Measuring productivity: output per input, labour measures and overall equipment effectiveness

“What gets measured gets improved” is a business cliché because it is mostly true. If you want faster processes, lower unit costs and more capacity from the same people and equipment, you need to know your productivity, how it is changing and where it is lost. Yet many small businesses track only revenue and profit. Those numbers are important, but they are too broad and too late to show where operations can improve.

This article explains what productivity is, how to choose meaningful measures for people and processes, how to measure equipment performance using overall equipment effectiveness (OEE), and how to avoid measures that encourage the wrong behaviour.

The basic formula

Productivity is simply:

Productivity = Output ÷ Input

Output is what you produce: units, jobs, orders, drawings, services or revenue. Input is what you use to produce it: labour hours, machine hours, materials, energy or money.

Everyday examples make the idea concrete. A dairy farmer might measure litres of milk per kilogram of feed. A cricket selector might look at runs per match. A fabrication shop might measure parts per labour hour, and a design office drawings completed per week. In each case productivity rises if output increases for the same input, or if the same output needs less input.

Choosing the right measures

There is no single correct productivity measure. Choose measures that reflect what matters in your operation.

Partial productivity measures relate output to one input:

• Labour productivity: units per labour hour, jobs per person per week, value added per employee.
• Machine productivity: parts per machine hour, utilisation percentage.
• Material productivity: yield, meaning good output as a share of material input, or scrap rate.
• Energy productivity: output per kilowatt-hour.

Financial measures relate value to inputs:

• Revenue per employee: simple but affected by prices and outsourcing.
• Value added per labour hour: revenue minus bought-in materials and services, divided by hours worked. It is a better measure of what the business itself contributes.

Process measures look at flow:

• Lead time: from order to delivery.
• Throughput: completed units or jobs per period.
• First-pass yield: share completed correctly first time, with no rework.
• On-time delivery: share of orders delivered by the promised date.

Pick a small number of measures, perhaps three to five for the business and one to three for each area, rather than dozens. Make sure they connect to things the business cares about: cost, delivery, quality and capacity.

Human and technological drivers

Productivity improves through two broad levers.

Human factors include skills, training, motivation, clarity of expectations, incentives and the way work is organised. Fair pay, recognition, incentives and opportunities for growth all influence effort and engagement. So do practical things: clear instructions, the right tools at hand and minimal waiting for information.

Technological factors include equipment, tooling, automation, software and layout. Better jigs reduce setup time, automation removes repetitive manual work, and good systems remove re-keying.

Usually both are needed. New equipment run by untrained, unmotivated people underperforms. Highly motivated people fighting poor tools and processes burn out.

KRAs and KPIs: linking measures to people

Two terms are widely used to connect productivity to individual roles:

• Key result areas (KRAs) are the main areas a role is responsible for. A production supervisor’s KRAs might be output, quality, safety and team development.
• Key performance indicators (KPIs) are the specific measures that show how well each KRA is going, such as units per shift, first-pass yield, lost-time injuries and training hours completed.

Good KPIs are specific, measurable, within the person’s influence and few in number. KPIs are often best expressed as ratios, such as output per hour, defects per thousand or orders per enquiry, because ratios allow fair comparison across periods and teams.

Measuring equipment: overall equipment effectiveness

For businesses where machines drive output, overall equipment effectiveness (OEE) is the most widely used measure of how well equipment is being used. It combines three factors:

• Availability: the share of planned production time the machine was actually running (losses: breakdowns, setups, waiting for material).
• Performance: how fast it ran compared with its ideal speed while running (losses: small stops, slow cycles).
• Quality: the share of output that was good first time (losses: scrap and rework).

OEE = Availability × Performance × Quality

A worked example

A machine is scheduled for an 8-hour shift (480 minutes) with 30 minutes of planned breaks, giving 450 minutes of planned production time.

• It is down for 60 minutes (a 40-minute changeover and a 20-minute fault), so it runs for 390 minutes. Availability = 390 ÷ 450 = 86.7%
• Its ideal cycle time is 1 minute per part. In 390 minutes it produces 330 parts, whereas it could ideally have produced 390. Performance = 330 ÷ 390 = 84.6%
• Of the 330 parts, 315 are good and 15 are scrapped or need rework. Quality = 315 ÷ 330 = 95.5%

OEE = 0.867 × 0.846 × 0.955 ≈ 70%

A simpler cross-check gives the same answer: good parts × ideal cycle time ÷ planned time = 315 × 1 ÷ 450 = 70%.

Each factor looks respectable on its own, yet combined they show that 30% of the machine’s potential is being lost. OEE figures around 85% are often quoted as world-class for discrete manufacturing, while many plants operate well below that. More important than any benchmark is your own trend, and understanding which losses are largest.

The six big losses

OEE losses are often grouped into six categories, each with typical remedies:

Recording downtime reasons, even on a simple sheet at the machine, quickly shows which losses dominate. In many small shops, setups and waiting for material or information are the biggest losses, not breakdowns.

Measuring knowledge work

Productivity in offices, design and administration is harder to measure but still possible. Useful measures include:

• Throughput, such as drawings issued, quotes sent or invoices processed per week.
• Turnaround time, for example from enquiry to quote or from order to drawing release.
• First-time-right rate, such as drawings released without revision or quotes without errors.
• Backlog, meaning work waiting, and how it changes.

Combine these with quality measures, because speed without accuracy simply moves the problem downstream.

Avoiding perverse measures

Measures shape behaviour, sometimes in unintended ways:

• Output targets without quality measures encourage rushing and defects.
• Utilisation targets can encourage overproduction of stock nobody ordered.
• Individual targets can undermine teamwork and hand-offs.
• Activity measures, such as calls made, can be inflated without improving results, so pair them with outcome measures.
• Measures people cannot influence demotivate.

Always pair an efficiency measure with a quality or outcome measure, and review measures with the people being measured. If a measure is being gamed, the measure or the incentive is usually the problem, not the people.

Finding the constraint first

In any process with several steps, total output is limited by the slowest step: the constraint or bottleneck. This idea is central to the theory of constraints, popularised by Eliyahu Goldratt. It has an important consequence for measurement. An hour gained at the constraint is an hour gained for the whole operation, while an hour gained elsewhere often just creates more work waiting in front of the bottleneck.

To find the constraint, look for:

• Where work in progress piles up.
• The step that is always busy while others wait.
• The resource most often blamed for late orders.
• The step with the highest utilisation against demand.

Once you know the constraint, focus productivity measures and improvement effort there first. Make sure it never waits for material, information, operators or maintenance. Move inspection or preparation work away from it where possible, and schedule it carefully. Only when it is no longer the constraint should attention move to the next one.

Many small businesses discover that the constraint is not a machine at all. It may be a person, such as the only qualified welder or the one estimator, or a process, such as drawing approval. Measuring flow through each step usually reveals it quickly.

Building a simple productivity dashboard

1. Choose measures: three to five for the business, one to three per area.
2. Define them precisely: formula, data source, frequency and owner.
3. Collect data simply: at the machine, from job cards or from existing systems. Avoid elaborate data collection that nobody maintains.
4. Display visibly: a whiteboard or screen in the work area, updated daily or weekly.
5. Review regularly: in a short weekly meeting, look at trends, identify the biggest loss and agree one improvement action.
6. Improve and repeat: track whether the action worked.

A worked example: improving a machining cell

A small precision engineering business runs two CNC machining centres on a single day shift. The owner feels the machines are always busy, yet orders are often late and overtime is common. Before buying a third machine, the owner decides to measure.

For four weeks, operators record on a simple sheet at each machine: start and finish times, each stop with its reason and duration, parts produced and parts rejected. The supervisor enters the sheets into a spreadsheet each afternoon.

The results surprise everyone. Over the four weeks:

• Availability averages 68%. The largest losses are setups (averaging 75 minutes per changeover, with several changeovers a day) and waiting for material, drawings or first-off approval.
• Performance averages 88%. Small stops for chip clearing and tool changes account for most of the gap.
• Quality averages 97%. Most rejects occur on the first part after a setup.

OEE is therefore about 0.68 × 0.88 × 0.97 ≈ 58%. The machines looked busy because people were always doing something around them, but they were cutting metal for well under two-thirds of the planned time.

The team focuses on the biggest loss first: setups. They introduce standard setup sheets, preset tools offline while the machine is running, build two quick-change fixtures for the most common part families and schedule similar jobs together. They also introduce a rule that material and current drawings must be kitted at the machine before a job is released. Within three months, average setup time drops to about 35 minutes and waiting time falls sharply. Availability rises to around 80%, and OEE to around 69%.

That improvement is roughly equivalent to adding a fifth more capacity across the two machines, enough to cut overtime and delay the purchase of a third machine by at least a year. The measurement cost a few minutes per operator per day and some spreadsheet work. The improvement was worth far more than the equipment budget.

Productivity in service and project businesses

Service businesses such as engineering consultancies, maintenance contractors and design offices measure productivity differently, but the principle is the same. Common measures include:

• Utilisation: billable hours as a percentage of available hours. It is useful, but targets set too high leave no time for improvement, training or business development.
• Realisation: the value actually invoiced compared with the standard value of the hours worked. Low realisation suggests write-offs, poor scoping or scope creep.
• Revenue or gross profit per person.
• Project margin: actual against estimated, by project type.
• Rework rate: hours spent correcting work as a share of total hours.

In project work, the biggest productivity losses often come from unclear scope, waiting for client information and rework caused by late changes. Better scoping and change control usually improve productivity more than asking people to work faster.

Questions owners often ask

Should I measure every machine and person? Start where the constraint is: the machine, process or team that limits overall output. Improving a non-constraint rarely improves total output.

Is higher utilisation always better? No. Running equipment to build stock nobody needs is waste, and keeping people 100% busy leaves no slack for problems or improvement. Aim for high utilisation of constraints on real demand.

How accurate does the data need to be? Accurate enough to show the biggest losses and trends. A simple manual record kept consistently is far more useful than an elaborate system nobody maintains.

How do I avoid people feeling watched? Explain the purpose, involve operators in choosing what to record and focus discussion on process losses, not individuals. When people see measurement leading to better tools and fewer frustrations, resistance usually fades.

Summary

Productivity is output divided by input. Choose a few meaningful measures across labour, machines, materials and process flow, and connect them to roles through KRAs and KPIs. For equipment, use OEE to combine availability, performance and quality, and attack the six big losses, starting with the largest. Improve both human and technological drivers, pair efficiency measures with quality and outcome measures, and review them regularly with the people who do the work. Measured this way, productivity becomes a steady source of extra capacity and lower cost.

---

Sources: small-business training notes on calculating productivity and setting KRAs and KPIs, together with standard operations-management practice on overall equipment effectiveness. The worked example is illustrative.