Overall Equipment Effectiveness is the most widely cited metric in manufacturing. It is also one of the most consistently misused.

Walk into almost any factory and you will find OEE on the daily management board. Ask the plant manager what their OEE is and they will tell you. Ask them whether they trust the number and — if you have built enough rapport to get an honest answer — most will hesitate.

That hesitation is important. It is telling you something that the metric itself cannot.

What OEE Is Supposed to Measure

OEE is elegant in its design. It captures three dimensions of equipment performance in a single number: Availability (is the machine running when it should be?), Performance (when it runs, is it running at the right speed?), and Quality (of everything it produces, how much is right first time?).

Multiply those three together and you get OEE. A score of 85% is frequently cited as world-class for discrete manufacturing. Most factories sit somewhere between 40% and 70% and believe they are doing reasonably well.

The intent is right. A single integrating metric that connects uptime, speed, and quality is exactly what operations leaders need. The problem is not what OEE measures. The problem is how it gets measured — and what it quietly ignores.

The Six Ways OEE Gets Gamed

Planned downtime manipulation. OEE is calculated against planned production time, not total available time. This means that any time classified as “planned” — changeovers, scheduled maintenance, team briefings, even shift handovers — disappears from the denominator. Factories with aggressive planned downtime classification can report OEE figures that look excellent while actually running far less productively than the number suggests. If you want an honest picture, calculate Overall Factory Effectiveness (OFE) against 24/7 calendar time. The gap between your OEE and your OFE is telling.

Ideal cycle time inflation. Performance is measured against an ideal or theoretical cycle time. In many factories, that ideal cycle time was set years ago and has never been updated. If the standard reflects an aspirational rate that the equipment was never consistently capable of, Performance will always look artificially low — and the organisation stops treating it as a real improvement target. Conversely, if the ideal cycle time is set too conservatively, Performance looks excellent while real capacity is being left on the table.

Quality measured at the machine, not at the customer. First pass yield at the point of production and quality as experienced by the customer are often very different numbers. Rework done offline, inspection sorting, concessions raised after the fact — none of these necessarily appear in the OEE quality calculation. A machine producing 95% first-pass yield sounds impressive until you discover that the 5% is being reworked by a team of three people downstream, costing more in labour than the original production saved.

Chronic losses normalised. Small stops — interruptions of less than five or ten minutes — are frequently not recorded as downtime. They are absorbed into Performance as speed loss or simply not captured at all. Yet in many factories, chronic small stops account for more lost production than the major breakdowns that everyone is focused on. The breakdown gets an action plan. The small stops get ignored because no single event is large enough to trigger formal attention.

Asset selection bias. OEE is commonly reported for the assets that are easiest to measure or that have the best data infrastructure. In plants with mixed equipment ages and data maturity, the OEE figure on the board may represent a curated subset of the factory floor rather than an honest whole. The bottleneck asset — which is the only one that truly matters for throughput — may not be the one being reported.

The benchmark trap. The 85% world-class figure is a composite benchmark from a 1982 study of Japanese automotive manufacturing. It is not a universal standard. The appropriate OEE target for a high-mix, low-volume job shop is fundamentally different from the appropriate target for a dedicated, high-volume automotive press line. Applying a single benchmark across dissimilar assets and production environments creates either false comfort or false alarm — neither of which drives improvement.

What OEE Does Not Tell You

Even a perfectly measured OEE has blind spots that operations leaders need to understand.

OEE does not tell you whether you are making the right products. A machine running at 90% OEE producing items that are not needed is not contributing to business performance — it is building inventory. OEE optimisation and scheduling optimisation are separate problems, and confusing them is how factories end up with excellent equipment utilisation and a warehouse full of slow-moving stock.

OEE does not tell you about the cost of the output. A line running at 70% OEE with low labour cost and minimal scrap value may be significantly more profitable than a line running at 85% OEE with high consumable costs and significant rework. Operational efficiency and economic efficiency are related but not the same.

OEE does not tell you about the system. A bottleneck machine running at 65% OEE is more valuable to improve than a non-bottleneck running at 50%. OEE is an asset-level metric. Theory of Constraints thinking reminds us that the constraint is what limits the system — and that improving non-constraints improves nothing except the OEE dashboard.

How to Use OEE Correctly

None of this means OEE should be abandoned. It means it should be used carefully.

Start with the bottleneck. Identify the asset or process step that constrains your throughput and measure OEE there with rigour. Improving OEE at the bottleneck by five percentage points is worth more to the business than improving it everywhere else by ten.

Measure losses, not the score. The OEE percentage is a summary. The value is in the loss decomposition — how much time was lost to breakdowns, to changeovers, to minor stoppages, to speed loss, to startup rejects, to scrap? That breakdown tells you where to focus. The score tells you nothing about what to do next.

Set asset-specific targets. Benchmarks are starting points for conversation, not destinations. A realistic OEE target for each significant asset should be based on its design capability, its maintenance history, its product mix demands, and the commercial priority it serves. A blanket target applied to a diverse asset base is a vanity metric.

Connect OEE to financial outcomes. The reason to improve OEE is to either reduce cost or increase throughput. Make that connection explicit. A 5% improvement in OEE on your primary bottleneck translates to X units of additional capacity, worth £Y of revenue or £Z of avoided capital expenditure. When leaders understand the financial leverage of OEE improvement, they invest in it differently.

Use trend, not point-in-time. A single OEE figure tells you where you are today. A twelve-month OEE trend tells you whether your maintenance strategy, your changeover programme, and your operator capability are improving. The trend is the signal. The point-in-time figure is just noise.

The Honest Conversation

The most valuable thing an operations leader can do with OEE is sit down with the team that produces the number and ask: do you trust this? What would the figure look like if we measured everything? What are we not capturing?

That conversation — uncomfortable as it sometimes is — is worth more than any dashboard.

OEE is a tool. Like every tool, its value depends entirely on how honestly it is used.

Adam