strict about what qualifies. The test is whether the cause is part of the method or an interruption to it. A stalled wire feed is abnormal. "A new operator was slower" is not — that is your process, run by the people who actually run it. The tool will not accept a ticked exclusion with no explanation.

The bottleneck

The step with the highest mean cycle time is the bottleneck, and it alone sets the rate of the line. Making a faster step faster produces stock in front of the constraint and nothing at the end of the line.

So compare the bottleneck with takt first. Under takt, the line can meet demand and the work is balance and stability. Over takt, it cannot meet demand however hard anyone works, and there are four honest options: move work off that step to one with room, reduce the work content there, add capacity or a second station, or extend available time. Everything else is exhortation.

The line total sums the step means — the work content in one unit. Divided by takt it gives the minimum number of balanced stations needed to hold the rate: rounded up, and a floor rather than a plan. A step counts as "at takt" when its mean sits within 2 % of takt either way.

What to do with the result

A finished study feeds two pieces of work. Write the method down for each step, with its time and key points, so that what you measured is what gets repeated — the job of the Standard Work Sheet tool. Then redistribute the work content so every station sits just inside takt — the job of the Line Balancing tool. In that order: balancing a process nobody has standardised only moves the variation around.

The spreadsheet workflow

If the readings already live in a spreadsheet, there is no need to retype them.

• Spreadsheet template in the toolbar saves a CSV whose headings are exactly this tool's column names — product or family, process step, station or operator, observation number, observed cycle time, observation date, observer, conditions, abnormal, notes — with a guidance row underneath showing what each expects: the date format, the accepted condition values and the unit for each number.
• Open it in any spreadsheet, fill in one row per timed observation, and delete the guidance row before saving as CSV.
• Import spreadsheet reads it back. Columns are matched by heading, so their order does not matter and extra columns are ignored; numbers, dates and list values are converted for you. Rows missing a required column, or failing a validation check, are skipped and reported by row number.

The file is read by this page in your browser: nothing is uploaded and the original is not changed. Importing adds to what is already here rather than replacing it.

FAQ

How many observations do I need? More than three. Three tells you roughly where a step sits; ten or more begins to describe how it behaves. The noisier the step, the more you need — which is inconvenient, because noisy steps are the ones people give up timing.

Does cycle time include walking between stations? Yes, if the walk is part of the cycle and happens every time. Time what happens, then argue for a shorter walk.

What if one machine makes several products? Record the product on every row and run a separate study per family. Different products have different work content, and averaging them gives a time that is right for neither.

We are under takt but still miss the rate. Why? Almost always availability rather than speed: breakdowns, changeovers and waiting eat time you assumed was available.

Saving your work

Observations, settings and the report header are written to this browser's local storage as you type. That storage belongs to one browser on one computer: another browser, a private window or a clean-up tool that clears site data will not have it.

Treat Export .json as the real save — one file containing everything, which Import .json restores anywhere. Export CSV gives you every filtered observation for spreadsheet work. Reset asks twice, then erases everything this tool has stored. There is no undo.

Accuracy & disclaimer

The arithmetic is elementary and the tool performs it faithfully. Everything that decides whether the answer is true sits underneath: whether the available time is honest, whether the demand is the real customer rate, whether the start and end points were consistent, and whether the exclusions were fair.

Three observations describe a tendency, not a process. Time more cycles before setting a standard, and re-time after any change to method, tooling, material or layout.