Line Efficiency, Labour Productivity and Utilisation: Check the Denominators

Three reports can describe the same shift differently

A factory meeting begins with three numbers. The sewing report shows efficiency of 81.1%. A wider line report shows 66.7%. The labour report shows 2.78 garments per worker-hour. Someone asks which number is correct.

All three can be correct, but they answer different questions. The first two compare standard work earned with different groups of available minutes. The third compares physical output with labour hours. None makes sense until the reader knows what went into the calculation.

This guide follows one fictional shift from the attendance record to the final metrics. It shows how the time records fit together. It also explains which output to count and how to read utilisation alongside efficiency. The figures are teaching examples, not industry targets or observations from a real factory.

The aim is to make a meeting about the process. A change in the report's denominator should not be mistaken for a change in how people worked.

Give each measure a question of its own

Line efficiency compares the standard work earned with the worker time included in the measure. Standard work is expressed in standard minutes, often called SMV. The unit in the numerator is therefore a standard minute, not a garment.

Labour productivity asks how much output was produced per unit of labour input. The BLS explanation of productivity defines labour productivity as output per hour worked. A factory can use a physical measure, such as garments per worker-hour. Different garments, however, contain different amounts of work.

Utilisation needs a further definition. Here, task-time utilisation is the share of assigned operator time spent on defined production tasks. It is a local, stated measure. It is not automatically machine utilisation, capacity utilisation or the availability component of OEE.

The terms become useful when the question appears beside the number. “How much of the assigned time was spent on tasks?” differs from “How much approved work was completed?” A line can be busy for most of the shift while completing little accepted output.

Begin with a time ledger that adds up

The fictional line has twenty operators on its paid roster. Each has 480 paid minutes for the shift. One operator takes a full shift of paid leave. The other nineteen attend. Two of those present spend 120 minutes each in training away from the line.

There is no overtime in this example. Short agreed breaks fall within the 480-minute window; any meal period sits outside it. Training does not overlap the recorded short breaks. The time records are complete and use the same shift cutoff.

These are different views of the same time record. Paid leave has not become a production delay. Training has not vanished. Both remain visible when the report moves from paid time to time assigned to the line.

Four helpers also work on the line for the full 480 minutes. Their 1,920 minutes bring the operator-and-helper total to 10,800 assigned worker-minutes. Supervisors, maintenance staff and other departments are outside this defined line measure.

Paid time is not the same as time worked

The ILOSTAT working-time definitions distinguish hours paid for from hours actually worked. Paid hours can include paid leave. Under this statistical definition, hours worked can include related duties, downtime and short rest periods.

This prevents an easy mistake. A minute without stitches is not necessarily a minute outside work. A person waiting for an approved panel release is still present in the production system. Removing that time from the denominator would change the question being measured.

The training took place away from the line, so its minutes move to a separate activity record. This is a line-allocation choice. It does not claim that all training is excluded from every definition of working time.

A whole-factory report still needs those minutes. Finance may also need all paid hours, including leave, for its own cost question. Reconciliation connects these reports without forcing them to use identical denominators.

If someone transfers between lines during a shift, use the actual time on each line. Counting the person for a full shift on both lines duplicates labour input. Assigning the person to neither makes work disappear from the factory total.

Build earned minutes from the correct styles

During the shift, the line completes 300 accepted garments of style A and 200 of style B. The approved standard for A is twelve minutes; B requires eighteen. Both standards cover the same defined sewing stage and the current approved methods.

Earned minutes are calculated style by style:

Earned minutes = (300 × 12) + (200 × 18) = 7,200.

The ILO and IFC assessment of GEAR explains this approach in its technical notes. The garment SMV converts completed pieces into output minutes.

Do not multiply all 500 garments by whichever style appears at the top of the report. Nor should an unweighted average of twelve and eighteen replace the actual mix. Here, the output-weighted average is 7,200 ÷ 500 = 14.4 minutes per garment.

The example assumes no material change in unfinished standard work across the shift. A sharp change in work in progress can move completed-output efficiency through timing alone. Read the WIP record before judging the shift's performance from that change.

Decide which workers belong in line efficiency

The broad measure includes all operators and helpers assigned to this sewing line. Its calculation is:

Line efficiency = 7,200 ÷ 10,800 × 100 = 66.7%.

The Work Progression and Productivity Toolkit evaluation uses this worker boundary. That report's calculation includes operators and helpers in available worker-minutes. It also warns that differences in factory SMVs can limit direct comparisons.

An operator-only input view instead gives:

Earned minutes per assigned operator-minute = 7,200 ÷ 8,880 = 0.81081.

Expressed as a percentage, that is 81.1%. The higher figure comes from a smaller denominator. No garment was added and no production method improved between the two calculations.

The operator-only view is not an individual skill rating. It credits line output against one subset of the labour supporting it. The helpers' contribution remains real, so comparisons must keep their treatment visible.

Suppose a helper starts carrying bundles that an operator previously carried. An operator-only report may improve while total line labour stays the same. The broad measure can help test what changed. Did the combined process improve, or did work move between job titles?

Reconcile the four time views before choosing one

Paid operator time is 9,600 minutes. Adding the helpers' 1,920 paid minutes gives 11,520 paid minutes for the whole group. Leave and off-line training explain why this exceeds the 10,800 minutes assigned to the line.

These ratios should not all be labelled simply “efficiency.” The first and last describe paid-resource coverage. The middle two describe assigned labour with different worker boundaries. Choose the denominator to fit the management question. A more attractive percentage is not a sound reason to change it.

The reconciliation also exposes incomplete records. The operator ledger must satisfy 480 + 240 + 8,880 = 9,600. Adding helpers then gives the wider totals. If the totals do not agree, check the time records first. A small movement in the ratio can wait.

A daily board may need only one main efficiency measure. The supporting worksheet can hold the other views. A reviewer can then explain differences without rebuilding the shift from memory.

The same ledger explains why a calculated headcount can be a fraction. Divide the 8,880 assigned operator-minutes by 480 minutes per shift. The result is 18.5 full-shift equivalents. Nineteen people attended, but two spent part of the day elsewhere. The figure 18.5 describes allocated time; it does not claim that half a person attended.

Using nineteen full shifts in the line denominator would put the 240 training minutes back into the line. That may suit a broader resource measure, but it would contradict the assigned-time rule used here. Where transfers are common, use the time total directly. Rounding the equivalent headcount to a whole number would change that total.

Equivalent headcount cannot show whether a station had safe staffing at a given moment. That question needs the actual roster and time sequence. The daily total does not show whether trained people were available together when needed.

Convert minutes to worker-hours before calculating productivity

The broad line total of 10,800 worker-minutes equals 180 worker-hours. Physical labour productivity for the defined group is:

500 accepted garments ÷ 180 worker-hours = 2.78 garments per worker-hour.

The shift itself lasted eight elapsed hours. Dividing by eight gives 62.5 garments per elapsed hour. That describes the line's output rate through clock time. It does not describe output per worker-hour.

The units expose the difference. One hour involving several people contains several worker-hours. A report that writes “pieces per hour” without naming the type of hour leaves the reader to guess.

The 2.78 figure also depends on product mix. More of the simpler style may raise pieces per worker-hour without improving the method. For mixed products, read pieces per worker-hour alongside standard work earned. Keep the style breakdown with both.

Standard-minute efficiency does not eliminate every comparison problem. A changed method, allowance or work boundary can change SMV. ApparelDiary's SMV calculation guide explains that foundation. Here, the task is to keep the standards and labour inputs consistent.

Separate task-time utilisation from earned work

The nineteen attending operators have 8,880 minutes assigned to the line after off-line training. Suppose complete activity records show 8,000 minutes on defined production tasks. They also show 500 minutes waiting and 380 minutes of agreed short breaks.

The categories do not overlap, and they cover the full assigned time:

8,000 + 500 + 380 = 8,880 minutes.

Task-time utilisation is 8,000 ÷ 8,880 × 100 = 90.1%.

The task category includes the handling and checks required by the defined work. It does not mean the needle was moving throughout those 8,000 minutes. A machine counter alone would not establish this whole category.

The 500 waiting minutes remain inside assigned time because the operators were still assigned to the line. They need cause records before anyone can decide what could be reduced. The 380 break minutes remain a separate, visible category; this calculation does not justify removing breaks.

The useful comparison is between activity and output. Suppose task time is high but earned output is low. The team might examine the method, quality, work mix or standard. It cannot establish which explanation is correct on its own.

Use the multiplication identity with matching boundaries

Let E mean earned minutes, A assigned operator-minutes and T logged task minutes. Provided the measures refer to the same output stage, workers and period:

E ÷ A = (T ÷ A) × (E ÷ T).

For this example:

(8,000 ÷ 8,880) × (7,200 ÷ 8,000) = 7,200 ÷ 8,880 = 0.81081.

Task-time utilisation is about 90.09%. The earned-minute ratio during logged task time is 90.0%. Their product gives the operator-only ratio of about 81.08%. Use the unrounded values when checking the identity.

Coats Digital's operator-efficiency guide separates available-time efficiency from time on standard work. This is a useful distinction. The local definitions still need checking. In our example, SMV contains its approved allowances, while the logged task category excludes the separately recorded breaks.

For that reason, E ÷ T is not a pure rating of physical pace. Nor can it be used to claim that every remaining minute could be recovered. The arithmetic is exact. What it means depends on how the standard and activity records were defined.

Do not multiply this operator utilisation by a machine-performance percentage from a different report. Both may be percentages, but they cover different people or equipment.

Keep quality and rework rules beside the numerator

The example credits a garment once when it is accepted at the defined sewing-stage boundary. If a garment needs repair before acceptance, the repair time remains in the line's labour records. Once accepted, it earns one garment's standard work. Passing through twice does not earn a second credit.

This rule prevents a busy repair loop from inflating output. It also means the reported accepted-output count differs from a first-pass count. The quality report should keep repairs visible and separate first-pass success from final acceptance.

If a fault is found after a garment was already credited, follow a clear correction rule. Preserve the original record and the reason for changing it. Otherwise, yesterday's output and today's rework may enter different systems with no visible connection.

A daily cutoff matters as well. A piece repaired after the shift cannot be silently treated as accepted before the cutoff. A consistent rule may create temporary timing differences, but those differences can be explained. An inconsistent rule creates a trend that nobody can interpret.

The same care applies to operations and complete garments. Do not add intermediate operation counts to finished-garment counts. That could credit the same work twice. A measure built at operation level needs its own clear counting rules.

Keep OEE separate from a labour-efficiency ratio

OEE is often placed beside line efficiency on a dashboard. Vorne's OEE calculation guide bases OEE on availability, performance and quality for a defined process. Its simplified formula uses good count, ideal cycle time and planned production time.

Those inputs differ from a labour ratio based on garment SMV and worker-minutes. The OEE guide also defines good parts through first-pass success, excluding parts that need rework. That is different from the final accepted-output rule in our teaching ledger.

A machine could run throughout an automatic cycle while its operator prepares another item. Conversely, an operator could perform a necessary manual check while the machine stands still. Swapping these time records would change what the measure describes.

Keep the process, time basis, cycle standard and quality rule attached to each metric. OEE can support an equipment investigation. Labour efficiency can support a workforce and work-flow investigation. Neither is a universal percentage of how well the whole factory is managed.

Combine line results by adding their inputs

A separate example shows why averaging percentages can mislead. Line X earns 900 standard minutes from 1,000 available worker-minutes. Line Y earns 4,500 from 9,000. Their efficiencies are 90% and 50%.

The simple average is 70%. Yet the combined calculation is:

(900 + 4,500) ÷ (1,000 + 9,000) × 100 = 54%.

The larger line supplies most of the available minutes, so it must have more weight in the combined result. Add earned minutes and available minutes first, then divide. Weighting each line's percentage by its available minutes gives the same answer.

This rule also applies when combining shifts or days that use the same definition. It does not solve a mismatch in standards, worker scope or quality rules. Harmonise those inputs before pooling them.

Retain the individual results too. A correct combined percentage can still hide an unusual line or a change in the share of work assigned to each one. The total is a useful summary, not a replacement for the underlying records.

Distinguish efficiency from meeting the plan

Return to the main shift and suppose its agreed planning assumption was 70% broad line efficiency. With 10,800 assigned worker-minutes, that plan called for 7,560 earned standard minutes. The actual 7,200 therefore achieved 7,200 ÷ 7,560 × 100 = 95.24% of the earned-minute target.

The line can meet 95.24% of this target while recording 66.7% actual efficiency. The two percentages have different denominators: planned earned minutes in one case, available worker-minutes in the other.

The shortfall is 360 standard minutes. Keep the same 60:40 style mix. Fifteen extra A garments and ten extra B garments would then represent that amount: (15 × 12) + (10 × 18) = 360. This is a mix-specific conversion, not a promise that twenty-five more garments were feasible during the shift.

The 70% planning assumption is invented for the example. It is neither a universal requirement nor proof of safe capacity. A useful review would ask why the plan was set. Were its inputs still valid, and which delays affected the result? It would also check quality and the unfinished-work boundary before treating the shortfall as lost production.

Check whether an apparent gain is only a reporting change

Suppose yesterday's report included helpers and today's excluded them. The displayed percentage could jump even if every physical record stayed the same. In the main example, this alone changes 66.7% to 81.1%.

A standard revision can have a similar effect. More standard minutes awarded to the same output raise earned minutes. That may correct an old error, but it is not evidence of a new production gain. Keep the effective date and reason for the revision.

When a change is necessary, show a bridge. Recalculate a comparison period on the new basis where the records support it. Preserve the originally reported result as well, so the correction remains traceable.

Do not alter the standard merely to fit a desired efficiency. If a result exceeds 100%, check the output, time allocation, method and standard. A standard-minute ratio is not automatically bounded like a share of recorded time. Investigate the unusual result before drawing conclusions about effort or data fraud.

Make the metric definition easy to find

A useful definition fits beside the worksheet. It names the output stage, accepted-count rule, SMV version, worker group, time basis and reporting cutoff. It also identifies who can approve a correction.

The source fields should be clear enough for another person to rebuild the number. Attendance shows who was present. Transfer records show their line assignments. Production records show accepted output by style. No single department's sheet proves all three.

Missing information deserves a visible status. An unconfirmed transfer should not quietly become zero minutes. A provisional result can still help if its unresolved input and likely effect are clear. Once the input is confirmed, correct the record and preserve the reason.

ApparelDiary's factory KPI dashboard guide explains how to present defined measures. The essential work happens before presentation: reconcile the ledger, preserve the work boundary and choose a ratio that answers the intended question.

In the fictional meeting, the three opening figures no longer conflict. Each can be traced to the same shift, with its own unit and denominator. That gives the team a sound basis for discussing delays, product mix and the next process improvement.

References

• US Bureau of Labor Statistics: productivity concepts and labour hours.

• ILOSTAT: wages and working-time concepts.

• ILO and IFC: assessment of the Gender Equality and Returns programme, 2025.

• Work Progression and Productivity Toolkit: final evaluation, 2018.

• Coats Digital: operator efficiency and standard-minute definitions.

• Vorne: OEE calculations, time boundaries and first-pass quality.