One path, read from many sheets.
Northline Components is fictional. Its Series 40 valve bodies follow one path from Helen Ward’s promise to Luis Ortega’s dock. Each sheet is a different view of that path.
Record the promise
- SIPOC
The promise starts when Helen Ward records it for Ruth Adeyemi, and it ends when Luis Ortega confirms the shipment to Hartwell.
- Voice of the customer
Ruth Adeyemi’s crew is booked to the confirmed morning. On-time to that promise day is a promise they can feel. The target is 97 percent.
- Process map
Helen Ward records the promise in Ledgerline. When it passes to Mara Ellison, the revision often stays in email.
- Run chart
On-time ship across the twelve weeks is 82 percent. The weekly median is 82.5, and the target is 97. This sheet has no control limits.
- Daily management
The Friday 2 October shift expected 18 bodies against promises already recorded, and no expedite before noon.
Release the lot
- Process map
Mara Ellison releases the lot in Ledgerline, and from memory where ShopPacket lags. She hands Andre Pell a printed list.
- Standardized work
Mara Ellison does not release a mismatch. The posted rule is: do not run if the traveler rev differs from Rev C.
- Box plot
When the rev is confirmed before release, the queue median is 3 hours. When the traveler rev is blank, the median is 13 hours. This is not a tested cause.
- Scatter plot
More unchecked travelers in the morning release go with more lots missing the promise that day. The dock-lag day had 0 unchecked and 1 late lot.
- A3
From 21 September, Mara Ellison does not release a mismatch. Expedites were 11 in a typical week before the rule, then 7, then 4.
Issue the material
- Process map
Andre Pell issues material from the stores cage against a printed list. The traveler he hands Devon Cho may still show Rev B.
Machine the seat face
- Process map
Devon Cho machines the seat face on cell 2 from the ShopPacket traveler. He hands Priya Shah the count in grease pencil.
- Value stream map
Posted touch time is 46 minutes under a median queue of 11 hours. The 46 minutes are already over the 25-minute takt.
- Spaghetti diagram
When the traveler rev is blank, Devon Cho walks from cell 2 to the ShopPacket printer, the drawing cabinet, and Mara Ellison’s desk, then back. That walk is 12 minutes.
- Operator balance
The posted cycle is 46 minutes against a takt of 25. The finish cut is 11 minutes, so another person cannot split it. The observed cycle is 58 when the rev is blank.
- Standardized work
The posted elements are load 5, rough cut 16, finish cut 11, deburr 6, in-cell check 4, and mark the rev 4. The 12-minute walk is not on the sheet. Do not run if the traveler rev differs from Rev C.
- Control plan
Every lot, Devon Cho matches the traveler to Rev C and stops to call Mara Ellison on a mismatch. He samples Ra on 1 of 5 and stops to call Priya Shah over 32 µin.
- Check sheet
In the 28 Sep–2 Oct bench sample, the vise jaw was marked for 9 scratches and the traveler header for 2 wrong revs.
- Pareto chart
Over twelve weeks, wrong revision leads with 86 and seat-face scratch has 71. In the last 10 production days, scratches lead with 17 and wrong revision is 4.
- Control chart
Daily seat-face rejects center at 2.75, with an upper limit of 8.91 and a lower limit of 0. The Thursday expedite at 9 is past the limit. The day at 8 is not.
- Histogram
The Ra readings have a mean of 24.54 µin. Two readings, 33 and 35, sit past the spec at 32. The bins are width 3 from 15 to 36.
- Process capability
A naive Cpk is 0.57, from a mean of 24.54 and a sample standard deviation of 4.37. It is arithmetic on an unstable series, not a claim that the seat can hold 32 µin.
- Measurement system analysis
Priya Shah and Noah Kim do not always put P3 and P4 on the same side of 32 µin. The readings are the evidence. There is no percent GRR.
- FMEA
A traveler still showing Rev B can leak at Hartwell. The plant marks that mode 9, 6, and 3 for severity, occurrence, and detection.
- Scatter plot
Scatter puts unchecked travelers against late lots. The marked point is the day cell 2 was clean and the confirm was still late: 0 unchecked, 1 late lot.
- Design of experiments
In eight runs, high clamp adds about six scratches per 20 seats. Coolant and deburr order do not. The high clamp ties to the vise-jaw marks.
- A3
Devon Cho stops the setup when the traveler does not match Rev C. The rule started 21 September. A posted cycle over takt is still open.
- Daily management
On Friday 2 October the cell expected 18 bodies and a rev circled on every traveler. By 14:00 it had 11 bodies and 3 blank travelers.
Inspect the seat
- Process map
Priya Shah inspects the seat at the bench. The count arrives in grease pencil. A hold lives in Hold Log, because the MES hold flag is not trusted.
- Control plan
The plan samples Ra on 1 of 5 and tells Devon Cho to stop and call Priya Shah when a reading is over 32 µin. The plan is not proof the call happened.
- Check sheet
Priya Shah’s bench sample for 28 Sep–2 Oct marks 9 scratches at the vise jaw and 2 wrong revs on the traveler header. It is not the twelve-week Pareto.
- Control chart
The individuals chart is the daily seat-face rejects from the bench. The mean is 2.75. A Thursday expedite at 9 is the point past 8.91.
- Histogram
The readings from the bench have a mean of 24.54 µin. The histogram has no time order, so stability is assumed, not shown.
- Measurement system analysis
Priya Shah measures first. Noah Kim, the Friday backup, measures second. On P3 and P4 the same part can pass with one person and fail with the other.
- FMEA
An inspect skipped on an expedite is marked 8, 4, and 7. The sheet names the mode. It does not chart how often the skip happens.
Kit and pack
- Process map
Luis Ortega kits and packs at the pack bench. Priya Shah’s hold reaches him through Hold Log, not through a trusted MES flag.
- Check sheet
The 28 Sep–2 Oct bench sample marks 2 missing seals at the pack bench.
- Pareto chart
Missing seal is 24 in the twelve-week count and 3 in the last 10 production days.
- FMEA
A seal missing at pack is marked 6, 5, and 4. The current control is the packing count against the list.
Confirm the shipment
- Process map
Luis Ortega confirms from Docknote, then types the confirm into Ledgerline later. It returns to Helen Ward the next morning.
- Value stream map
The median lag from truck departure to the Ledgerline confirm is 16 hours. That is information delay, not touch time.
- Control plan
Docknote and Ledgerline have to agree before Helen Ward writes to Ruth Adeyemi. The plan does not prove the check happened.
- Check sheet
The dock inbox sample for 28 Sep–2 Oct marks 4 confirms on the wrong day.
- Pareto chart
Confirm posted on the wrong day is 22 across twelve weeks and 6 in the last 10 production days.
- FMEA
Docknote and Ledgerline can disagree. The plant marks that mode 7, 6, and 5.
- Daily management
By 14:00 on Friday 2 October, Luis Ortega was already asking for 6 bodies to make a pickup.
One reading
The voice of the customer sets the 32 µin limit. The histogram places the readings against it. The gage asks whether Priya and Noah can tell 32 from 35. The control chart asks whether those readings came from a stable run. Capability sits in the middle of that argument and does not replace it.
What each sheet leaves behind
Boundary
SIPOC
The suppliers, inputs, seven steps, outputs, and customers of one Series 40 promise.
Takes The customer, the promise, and the names of the people and systems on the path. Leaves A bounded path the other sheets can share. It carries no times and no counts.
Voice of the customer
Three sentences in Ruth Adeyemi’s voice, and the CTQs that follow from them.
Takes What the install crew loses when a lot is late or a seat weeps. Leaves On-time to the promise day, Ra at or under 32 µin, one revision in the lot, and a packing count that matches the list.
Path
Process map
Who acts at each step, which system they use, and the six places the lot changes hands.
Takes The seven steps, the person on each, and the system that holds the record. Leaves Six handoffs. The revision often stays in email. The hold lives in Hold Log. The confirm returns the next morning.
Value stream map
Touch, wait, and where the record lives, including the delays in the record itself.
Takes The seven steps, and the times already known for queue, touch, and confirm lag. Leaves Machining marked because 46 minutes of touch sits under 11 hours of queue, and because 46 is already over the 25-minute takt. The email revision and the 16-hour confirm lag marked as information delay.
Spaghetti diagram
Devon Cho’s walk when the traveler revision is blank, and Priya Shah’s walk to the Hold Log.
Takes The places the record can sit: the desks, the cage, cell 2, the cabinet, the printer, the bench, the laptop, the pack table, and the dock. Leaves Devon’s loop when the rev is blank: cell, printer, cabinet, Mara, cell. Priya’s walk from the bench to the Hold Log laptop, which is not at the bench.
Operator balance
The posted 46 minutes stacked against a takt of 25, and the observed 58 when the rev is blank.
Takes The six posted elements and their minutes, the takt, and the observed cycle. Leaves A posted cycle of 46 against a takt of 25. The finish cut is 11 minutes, so another person cannot split it. The observed cycle is 58.
Standard
Standardized work
The posted sequence, the Rev C rule, and the walk that is not on the sheet.
Takes The six elements and times, drawing Rev C, and the rule not to run a mismatched traveler. Leaves A posted method of 46 minutes, a stop rule when the traveler is not Rev C, and a 12-minute walk the sheet does not list.
Control plan
The rev check and the Ra sample on the machine, and the confirm check before Helen writes to Ruth.
Takes The checks the plant already says it runs, and the reaction on a miss. Leaves Rev match every lot, stop and call Mara on a mismatch. Ra sample of 1 of 5, stop and call Priya over 32 µin. Docknote and Ledgerline agree before Helen writes to Ruth.
Evidence
Check sheet
The marks from the bench sample of 28 Sep–2 Oct, by place and by defect.
Takes Marks made in that sample window, not the twelve-week totals. Leaves Traveler header, wrong rev, 2. Vise jaw, scratch, 9. Pack bench, missing seal, 2. Dock inbox, confirm on the wrong day, 4.
Pareto chart
The twelve-week counts and the last 10 production days after the rev rule.
Takes Counts by category for the full window and for the recent window. Leaves The full window is led by wrong revision, at 86. The recent window is led by scratches, at 17, with wrong revision at 4.
Run chart
The twelve weekly on-time percents, the median, and the target.
Takes The weekly on-time percents for the twelve weeks ending Friday 2 October 2026. Leaves A window at 82 percent on time, a median of 82.5, and a target of 97. No control limits.
Control chart
Twenty daily seat-face reject counts against a center line and moving-range limits.
Takes The twenty daily seat-face reject counts. Leaves Mean 2.75, average moving range 2.32, upper limit 8.91, lower limit 0. The value 9 is past the limit. The value 8 is inside.
Histogram
The Ra readings in bins of width 3 from 15 to 36, against the spec at 32.
Takes The Ra readings in the sample. Leaves A mean of 24.54 µin, and two readings, 33 and 35, past the spec at 32.
Box plot
Queue hours when the rev was confirmed before release, beside queue hours when the traveler rev was blank.
Takes The two queues: rev confirmed before release, and rev blank on the traveler. Leaves A confirmed median of 3 hours and a blank median of 13 hours. Not a tested cause.
Process capability
The Ra mean, the sample spread, and the spec, with the arithmetic Cpk kept in its place.
Takes The Ra mean, the sample standard deviation, and the upper spec. Stability comes from the reject chart, not from this arithmetic. Leaves A naive Cpk of 0.57 that is not a capability claim.
Measurement system analysis
Priya Shah’s and Noah Kim’s two trials on five parts, against the spec at 32.
Takes Two trials on each of P1 through P5, in µin Ra. Leaves P3 and P4 cross 32 depending on who measures. No percent GRR.
Cause
FMEA
Four failure modes, the plant’s current 1–10 marks, and the current control in words.
Takes The modes the path already knows, and modest severity, occurrence, and detection marks. Leaves Traveler Rev B at 9, 6, and 3. Skipped inspect at 8, 4, and 7. Missing seal at 6, 5, and 4. Docknote and Ledgerline disagreeing at 7, 6, and 5. Each row names its current control.
Scatter plot
Morning unchecked travelers against lots that missed the promise that day, including the dock-lag day.
Takes Twelve pairs: unchecked travelers in the morning release, and lots that missed the promise. Leaves The pairs as recorded. The dock-lag day is 0 unchecked and 1 late, when the cell was clean and the confirm was late.
Design of experiments
Scratches per 20 seats under clamp, deburr before wash, and fresh coolant.
Takes Eight runs of clamp, deburr before wash, and fresh coolant, with the scratch counts. Leaves High clamp adds about six scratches. Coolant and deburr order do not. The high clamp ties to the vise-jaw marks on the check sheet.
Response
A3
The condition, the target, the why, the countermeasure, the result so far, and what is still open.
Takes On-time ship, the long Pareto, and the fact that Mara is the integration layer between Ledgerline and ShopPacket. Leaves A rule, started 21 September: Mara does not release a mismatch, and Devon stops the setup. A typical week had 11 expedites. The two weeks after were 7, then 4. Still open: confirm lag, and a posted cycle over takt.
Daily management
Friday 2 October, expected against actual, and the list that is still open.
Takes What the shift expected: 18 bodies, the rev circled on every traveler, and no expedite before noon. Leaves By 14:00, 11 bodies, 3 blank travelers, and Luis already asking for 6 to make a pickup. Still open: confirm lag and cycle over takt. The rev rule is in trial, not finished.