Value stream mapping: the walk, the argument and the map nobody reads

A product family narrow enough to draw truthfully

Group products by the process steps and equipment they share, not by commercial category or customer. A family that covers everything the plant makes forces every box on the map to become an average, and averages conceal exactly the variation you were looking for. Choosing well usually means a family that accounts for a meaningful share of volume and passes through the areas you suspect are the problem. Where the mix is genuinely diverse, map a representative order rather than a family, accept that the picture is one case, and say so on the drawing so nobody later treats it as universal.

Data collected by walking the route, not by querying the system

The planning system holds the routing as designed, which is frequently not the route material takes. Walk it backwards from despatch with a clipboard: count what is physically sitting between each pair of operations, note where it is stored, ask the operator what they actually do rather than reading the method sheet, and time a few cycles yourself. Record who tells each operation what to make and when. The differences between what people believe and what the walk shows are the most valuable output of the exercise, and they only appear if the team goes and looks rather than delegating collection to an analyst.

The argument that the current state is supposed to provoke

Drawing the current state in front of planners, supervisors and engineers reliably produces disagreement, because each function has been managing a piece of the stream on assumptions the others do not share. Let the argument happen; it is more useful than the drawing. The failure mode is a facilitator who smooths over conflicts to finish the map, or a team that draws the process as it should work rather than as it does. Insist that anything on the map can be pointed at on the floor, and mark contested items as unknown rather than guessing, because a single fabricated figure discredits everything else.

A future state is a set of commitments, not a drawing

The second map is worth nothing until it is decomposed into changes with named owners and agreed review points, and until those changes sit in the same plan as the plant's other work. Sequence matters: stabilising a process usually has to precede connecting it to the next one, and pull cannot be laid over an operation that stops unpredictably. Be explicit about what each change requires, whether that is engineering time, tooling money, a planning rule change or a conversation with a customer. Anything that requires capital should go into the capital process immediately rather than waiting for a review meeting to notice it.

What a map cannot tell you

It is a scoping instrument. It shows where time and stock accumulate and how work is triggered, which is enough to decide where to look. It says nothing about why a machine loses availability, how variable a process is, what capacity a shared resource really has under a mixed load, or what a defect costs. Teams that treat the map as analysis end up proposing flow through an operation that cannot hold its process. Use it to choose the target, then use the appropriate technique on that target, and revisit the map only when the stream itself has changed.

Frequently asked questions

Who needs to be in the room when a value stream is mapped?

Someone from each function that touches the stream, including planning, the production areas involved, quality, maintenance and despatch, plus a manager with the authority to change a planning rule or release engineering time. Operators from the mapped areas should walk it with the team. If the person who can authorise change only sees the finished map, the exercise turns into a presentation and the commitments never get made, which is the most common reason mapping produces nothing.

Should we map the office and order-handling steps as well?

Often yes, because a surprising share of elapsed time sits before anything is cut or moulded: quotation, order entry, drawing approval, credit checks, material release. Those steps are usually invisible to the factory and unmeasured by anyone. Map them at a coarser level than the physical process, focusing on handovers and waiting rather than on task detail, and be prepared for the finding that the fastest available improvement is administrative rather than mechanical.

How often should a map be redrawn?

When the stream materially changes, not on a calendar cycle. A new product family, a relocated operation, a changed planning approach or a shift in volumes all justify redrawing. Refreshing a map that nothing has changed underneath consumes the same people for no new insight and encourages the habit of treating mapping as the improvement activity itself. Between redraws, track the small number of measures the map identified rather than reproducing the whole picture.