Value stream mapping (VSM)

What is value stream mapping?

Value stream mapping (VSM) is drawing the whole flow of one product or service on a single sheet, from the moment a customer asks for it until they have it, with the times and the quality figures written under every step. The Lean Enterprise Institute defines it as diagraming every step involved in the material and information flows needed to bring a product from order to delivery.

Once the waiting sits on paper right next to the working, everyone in the room can see the process is slow for reasons nobody was measuring. That is the point of the afternoon.

Toyota developed the method and calls it material and information flow mapping. Mike Rother and John Shook wrote it down for a wider audience in the workbook Learning to See, published by the Lean Enterprise Institute in May 1998, and their introduction says the method they present is based on the material and information flow maps Toyota uses.

You map one product family at a time, never the whole company. A family is a group of items that go through roughly the same steps: small repair jobs, custom orders, standard orders from stock. Mix two families on one sheet and every number becomes an average of two processes, which describes neither.

What is on a value stream map

A map has four layers, and each one answers a different question.

The process boxes run left to right across the middle, one per step. Under each box sits a data box with what you measured while standing there: cycle time, changeover time, uptime, batch size, number of people. The Lean Enterprise Institute defines cycle time as the time required to produce a part or complete a process, as timed by actual measurement, and changeover time as the period between the last good piece of one run and the first good piece of the next. Both are stopwatch numbers, not estimates pulled from a system.

The triangles between the boxes are the inventory or the queue: parts on a shelf, or files in an inbox with the age of the oldest one. This layer is where most of the time hides, so a map that skips it tells you nothing you did not already know.

The information flow runs across the top, from the customer on the right, through whoever schedules the work, down to each box. In Shook's words, information flow tells each process what to make or do next and when to do it. Weekly plans, a spreadsheet on a shared drive, a foreman walking over and saying start on this one: they all go on the map. This top half is what separates a value stream map from a flow chart, because nothing else you draw shows that a step is slow because of how it gets told what to do.

The timeline at the bottom alternates between the working time inside each box and the waiting time in each triangle, and ends in two totals: the sum of the process times and the sum of the lead times. Rother and Shook drew it as a square wave with waiting on the peaks. Karen Martin and Mike Osterling, who adapted the method for office work, put process time above a straight line and lead time below it, because teams kept mixing up which number belonged where.

Touch time against lead time

Those two totals are the headline the map produces. Divide the working time by the elapsed time and you have what Martin and Osterling call the activity ratio: the sum of the timeline process times over the sum of the timeline lead times, times one hundred.

The Lean Enterprise Institute defines value-creating time as the time of those work elements that actually transform the product in a way the customer is willing to pay for. Everything else on the timeline is enabling work or plain waiting, and in office streams the value-creating share routinely lands in the low single digits. A team that has been told to work faster gets to point at a strip of paper where their own work is a thin sliver and the queues around it are the rest. The entry on cycle time, lead time and waiting time sorts out which word means which clock.

Current state, future state, action plan

The current state map is what actually happens, drawn from a walk: you start at the customer end, work backwards up the stream, and ask the person doing each step how long it takes and what arrives wrong. Draw it from a system export instead and you get the process as designed, which is not the one you are trying to fix.

The future state map is what the flow should look like at a horizon you pick, usually six to twelve months out. The Lean Enterprise Institute calls it a target image of how material and information should flow through the value stream, and it goes on the same timeline so the new totals sit next to the old ones. Then comes the part most maps never get: every difference between the two sheets becomes a line with an owner, a date, and the number on the timeline it is meant to move. A map without that list is a poster.

Mapping an office process instead of a factory floor

Shook's own warning about mapping away from the plant floor is that the split between material flow and information flow blurs, because in an office the information travels with the work itself. Both halves start describing the same thing, and the map slides toward being a flow chart with times on it. The way out is to keep asking who tells this step to start and where that instruction comes from, even when the answer is an email.

Inventory triangles become inboxes. Batch size becomes a habit: the owner who checks all the quotes on Friday evening is running a weekly batch, and it costs the same days a weekly production run costs. Uptime becomes the availability of the person who has to act, which is how one week of holiday turns into the entire reason a step takes eight days.

The office adaptation also adds a quality figure per box, percent complete and accurate, written %C&A. You get it by asking the next step down what share of the work arrives usable as is, with nothing to correct and nothing to phone back about. A box at 70 percent means three in ten items loop backwards, and that loop is why the same step turns up twice on the timeline.

Worked example: an installer from quote to invoice

A heating installer with eleven people maps one family: small service and repair jobs under about three thousand euro, roughly twelve a month, from the customer's call to the invoice. Five process boxes, four queues, measured over one month of real jobs.

• Intake and planning, office, 15 minutes. Then 6 working days waiting for a survey slot in a technician's calendar.

• Site survey, technician at the customer, 60 minutes. %C&A is 70 percent: three in ten sets of notes come back missing a measurement or a photo. Then 2 working days before the notes reach the office.

• Drafting the quote, office, 75 minutes. Then 7 working days waiting for the owner to check the prices, because he does quotes on Friday evenings.

• Sending the quote and following it up, office, 30 minutes. Then 3 working days until the customer says yes and the job gets slotted in and done.

• Invoicing, office, 60 minutes.

The totals: 4 hours of touch time against 18 working days of queue, so about 18 and a half working days end to end, an activity ratio of 4 hours over roughly 148 hours, a bit under 3 percent. The technician's hours on the job itself are the value-creating work the customer pays for and sit outside this count; the map measures everything wrapped around them.

Two queues carry 13 of the 18 days. The 7-day Friday sign-off is a batch, and a price list plus a threshold (standard materials under three thousand euro go out without a check) empties it for most jobs. The 6-day survey slot is a scheduling rule, and for the eight job types this installer has done a hundred times, quoting from photos and the installation record already in the system removes the wait and the 60 minutes with it. Future state on the same timeline: about 5 working days.

An AI agent that drafts the quote from the survey notes takes those 75 minutes down to a few, which on this map is 75 minutes off 18 days. Draw the timeline before you buy anything, because it tells you in advance whether a tool that shortens touch time will move the number your customer actually feels. Here it does not, until the Friday batch is gone.

How it differs from a swimlane and from process mining

A swimlane diagram splits the process into horizontal lanes per role or department and shows which lane each step belongs to. It is the right picture when the argument is about responsibility or where a step should live, and it carries no clock. A value stream map has one lane and a clock. It cannot show you cleanly that three departments touch a file, but it can show that the file spent nine of its eleven days in a queue.

Against process mining the dividing line is where the data comes from. Process mining reconstructs the flow from timestamps already sitting in your ERP, CRM or ticketing system, so it covers every case, it is repeatable next month, and it can rank a hundred variants by how often each one happens. Value stream mapping is drawn by hand from a walk, so it covers one typical run of one family and is only as good as the day you walked it.

In an SME the walk is often the only option, because the systems do not log the waiting. A quote request lives in a mailbox until someone opens it, and the ERP records the moment the quote was created, not the six days it sat there. Process mining reads what the software knows; the map records what the person tells you when you ask why this pile is here.