Arithmetic on the times and percentages you typed. Averages only, not a forecast, a staffing plan or a promise date. Lead time, yield and rework here follow from your figures and nothing else. Waits are whatever you entered, not something worked out from how busy people are, and an average says nothing about the worst case. Used at your own risk. Open for the full scope limits.
What this tool actually does
It treats the process as a chain of steps where work leaves each step by one of its exits, in the proportions you gave. From that it works out how many times a unit of work passes through each step on average, and multiplies by the times you entered.
- Every figure is an average per unit of work. Lead time is the mean. Half of the work, or far more than half, can take longer.
- A wait is an input. The tool does not model queues. If a step waits two days in your list it waits two days here, whether one request a week arrives or a thousand.
- Percentages are taken as fixed and independent. A request sent back for the third time is given the same chance of passing as one seen for the first time. Real rework usually is not like that.
- Exits that do not add up to 100% are scaled, and exits with no percentage share what is left evenly. Each case is listed under "Check these" with the reading used. If the reading is wrong, every figure downstream is wrong and looks no different.
- Rework is chosen by typing order. The exit whose line closes a loop, or leads back to an earlier phase, is the one counted as rework. That decides first-pass yield and the rework figure. The exits counted are listed and drawn dashed.
Not modelled, at all
- Work done in parallel. Every exit means "one of these happens". Two lanes working on the same item at once, then joining, is not represented, and a step written that way is read as a split.
- Capacity. Nothing here knows how many people are in a lane, what else they are doing, or whether they can keep up. "Work per week" is hours of work arriving, not a utilisation.
- Queues, batching and priorities. Waiting that grows with load, work released in batches, expediting and work that jumps the queue are all outside the model.
- Variation. No spread, no percentiles, no worst case, no confidence interval.
- Calendars. A day is the number of working hours you set and a week the number of days you set, for every lane alike. Shifts, holidays, time zones and a lane that works different hours are not represented.
- Whether the map is the real process. The tool checks that the list is well formed. It cannot know about the step everyone does and nobody wrote down.
- Cost, quality and risk. Only time and counts.
The usual error is in the waits. People remember how long a job takes and badly underestimate how long it sat. A lead time built from guessed waits is a guess with decimals.
What the picture can mislead about
- The tint is relative. The darkest step is the largest on this map, not large in any absolute sense. A map where nothing matters still has a darkest step.
- Position means nothing. Columns follow the order of steps, not time. Distance across the page is not duration.
- Lane colour. The palette has eight colours and repeats after that.
- Exports. A PNG carries the page font for Latin text only, and an SVG uses whatever fonts the program opening it has. Check an export before it goes into a report.
Never use this for
- A delivery date, service level or turnaround promised to a customer, a regulator or a court.
- Deciding how many people to hire, lay off, or schedule.
- Clinical, emergency, safety-critical or life-safety response times.
- A compliance, audit or validation record of how a process is performed.
- A business case, contract or penalty clause without an independent analysis behind it.
Projects are stored in this browser only
Up to ten projects are kept in this browser's local storage, on this device, and are never sent anywhere. Clearing site data, using a private window, or switching browser or device loses them. It is a convenience, not a backup: use Export file to keep a copy of anything that matters.
Before you rely on a result
Measure the waits instead of estimating them. Check the percentages against counts from real work. Walk the map with the people who do the job. Have someone qualified in the process review the result before it supports a decision, a commitment or a report.
Process
One step per line: who does it, how long it takes, where it goes next.
Syntax
Sales: Enter order -> Check credit- Sales does "Enter order", and the work then goes to "Check credit".
Sales: Enter order {10 min, wait 2 h}- Ten minutes of work, after two hours of waiting to be started. Units are
s,min,h,dandwk. Either part can be left out. Finance: Credit OK? -> yes 85%: Pick goods, no 15%: Tell customer- Two exits with a label and a share each. A name ending in
?is drawn as a decision. Customer: (Order placed)- Brackets mark a start or an end. A step named only as a target needs a line like this to say which lane does it.
== Fulfil ==- Starts a phase. Steps below it are grouped, and the phase gets its own share of the lead time.
lanes: Customer, Sales, Finance- The order of the lanes, top to bottom. Without it they appear in the order first typed.
arrivals: 30 / wk- How much work arrives. Adds hours of work per week to each lane.
@ Credit OK?: limit is 30 days overdue- A note kept with that step and shown when you hover over it.
: Finance #4a3aa7- A colour for that lane.
// comment- Ignored, as is a line starting with
#.
An exit that leads back to a step already passed is rework. Which exit of a loop counts as the return is decided by typing order: the line that closes the loop. A name containing a comma, a colon or an arrow goes in double quotes.
Working hours, which is what d and wk mean in your list and how results are turned back into days. One shift is 8, two are 16, round the clock is 24. "Wait 2 d" on an 8 hour day is 16 working hours.
Advanced options
Export
Result
Darker steps take more of the lead time. Dashed connectors are rework. Hover or focus a step for its figures. A wide map is fitted to the panel down to 60% and scrolls sideways past that; the size is under Advanced.
These are averages worked out from the times and percentages you typed. Waits are taken as entered, not derived from how busy anyone is, work done in parallel is not modelled, and nothing here says how long the slowest case takes. Used at your own risk. See the full disclaimer at the top of the page.
Where the time goes
No times entered yet. Add {10 min, wait 2 h} after a step.
By lane
Where work ends up
By phase
Every step
Times are per unit of work, with repeat passes counted.
Every exit
Per unit is how often the exit is taken, on average.