Setup + Stream Build
Enter the stream header, then add process steps and inventory in the order material moves. Drag a card by its handle to reorder, or use Up and Down. Every field feeds the map, the metrics, and the ranked opportunities. Leave a field blank and it drops out of the math instead of being assumed.
Stream Header
Demand + Available Time
Shifts
One row per shift the area operates. Paid minutes less breaks sets the operating hrs each shift carries. Steps then pick the shifts they run and the crew on each.
Steps + Inventory
Current State Map
Process boxes carry the data collected on the walk. Triangles carry inventory between steps. The ladder underneath runs high through wait and low through process time. Lead time is the whole ladder, value add is the low sections only.
Stream Metrics
Every figure below shows its derivation. Verification status comes from the basis set on each node: observed on the floor, pulled from a system, or estimated.
Time + Flow
Step Capacity
Effective cycle time carries setup, downtime, speed, and defects. Capacity is good units per day on the shifts that step runs.
Where The Hours Go, Scheduled To Fully Productive
Each step starts from the hrs it is loaded and loses time in order: setup, then unplanned downtime, then speed, then defects. What is left is fully productive time, and that is what capacity comes from.
Labor By Shift
Crew comes from the shifts each step runs. Labor hrs required splits a step's work across the shifts it runs, weighted by the hrs each shift carries.
Inventory
Quality
Largest Areas Of Opportunity
Every line reads the same way: change one variable from what it is now to what it could be, and here is what the area gets. The target shown is the full gap, the floor or ceiling that variable can reach. Nothing here is assumed: if an input is missing, the line does not appear.
Wait, Change This To That
Capacity, Change This To That
Balance + Quality
Flow Read
Area Capacity Versus A Target
Enter the rate the area has to hit. The file compares it to what the area can run today, names every step that cannot hold that rate, and shows what each variable has to move to for that step to get there on its own.
Steps That Cannot Hold The Target
Every Step Against The Target
Combine Streams + Shared Resource Load
One map answers what happens to one part family. It cannot tell you whether a machine several families share can carry all of them. Load two or more saved maps here, tell the file which steps are the same physical resource, agree on one set of numbers for that resource, and it adds the hrs every family puts on it.
Streams In The Library
Header Reconciliation
These have to agree before the hrs can be added. A stream on a different clock or a different demand window is not comparable, it is a second opinion.
Step To Resource Mapping
Every step from every stream gets a resource name. Two steps with the same resource name are treated as the same asset or the same team. Different families keep their own cycle time on that resource, that part is supposed to differ. Uptime, speed, first pass yield, and scheduled hrs belong to the resource, so those get one agreed value.
Agreed Numbers Per Resource
Blank uses the default shown. The default for scheduled hrs is the largest any stream claims, and for the loss numbers it is the hrs weighted average across the streams that use it. Where the streams disagree the row is flagged, and the flag stays until someone enters the agreed value.
Combined Load On Every Resource
What Each Stream Consumes
Answering The Objection
The objection is right about one thing and wrong about the other. It is right that a single map cannot size a shared machine, because it only carries one family's demand. It is wrong that this means the area cannot be modeled. The order of work is this:
- Map each family on its own. A family is a set of parts that run the same steps on the same equipment in roughly the same order. Two or three families usually cover most of the volume, and the rest gets carried as an other bucket with its own hrs.
- Give every step a resource name that matches the physical asset or team, not the local nickname. Two maps saying Mill 3 and Big Bridgeport are the same machine and have to end up on one line.
- Load the maps here and reconcile the header first: same shifts, same clock, same demand window, same unit.
- Agree the loss numbers per resource. The part carries the cycle time, the resource carries the uptime, the speed, the first pass yield, and the hrs it is scheduled.
- Read the combined load. The resource over 100 percent is what stops the mix, not the step someone complains about.
- Scale the mix. Change any stream's demand in the library and every resource reloads, which is what a schedule argument actually needs.
What this does not do is sequence jobs. It sizes whether the hrs exist across the whole mix. Sequencing is the scheduling system's job, and it cannot fix a resource that is short of hrs.
Reconciling Different Assumptions And Names
- Name the resource by what it is on the floor: asset number, work center, or team. Nicknames go in the notes, not in the resource name.
- One clock. Paid minutes, breaks, meetings, cleanup, and maintenance have to be the same on a shift that two maps both use. If one map says 7.08 hrs and another says 7.50 for the same shift, someone counted breaks and someone did not.
- One demand window. All maps use the same weeks of demand, or the load is a comparison of different months.
- Cycle time is allowed to differ by family. That is the whole point of a mix. Uptime, speed, first pass yield, and scheduled hrs are not allowed to differ on the same resource.
- Where they do differ, the file shows every value and who claimed it. Pick one and enter it as the agreed number, or go back to the floor and settle it.
- Watch the basis. A resource where one map says observed and another says estimate carries the weaker of the two into every number built on it.
- No double counting. If two maps both include a shared inspection step, that step appears once per family with its own demand, which is correct. If the same family appears in two maps, remove one.
- Changeover belongs to the resource across the mix. More families on one machine means more changeovers, and the file adds those hrs per family as entered.
Material Flow Actions + Modeled Effect
Every action names the one variable it moves, shows what that variable reads today, takes the value you want it set to, and returns what the area gets. Change changeover at Mill from 45 min to 20 min and the file says how many more units per day that is. Blank target means nothing is modeled.
Actions come from the Material Flow and Work In Process Playbook, items 1 to 31, in the playbook's own priority order. Items that the map can size appear against the node their pool belongs to. Items that hold gains or feed data appear in the reference table with no number attached, and item 24 sits out because it carries equipment moves. Add playbook item covers anything the area runs that is not on the list.
Modeled Effect On The Area
Change This, Get That, Ranked
Actions By Node
Nodes are ordered by the size of the gap their own data shows. The top three open by default, the rest expand on click, and everything opens when the page prints.
Playbook Coverage
Future State Model
Set targets, see the recomputed stream. Targets left blank hold at current. This models the arithmetic only: the plan to hit the target is separate work.
Current Versus Future
Buffer Detail
Future State Questions
- What is takt time for this family, and does the stream run to it
- Does the stream build to ship direct, or to a finished goods supermarket
- Where can continuous flow run, one piece or small transfer batch, with no buffer between steps
- Where does flow have to stop, and a supermarket pull set the release instead
- Which single step is the pacemaker, the one point scheduled
- How is mix leveled at the pacemaker
- What increment of work is released, and how often the release is checked
- What process improvements have to land for the design to hold: changeover, uptime, yield, staffing, layout
How To Build A Value Stream Map
A value stream map is one page that shows material flow, information flow, and the time a part spends in the stream, for one part family, door to door. It is drawn on the floor with a pencil and a stopwatch, then entered here.
Step 1: Pick The Part Family
- One family per map. A family is a set of parts that run the same steps on the same equipment in roughly the same order.
- Build a matrix: parts down the side, process steps across the top, mark where each part goes. The clusters are the families.
- Name the family and the volume it carries. If you cannot state demand per day for the family, the map has no takt and no inventory math.
Step 2: Set The Scope
- Door to door first: receiving dock to ship dock inside one site. Extend to supplier and customer once the internal stream is understood.
- Multi site streams map per site, then link on the pegged network need date. One order does not travel between sites.
- Write the start point and end point down before you walk. Scope creep on the walk kills the map.
Step 3: Walk The Flow Backward
- Start at shipping and walk upstream. Backward keeps you anchored to what the customer actually receives.
- Walk the whole flow yourself first, fast, no data. Then walk again and collect.
- Collect your own times. Do not pull routing standards and call them observed. A standard is what the work should take, not what the part experienced.
- Count the inventory yourself at every point material sits: floor stock, racks, carts, staging, work in process at the machine, quarantine, rework cages.
Step 4: Draw The Boxes And Triangles
- One process box per point where flow stops and the part changes hands or changes state. Boxes are objects, not owners.
- One triangle wherever material sits between boxes, with the count you took.
- Data box under each process box: cycle time, changeover time, uptime, first pass yield, operators, batch size.
Step 5: Add The Information Flow
- How does each step know what to run next: a system release, a printed list, a verbal push, a kanban signal.
- Mark every point that receives a separate schedule. A stream scheduled at every step is a push stream by definition.
- Note the release frequency and the horizon: daily, weekly, firmed window.
Step 6: Draw The Timeline
- Ladder under the map: high through wait, low through process time.
- Wait at a triangle in days = count on hand / demand per day. Convert to hrs on the same clock you use everywhere else on the map.
- Lead time = all wait + all process time on the longest path. Value add = process time that changes the part toward what the customer bought.
Step 7: Read It Before You Fix It
- Compute takt, process cycle efficiency, the constraint, and the top three inventory holds. Enter the data in the Build tab and the Metrics and Opportunities tabs do this.
- Do not design the future state on the walk. Finish the current state first, with the numbers standing up.
Step 8: Design The Future State And The Plan
- Work the eight future state questions in the Future State tab, right to left, customer end first.
- Every gap becomes a kaizen item with an owner, a date, and a number it is expected to move.
- Remap on a set interval. A map that is never redrawn becomes wall art in about six weeks.
Why The Map Works
- It puts wait and work on the same page. Most streams are 1 to 5 percent value add, so the answer is almost never to work faster.
- It sizes problems in hrs instead of opinions. The biggest triangle wins the argument.
- It shows the whole part path, so local improvements that move a problem downstream get caught before they are funded.
- It gives one shared picture across production, planning, quality, and engineering, built from the same walk.
- It exposes the schedule mechanism. Push, pull, or verbal shows up in the information flow, not the material flow.
- It sets a baseline the future state gets measured against, in lead time, process cycle efficiency, and inventory days.
- It is fast. A door to door map on one family is a day or two of work, not a quarter.
Data To Collect
Collect what the map computes with. Each field below lists where it comes from and what it drives. Set the basis on every node so the verification status travels with the number.
Stream Level
| Field | Definition | Source | Drives |
|---|---|---|---|
| Demand per day | Good units the customer pulls per work day for this family | Firm orders, shipment history, forecast for the window mapped | Takt • inventory days of supply • capacity gap |
| Shifts | One row per shift the area operates, paid minutes less breaks | Shift schedule | Plant operating time • takt • wait clock |
| Shifts run at a step | Which shifts that step is manned and running | Staffing plan, observation | Operating hrs at the step • capacity • load |
| Crew per shift | Operators on that step during that shift | Observation, staffing plan | Labor hrs available and required by shift |
| Work days per month | Days the stream runs | Plant calendar | Monthly demand to daily demand |
| Wait clock basis | Whether wait is counted on work hrs or on a 24 hr calendar | Decision, stated once and held | Lead time • process cycle efficiency |
| Target buffer days | Days of supply a buffer is allowed to hold | Policy or the future state design | Opportunity sizing |
Process Step
| Field | Definition | Source | Drives |
|---|---|---|---|
| Cycle time | Time between good units coming off the step, per unit, at current staffing | Stopwatch on the floor, several units, not one | Constraint • capacity • process time on the ladder |
| Value add time | Portion of cycle time that changes the part toward what the customer bought | Observation, split from cycle time | Process cycle efficiency |
| Changeover time | Last good unit of the prior part to first good unit of the next | Stopwatch across a real changeover | Available time lost • batch size pressure |
| Changeovers per day | How many times the step changes over in a day | Observation, run history | Hrs per day lost to setup |
| Uptime | Percent of scheduled time the step is able to run | Downtime logs, transaction history, observation | Effective cycle time • capacity • loss hrs |
| First pass yield | Percent of units through the step correct the first time, no rework | Inspection records, scrap and rework counts | Rolled throughput yield • effective cycle time • rework hrs |
| Operators | Default crew on the step, used for any shift with no crew entered | Observation | Labor content • staffing sizing |
| Batch size, or every part every interval | Units run before the step changes to another part, or the interval between runs of the same part | Run history | Minimum inventory the batch forces |
| Basis | Observed, system, or estimate | Set by the mapper | Verification status carried into every figure |
Inventory Point
| Field | Definition | Source | Drives |
|---|---|---|---|
| Count on hand | Units sitting at that point on the day of the walk | Physical count on the walk, system stock as a cross check | Days of supply • wait hrs • lead time |
| Days of supply | Count on hand / demand per day | Computed | Ranking of the largest holds |
| Wait hrs, direct | Measured dwell where a count is not meaningful, cure, cool, test soak, transit | Observation or process spec | Lead time on non queue waits |
| Reason held | Why material sits: batch, changeover, quality hold, staffing, transport, awaiting paperwork | Ask at the point | Which countermeasure fits |
What Varies The Productive Hours In A Day
Every factor below moves the hrs a step can actually produce in. The model runs them in order: calendar, then schedule, then planned stops, then setup, then downtime, then speed, then defects. Enter each one where it belongs and the loss shows up once, not twice.
| Layer | What eats the hrs | Where it goes in this file | Source |
|---|---|---|---|
| Calendar | Total clock, 24 hrs a day, 168 hrs a week. Nothing is lost yet | Fixed | Calendar |
| Schedule | Shifts not staffed • days not run in the week • weekends • holidays and shutdowns • no orders released | Shift rows: paid minutes, days per week, and how many shift rows exist | Shift schedule, plant calendar |
| Planned stops | Breaks and meal • shift start huddle • pass-down at handoff • cleanup and 5S • planned maintenance window • training • safety stand-down • inventory counts | Shift rows: break, meetings, cleanup, maintenance minutes, and the break covered flag when relief keeps the asset running | Shift routine, observation, maintenance plan |
| Setup | Changeover, tool and fixture changes, first article and warm-up before the run | Step fields: changeover time and changeovers per day. Setup comes off loading time before uptime is applied | Stopwatch across a real changeover |
| Unplanned downtime | Breakdowns • jams • material starvation • blocked downstream • waiting on inspection, paperwork, crane, or forklift • waiting on an operator | Uptime percent, per step and per shift | Downtime logs, transactions, observation |
| Speed | Running under rate • minor stops under the logging threshold • learning curve on new operators • mixed model rate differences | Speed percent, per step and per shift | Counts against cycle time over a run |
| Defects | Startup rejects • in process defects • rework loops | First pass yield percent, per step and per shift | Inspection records, scrap and rework counts |
| Labor | Absence and call outs • vacation • training hours • indirect work pulled off the job • break relief coverage • crew size and qualification • overtime fatigue • new hire ramp | Attendance percent on the shift row, crew per shift on the step, and the Labor By Shift table | Time and attendance, staffing plan, skills matrix |
Equipment effectiveness is availability times speed times quality, measured inside the hrs the step is scheduled to run. Total effective equipment performance is that same number times the share of the calendar the area is scheduled at all, which is what shows whether the answer is a better asset or more scheduled hrs. Both are on the Metrics tab.
Terms In Plain Words
| Term | What it means |
|---|---|
| Value stream map | One page showing the path one part family takes, the time it spends working, and the time it spends waiting |
| Cycle time | The time between good units coming off a step |
| Changeover | Last good unit of one part to first good unit of the next |
| Takt | The customer pace. One unit has to finish this often to keep up with demand |
| Lead time | Total time a part is in the area, waiting plus working |
| Value add time | The part of the time that changes the part toward what the customer bought |
| Process cycle efficiency | Value add time as a share of lead time. Most areas land between 1 and 5 percent |
| Days of supply | Units sitting at a point divided by daily demand. How many days that pile would cover |
| Work in process | Everything started and not yet finished |
| Uptime | Share of scheduled time the step is able to run |
| Speed, percent of rate | How fast it runs against the rate it should run, counting small stops and slow running |
| First pass yield | Share of units correct the first time, no rework |
| Rolled throughput yield | First pass yield of every step multiplied together, the odds a unit clears the whole stream clean |
| Equipment effectiveness | Uptime times speed times quality, measured inside the hrs the step is scheduled to run |
| Total effective equipment performance | Equipment effectiveness times the share of the calendar the area is scheduled at all |
| Scheduled running hours | Paid hrs less breaks, meetings, cleanup, and planned maintenance. The hrs the asset is available to run |
| Fully productive hours | What is left after changeover, downtime, slow running, and defects. Capacity comes from this |
| Constraint | The step with the highest load, the one that sets what the area can ship |
| Load | Demand divided by capacity at a step. Over 100 percent means it cannot keep up |
| Full time equivalent | Hours expressed as a share of one person working a full shift |
| Pull signal | A card, bin, or space that tells the step before it to make more, instead of a schedule pushing work in |
| Pacemaker | The one step that gets a schedule. Everything before it refills, everything after it runs oldest first |
| Little's law | Time in the area equals work in process divided by units finished per day |
Reading The Map
Six reads, in order. Each one has a number behind it and a countermeasure family that fits. The Opportunities tab runs these against the entered data.
Read 1: Wait Versus Work
- Process cycle efficiency is value add time divided by production lead time. Manufacturing streams commonly land between 1 and 5 percent.
- Low process cycle efficiency with a healthy constraint says the problem is waiting, not speed. Time studies and faster operators will not move it.
- Countermeasures: smaller transfer batches • release control • buffer caps • combining steps into flow.
Read 2: The Constraint
- Effective cycle time is scheduled running hrs divided by capacity, so it carries setup, downtime, speed, and defects. The highest load, demand over capacity, is the constraint, not the step people complain about.
- Compare demand times effective cycle time against the hrs that step actually runs. A step on one shift carries one shift of hrs even when the plant runs three. Above 100% load, the stream cannot meet demand at the shifts and staffing entered. Below takt with long lead time, the problem is release and queueing, not capacity.
- Countermeasures at the constraint: uptime recovery • setup reduction • yield fix • staffing • offload work content to non constraint steps.
Read 3: The Largest Triangles
- Rank the inventory points by hrs held, not by unit count. The count is meaningless without demand behind it.
- Ask at each one why it is there. Batch size, changeover, quality hold, staffing mismatch, and transport all look identical on the map and need different fixes.
- A buffer in front of the constraint is intentional protection. A buffer everywhere else is usually schedule and batch residue.
Read 4: Balance
- Plot cycle time per step against takt. Steps far under takt hold idle capacity, steps at or over takt set the pace.
- Wide spread between steps means the stream cannot flow without inventory between them.
- Countermeasures: rebalance work content • move operators to the pace setting step • split or combine steps.
Read 5: Quality
- Rolled throughput yield is the first pass yield of every step multiplied together. It falls fast across a long stream.
- Scrapped and reworked units consume constraint hrs twice and inflate the demand every upstream step sees.
- Countermeasures: fix at the step that creates the defect, not at the inspection that catches it.
Read 6: Variability And Load
- Queue grows with utilization and with variability, both arrival and process. At high utilization a small variability change moves wait time a lot.
- Signals on the map: uneven inventory at the same point day to day • expedites • overtime with idle time in the same week.
- Countermeasures: level the release • cap work in process • reduce setup so smaller lots are affordable • protect the constraint with a sized buffer instead of a random one.
Little's Law Check
Lead time equals work in process divided by throughput. Count work in process in the stream, divide by units shipped per day, and compare to the lead time the ladder produced. If the two disagree by more than a small margin, either the count missed material or the demand rate is wrong. Fix the disagreement before the map goes anywhere.
Common Errors
- Routing standards entered as observed times. The map then describes a plant that does not exist.
- Inventory taken from a system without a physical count. Floor stock, staging, and rework cages get missed.
- Averaging several parts into one map, then chasing a number no single part has.
- Mixing wait clocks. Some triangles on calendar hrs, others on work hrs, process cycle efficiency meaningless.
- Mapping every part number in the shop. One family per map.
- Jumping to the future state during the walk, before the current state numbers stand up.
- Naming the map's constraint from memory instead of from effective cycle time.
- Treating the map as the deliverable. The deliverable is the plan, the owners, and the remap.
Class Guide
How to run this file in a session. Everything is offline. No install, no account, no network.
Using The File
- Open in any browser. Enter data in the Build tab, everything else recomputes.
- Reorder by dragging: cards on the Build tab by their handle, boxes and triangles on the Map tab. Up and Down buttons do the same on a touch screen.
- Combine Streams tab: load two or more saved maps, map each step to the resource it runs on, settle the numbers the streams disagree on, and read the load every family puts on the shared machines. This is the answer to "our schedule is too complicated for one map."
- Capacity To Target tab: enter the rate the area has to hit and the file names every step that cannot hold it and what each variable would have to move to on its own.
- Playbook Actions tab: items 1 to 31 of the Material Flow and Work In Process Playbook. Each action names one variable, shows what it reads now, takes the value you set it to, and returns what the area gets.
- Save file writes a JSON of the stream. Open file reads it back. Send the JSON, not screenshots.
- Download actions list writes a CSV of every action, its pool, the capture entered, and what it drives. Download summary writes a plain text report of the stream, the metrics with their derivations, and the actions, ready to paste into an email.
- Test data loads a worked stream with captures already set. Two sets: make shop, asset bound, and assembly shop, labor bound. Clear wipes the stream, the captures, and the added items.
- Print produces a handout of every tab, map included.
- Clear the test data before the class enters real numbers.
Session Flow, Half Day
| Time | Block | Output |
|---|---|---|
| 20 min | Why the map, wait versus work | Class states process cycle efficiency in their own words |
| 20 min | Family selection and scope | One family named, start and end point written |
| 30 min | Data box fields and how to time them | Everyone can define cycle time, changeover, uptime, first pass yield |
| 60 min | Floor walk, backward, counts and times | Pencil map with every triangle counted |
| 30 min | Enter into this file | Current state with basis set on every node |
| 30 min | Read the map, six reads | Top three opportunities in hrs |
| 40 min | Material flow actions, capture by node | Action list with the hrs each one drives |
| 40 min | Future state questions and targets | Modeled lead time and process cycle efficiency, gap list |
| 20 min | Plan out | Items with owner, date, number to move |
Exercise Rules
- Count it, do not remember it. Every number on the map came from the walk or gets marked estimate.
- One mapper holds the pencil, the group holds the argument.
- Ask at every triangle: what is this waiting on. Write the answer, it is the countermeasure.
- No solutions until the current state is complete.
- State assumptions out loud, one line each, then set the basis field to match.
Facilitator Checks
- Does demand per day match what actually ships
- Does the Little's law check agree with the ladder
- Is the constraint the highest load, or the loudest step
- Is any figure carrying an estimate basis into a decision
- Does the future state hold if T1 is not achieved