The 8 wastes of lean: how to spot waste in manufacturing
In one sentence
The eight wastes are classes used to make losses visible; the elapsed time of an event is assigned to one main class, while other relationships are kept as explanations.
By the end of this lesson
- Distinguish the eight wastes using concrete observation criteria.
- Assign the durations of 24 events to classes without duplication and verify the total time.
- Calculate the lost-time share and explain why the remaining time is not automatically value-added.
- Correct a calculation that assigns the same waiting to two classes and select a safe trial.
Why it matters
Many containers in a work area do not show that the required part is available on time. There may be accumulated material of the wrong type while the right part is awaited. The eight wastes help name these different problems. Overproduction means producing before, or in excess of, verified need. Inventory is the state of accumulated material in the process. Waiting is an interval during which the selected person or product cannot progress. These may cause one another, but they are not the same thing. Overproduction may lead to inventory, and inventory to access difficulties. Adding all related losses to the same duration incorrectly inflates the total rather than enriching the explanation.
Motion means human movements beyond what the work requires, such as reaching, turning or searching for tools. Transportation is unnecessary movement of material. Overprocessing is extra work that does not contribute to the agreed need. Defects include nonconformity and the resulting correction or remanufacturing work. Unused talent concerns failure to make use of the knowledge, skills and contributions of the person doing the work. Measuring this last waste only through suggestion counts or speaking time would be incorrect. Investigate how a problem reported by a worker was examined and answered. Seek observable evidence for each category; applying a negative description to a person is not a waste record.
Muda means waste, mura unevenness in workload, and muri pushing a person or equipment beyond safe limits. These concepts make different questions visible. A station that is idle at the beginning of the day and later pressured to catch up with accumulated work may require investigation of unevenness and overburden. Simply trying to reduce idle minutes may not change the source of the problem. A loss may result from poor work allocation or unreliable material arrivals. When selecting a solution, examine how work arrives, the tools required and safe working conditions as well as the worker's movement speed. A waste class is not proof of a cause.
Waste observation is not a ranking of workers. If the same loss can be reduced through a different layout or material supply arrangement, the investigation should focus on improving the system. Quantity, distance, inventory level and time are different measures; they cannot be added directly. Also, having no time recorded in a class does not prove that the class is absent throughout the business. The selected observation represents only limited conditions. In this lesson, 24 events are recorded on a single worker's timeline. All eight categories appear in the report; no times are invented for unobserved categories. This ensures all classes are checked while the total loss consists only of intervals actually measured.
Formula
Total lost time (min) = Sum of durations of non-overlapping loss events
Event count (events) = Sum of unique event counts in the eight classes
Lost-time share (%) = Total lost time ÷ Observation time × 100
Time on other activities (min) = Observation time − Total lost time
How to do it
- 1
Select what to follow and the unit of measurement
Follow a worker, product or piece of equipment; do not change perspective within the same total. If you are examining K01's time, do not add other workers' simultaneous waiting to this timeline. Set the start and end and record the observation scope. Record time consistently in minutes or seconds. Separate columns may be used for inventory quantity in units or transportation distance in meters. These values do not enter the time total. Seeing the same event on another camera or in another note does not make it a new event. Check uniqueness using the record ID, location and time. Stating the measurement boundary clearly makes it possible to understand later whether a high total really reflects loss or confusion over scope.
- 2
Separate observations into events without interpretation
Record searching for a tool in a drawer, stopping until material arrives and moving a container as separate events. Each interval must have a visible start and end. Instead of an interpretation such as the worker is unwilling, record which activity continued until the tool was found. If time was not measured, leave it unknown; do not enter zero because you think it was brief. If searching ends and transportation begins within an event, create two records at the distinguishable boundary. Do not copy the full duration of the same interval into two rows. You can investigate the reason for the observation later; the initial record must preserve what was actually seen. This distinction makes both repetition counts and the time total auditable again.
- 3
Assign one main class and a clear reason
Explain the classification rule before observation. Searching for a tool empty-handed may be recorded as motion, and taking a container an unnecessary distance as transportation. Correcting an incorrect connection is recorded under defects. Do not also add the entire correction interval under overprocessing. Keep the secondary relationship in an explanation column. If a class cannot be selected, leave the record uncertain and verify it instead of forcing a completed classification. The rule is not intended to explain all the event's causes in one word; it ensures the same time is counted once in the total. If observers classify the same record differently, examine the example together. Comparing class percentages before establishing a common definition is misleading.
- 4
Check overproduction, inventory and waiting separately
To identify overproduction, compare the quantity produced with verified need and the time it is needed. Inventory requires quantity, product type and location. Waiting requires recording the time during which the tracked entity cannot progress. A container's waiting and a worker's waiting are not the same timeline. If inventory quantity is unknown, do not estimate units from waiting time. If accumulation is justified as protection against interruptions, ask how the required quantity was determined. Explaining the relationship among these three categories is useful in observation; adding a single interval three times is not. A category with no records may also indicate a data need. Further observation with suitable scope may be needed before concluding that it is absent.
- 5
Examine unused knowledge through evidence
Investigate the record of an access problem K01 reported earlier and the response given. A suggestion not being implemented does not alone show that knowledge was unused; it may have been rejected for safety or feasibility reasons. Check whether a reasoned response was given. Do not automatically record the duration of a discussion as lost time. Discussion needed for improvement is useful support work. Additional work caused by failing to use knowledge can be measured separately; if the same additional time is already in another record, do not add it again. Do not decide on the solution in advance while listening to the worker. Try to understand where the breakdown lies among reporting the problem, assessment, trial and feedback.
- 6
Reconcile event counts and total time
Calculate class totals separately for event count and time. Verify that the event count matches the source records. Then check that none of the intervals overlap on the selected timeline. When tracking a single worker's time, total lost time cannot exceed observation time. Staying below it is not sufficient proof of correctness either; smaller double counting may still exist. Show the remaining time as other activities. Do not label this whole section as value-added work; it may include necessary inspection and other support work. Calculate the ratio only from complete time records. If time is missing, state explicitly which part is covered.
- 7
Turn the priority into a safe trial
Assess duration, recurrence, safety and quality together. The longest loss is not always the first intervention; a serious hazard may require priority. Options may include arranging the point of use to reduce tool searching or changing supply conditions to reduce material waiting. Verify that the new position does not block passage and that the material's weight can be handled safely. Telling the worker to walk faster does not correct a distant layout. Measure again with the same product under similar conditions. If the loss has moved to another class or station, do not hide it from the result. The trial should improve safe flow of an accepted product rather than reduce a single class score.
Worked example
| Item | Value |
|---|---|
| Observation scope | K01; 52 continuous min; 24 loss events |
| Overproduction | 0 events; recorded time 0 min |
| Inventory | 0 events; recorded time 0 min |
| Waiting | 6 events; 2 min + 2 min + 2 min + 2 min + 2 min + 1 min = 11 min |
| Motion | 6 events; 1 min + 1 min + 1 min + 1 min + 1 min + 2 min = 7 min |
| Transportation | 5 events; 1 min + 1 min + 1 min + 1 min + 1 min = 5 min |
| Overprocessing | 4 events; 1 min + 1 min + 1 min + 1 min = 4 min |
| Defects | 3 events; 1 min + 1 min + 1 min = 3 min |
| Unused talent | 0 events; recorded time 0 min |
| Event reconciliation | 0 + 0 + 6 + 6 + 5 + 4 + 3 + 0 = 24 events |
| Time reconciliation | 11 min + 7 min + 5 min + 4 min + 3 min = 30 min |
| Result | 30 min ÷ 52 min × 100 ≈ 57.69% loss; 52 min − 30 min = 22 min of other activities |
The records are non-overlapping intervals on K01's same timeline. Waiting relates to material shortages, motion to tool searching, transportation to an unnecessary container route, overprocessing to an unjustified second wipe, and defects to correcting a nonconforming connection. Each event's duration appears under only one category. All eight classes are visible in the table; three classes have no events included in this time record. This does not prove that the same waste is absent on another shift or within another scope.
First, reconcile the event total to 24 events. Then find the time total of 30 min. More events do not necessarily mean the greatest time impact: motion and waiting have the same number of events, but total waiting is longer. The lost-time share is 30 min ÷ 52 min × 100 ≈ 57.69%. Presented to one decimal place, it is 57.7%. The tasks within the remaining 22 min must be examined separately. This remainder is not automatically value-added.
The table supports investigating material waiting for the first trial; it does not prove that the cause has been resolved. Examine where material is prepared and in response to which request. If there is a safety problem on the transportation route, address it without waiting for its turn in the duration ranking. Another option, such as changing the tool location, may also be tested under the same product conditions. In the next measurement, check whether incorrect connections or load handling increase while time decreases.
When interpreting the class distribution, also state the limitation caused by the choice of observation. Deliberately observing a startup known to lack material can help explain how the problem arises; that observation does not directly give the share for all normal working time. Hiding a nonrandom selection and extending the loss percentage to the whole day is misleading. For the same reason, a record observing only trouble-free work does not prove low losses either. Explain in the report why this interval was selected. If a decision on frequency is needed, plan additional observation covering different conditions; do not fill the gap by extending the existing time total through estimation.
When the class name changes after improvement, check separately whether the actual work has changed. If tool searching is replaced by waiting for a tool to be brought over, motion may fall while waiting increases. This change alone is not considered good or bad; assess total time, conformity and safe access together. Track the time of another person performing new support work under a separate scope too. Work removed from one worker's timeline may not have left the whole system. The trial record therefore explains newly created work as well as the category that decreased. This prevents resolving a loss from being confused with moving it elsewhere.
This dataset was generated synthetically for training; it does not represent a real business, employee or customer study.
When the method misleads
Adding the same waiting twice
After assigning 11 min of material waiting to the waiting class, an observer explains the material's absence from the shelf as an inventory problem. The same 11 min is then added under inventory too. The total becomes 30 min + 11 min = 41 min, and the ratio 41 min ÷ 52 min × 100 ≈ 78.85%. The explanation may point toward a valid investigation of causes; however, the time calculation is wrong because the same interval has been counted twice. A total below 52 min does not remove this error.
Keep the source events when correcting the calculation. Count the 11 min waiting once; move the possible inventory-related cause to the explanation column. Total lost time returns to 30 min, and other activities to 22 min. In the incorrect calculation, other activities would appear as 52 min − 41 min = 11 min. A single duplication therefore also hides other observed work. If inventory needs assessment, measure product type and quantity separately; waiting minutes cannot replace inventory units.
Similarly, it is incorrect to add 3 min of correction under both defects and overprocessing. Select defects as the main class and note which extra operations the correction caused. This choice does not mean there is only one cause. The aim is to preserve causal relationships while respecting the physical limit of elapsed time. After the solution, rebuild the same total; time that appears to disappear through renaming must not be accepted as a real gain.
This dataset was generated synthetically for training; it does not represent a real business, employee or customer study.
Common mistakes
- Double counting: Do not add an interval at full duration under two waste categories; select one main class and keep the secondary relationship in the explanation.
- Mixing units: Do not combine inventory units, transportation meters and lost minutes into one total; each answers a different investigation question.
- Inferring absence from zero: Do not assume a class with no records is absent throughout the business; examine whether the observation boundary suits that question.
- Blaming the person: Do not interpret searching or waiting as unwillingness; investigate the layout, material supply and working conditions using evidence.
- Confusing speed with a solution: Change the cause of an unnecessary route rather than walking faster to a distant tool; preserve safe access and product conformity.
Quizzes and labs open with Yamazo One
A free Academy membership is for reading the lessons. Quizzes, exercises, cases, module tests, labs and the certification exam are open to staff of companies with a Yamazo One license and to Yamazo partners.
If your company has a Yamazo One license, sign in to the Academy with your work email at that company. If you are a partner, sign in with the email of your partner account.
See Yamazo One