A central kitchen plan can include an assumed allowance for changing from one product to another. Too little allowance can make the schedule look feasible when it is not; too much can hide usable time. A clearly bounded observation helps the planner examine that input before using it in a production plan.
This guide helps a planner decide whether an observed changeover is clearly bounded enough to record, what it does and does not show, and what still needs observing. It is not an improvement method, a setup procedure or a food-safety process. The example, CO-026, is original fiction.
Why an assumed figure needs a boundary
The central kitchen planning and institutional kitchen guides treat changeover as part of a complete batch cycle, and the prep list guide uses a fixed set-up time. Those are inputs, and an input is only as good as its definition. Two people timing the same changeover can produce different numbers simply because one starts when cleaning begins and the other when the last product leaves.
So fix the boundary first: one shared resource, one product direction, one occasion, a common clock and a stated observation resolution. In CO-026 that is LINE-026 switching from Product A to Product B on 6 November 2026, timed on one clock to the whole minute, and reviewed on 7 November at 17:00 IST.
Define the two boundary events
The start is the last completed output of the prior run. That is the last piece, not the last good piece. If that output was later rejected, it is still the boundary; record its quality status in a separate field. In CO-026 the last Product A output, LAST-A026, came off at 10:00.
The end is the production time of the first next-run output judged acceptable under the site's existing authorised criteria. This guide does not set those criteria. Keep three times apart: when that output was produced, when acceptance was recorded, and when observation stopped. In CO-026, FIRST-B026 was produced at 10:18, acceptance was recorded at 10:20 by the authorised reviewer under unchanged criteria QC-026, and observation ended at 10:25. Trial output TRY-B026 at 10:16 was not accepted; keep its separate evidence rather than using it as the endpoint. The 10:20 entry records acceptance; it does not by itself establish the exact moment the reviewer made the judgement or authorise operational release at 10:18.
Do not substitute a button press, the first attempted output or a later sign-off. If no accepted output or end evidence exists, the interval is unresolved, not zero.
Worked example: CO-026 activity rows
Each row is one person's continuously observed active interval. Split waiting and interruptions out rather than counting the whole task span as active work. P2's packet retrieval overlaps P1's work; this is a fictional observation, not permission to move tasks or skip checks. The later acceptance entry at 10:20 is outside these activity rows, and its effort is not asserted.
| Row (fictional) | Person | Observed active work | Start | End | Active minutes |
|---|---|---|---|---|---|
| T1 | P1 | Required changeover work | 10:00 | 10:08 | 8 |
| T2 | P2 | Retrieves approved job packet | 10:03 | 10:10 | 7 |
| T3 | P1 | Required readiness checks | 10:08 | 10:14 | 6 |
| T4 | P1 | Next-run output sequence | 10:14 | 10:18 | 4 |
| Recorded person-minutes | All rows | Sum of valid rows | 25 | ||
| Elapsed changeover interval | Boundary | LAST-A026 to FIRST-B026 | 10:00 | 10:18 | 18 |
Elapsed interval versus recorded person-minutes
The two totals answer different questions. The elapsed changeover interval, 18 minutes, is how long LINE-026 went between the last Product A output and the first acceptable Product B output. The recorded person-minutes, 8 + 7 + 6 + 4 = 25, is the observed active work in the rows. It is not 25 minutes of downtime, because people can work at the same time.
Eighteen elapsed minutes does not establish how much of the interval was idle. P2's other time is simply not covered by this activity log. Twenty-five is recorded effort, not proof that all effort was captured. P1's rows touch at 10:08 and 10:14 without overlapping. Simultaneous work by different people can contribute separate person-minutes; overlapping entries for the same person's active time would count it twice.
Whole-minute timestamps limit the precision of both measures. A displayed zero for events recorded in the same minute does not prove an instantaneous changeover or no effort. Keep that limit visible and use a finer observation method when the shorter interval matters.
What one observation can and cannot support
CO-026 is one bounded observation. It is not a representative planning standard, a benchmark or a target. Before using it in place of an assumption, the planner should decide what still needs observing: repeated changeovers in the same product direction, on the same resource, under the same procedure, with comparable crew and units, keeping interruptions on record rather than dropping awkward runs.
No saving, critical path, labour cost, throughput or readiness is calculated here. If the observations suggest trying a different arrangement, use the restaurant kitchen workflow guide for authorised trials and the equipment acceptance guide for release decisions, rather than redesigning the changeover from this record. Required cleaning, checks and authorised procedures are referenced as they stand and are not changed by measuring them.
Where the definition comes from
The Lean Enterprise Institute treats a changeover as a transition between products on shared equipment. Its timing boundary includes the end of the outgoing run and extends through the first acceptable output of the incoming run. That general definition informs this guide; its industrial setup examples and linked improvement methods are not applied to a kitchen process. All people, tasks, times and calculations here are original illustrations.
Put it into practice
Your action checklist
Your progress saves in this browser. No sign-up needed.
0 of 10 complete
From reading to doing
Changeover Timing
Open Changeover Timing for one changeover on one shared resource in one product direction. Required inputs are the boundary event times and evidence IDs, confirmation of a common clock and whole-minute resolution, the acceptance recorded time and the observation cutoff. The sheet calculates elapsed minutes from valid bounded event timestamps, row active minutes, and a separately labelled recorded-person-minutes total from up to 12 individual-person activity rows. Each row is one person's continuously observed active interval; split waiting and interruptions out. It detects missing, reversed or out-of-window row times, duplicate row IDs and overlapping active intervals for the same person; touching endpoints are allowed, different people may overlap and row order does not matter. The boundary elapsed result stays independently labelled if the activity log is invalid, and person-minutes are withheld until every populated row is valid. No activity rows means no effort result, not zero. Manual fields cover coverage gaps, the rejected trial output, required procedure references, conditions, actions and review history. It does not calculate a critical path, infer idle time or give operational approval.
Open the Changeover Timing tab in the launch workbook. Keep a copy for your own business.
Download the free workbook See what’s inside the toolkit →Sources & further reading
- Changeover (Lean Enterprise Institute, Lean Lexicon) ↗
Retrieved 8 November 2026. General reference for changing products on shared equipment and defining the elapsed interval through the first acceptable output of the next run. No industrial setup instructions, SMED targets, food-safety parameters, cleaning or engineering controls are adopted. CO-026, its people, tasks, timestamps and arithmetic are original fiction; the worksheet does not authorise operation or establish savings.
Published by Keystonne Grow. Product links connect to the Keystonne ecosystem; check the linked product or service for its current scope.