How to Calculate OEE (Overall Equipment Effectiveness), Step by Step

The number your press brake is already generating — whether you track it or not

It's Tuesday morning and your production supervisor flags you: Line 3 ran all shift, but output was 14% below plan. The machine logged no alarms. The parts that did come off passed QC. So what happened?

The answer lives in three ratios you may not be tracking together: how much of your scheduled time the machine actually ran, how fast it ran relative to its rated capacity, and what fraction of output was good on the first pass. Multiply those three percentages and you get OEE — Overall Equipment Effectiveness — the single number that tells you how much productive work a machine actually delivered versus what it theoretically could have delivered.

World-class OEE sits at 85%, built from Availability ≥ 90%, Performance ≥ 95%, and Quality ≥ 99.9% (Tractian citing Nakajima/TPM, 2026). The average across industries is around 60% (InfluxData, corroborated by LeanProduction/Fabrico, 2024). That gap is where margin lives.

By the end of this guide you'll know how to calculate OEE step by step, what inputs you need, and — critically — which factor maintenance has the most direct lever on.

---

The OEE formula in one line

OEE = Availability × Performance × Quality

All three factors are expressed as decimals (0.0–1.0) before multiplying. The result converts back to a percentage for reporting. Every term has a precise definition — here's each one, built from the ground up.

---

Step 1 — Calculate Availability

Availability measures how much of your planned production time the equipment was actually running. Unplanned breakdowns and excessive planned downtime both drag it down.

The formula:

Availability = Run Time ÷ Planned Production Time

• Planned Production Time — the scheduled shift or production window, minus any pre-agreed planned stops (scheduled breaks, planned changeovers, planned maintenance windows). This is the time you committed the machine to production.
• Run Time — Planned Production Time minus all unplanned downtime (breakdowns, unplanned changeovers, material shortages that idle the machine).

Worked example — illustrative inputs:

Input	Value
Planned Production Time	480 min (one 8-hour shift)
Planned stops (breaks, scheduled changeover)	30 min
Net Planned Production Time	450 min
Unplanned downtime (two breakdowns)	45 min
Run Time	405 min

Availability = 405 ÷ 450 = 0.90 (90%)

A 90% Availability result meets the world-class threshold (Tractian, 2026). Each additional unplanned-downtime event chips directly into this factor — which is exactly why maintenance teams who track MTBF and MTTR find themselves looking at OEE Availability as a downstream outcome of those leading indicators. A shorter mean time between failures, or a longer mean time to repair, will both pull Availability below 90%.

For a deeper breakdown of how Run Time and Planned Production Time relate to equipment availability calculations, see the companion guide.

---

Step 2 — Calculate Performance

Performance captures speed loss. Even when a machine is running, it may be cycling slower than its rated or ideal speed — due to micro-stops, reduced feed rates, worn tooling, or operators compensating for a marginal condition.

The formula:

Performance = (Ideal Cycle Time × Total Pieces Produced) ÷ Run Time

• Ideal Cycle Time — the fastest cycle time the machine is theoretically capable of under optimal conditions (from the OEM spec sheet or your own established baseline).
• Total Pieces Produced — all pieces produced during the run window, good and bad.
• Run Time — from Step 1.

Alternatively, if you work in units per hour:

Performance = Actual Output Rate ÷ Ideal Output Rate

Worked example — continuing from Step 1:

Input	Value
Ideal cycle time	1.2 min/piece
Run Time	405 min
Theoretical maximum output	405 ÷ 1.2 = 337.5 pieces
Total pieces actually produced	295 pieces

Performance = 295 ÷ 337.5 = 0.875 (87.5%)

Performance of 87.5% is below the 95% world-class threshold (Tractian, 2026) — meaning the machine ran, but at roughly 87.5% of the speed it should have. That 12.5% speed loss is silent; it shows up as output that's below plan even on a "clean" shift with no alarms.

---

Step 3 — Calculate Quality

Quality measures the proportion of output that meets specification on the first pass, without rework.

The formula:

Quality = Good Pieces ÷ Total Pieces Produced

• Good Pieces — parts that pass first-pass inspection with no rework required.
• Total Pieces Produced — all pieces, including rejects and pieces requiring rework.

Worked example — continuing:

Input	Value
Total pieces produced	295
Rejects + rework pieces	6
Good pieces (first-pass)	289

Quality = 289 ÷ 295 = 0.9797 (≈ 98.0%)

A 98% Quality result is below the 99.9% world-class standard (Tractian, 2026). Six bad parts out of 295 sounds small, but at scale — across multiple assets, multiple shifts — scrap and rework accumulate into meaningful cost.

---

Step 4 — Multiply to get OEE

Now combine the three factors:

OEE = 0.90 × 0.875 × 0.9797
    = 0.771 (≈ 77.1%)

What 77.1% means in plain language: of everything this machine could have produced during the planned shift, it delivered roughly 77 cents of productive output for every dollar of available capacity. The remaining 23% was consumed by downtime, speed losses, and defects.

That result sits above the roughly 60% average OEE seen across industries (InfluxData / LeanProduction / Fabrico, 2024) but meaningfully below the 85% world-class benchmark (Tractian, 2026). The industry breakdown matters too: average OEE in discrete manufacturing has been measured at approximately 66.8%, with highs around 78.2% in medical devices and lows around 57.2% in trailers and RVs (Godlan, 2025).

The multiplication effect is unforgiving. Three individually solid factors — 90%, 87.5%, 98% — combine to 77.1%, not the ~92% average a simple mental model might suggest. This is why improving one factor in isolation has diminishing returns: world-class OEE requires all three factors to be simultaneously high.

---

Where maintenance directly moves OEE

Of the three factors, Availability is where maintenance has the most direct, controllable lever.

Performance and Quality are influenced by process engineering, tooling, operator training, and material quality — all of which sit partly or mostly outside the maintenance department's scope. Availability is different: every unplanned breakdown that consumes Run Time is a maintenance event.

The causal chain runs like this:

1. A PM interval is set too long (or not set at all).
2. A component degrades past its failure threshold before the next scheduled service.
3. The machine goes down unplanned.
4. Run Time drops, Availability drops, OEE drops.

Facilities that manage OEE as a primary KPI have been associated with up to 25% lower maintenance cost and 10%–20% throughput improvements over 18 months (McKinsey, via Cryotos, 2026). The mechanism is straightforward: tighter PM intervals reduce unplanned failures, which protect Run Time, which holds Availability at or above the 90% world-class threshold.

This is also why the connection between MTBF, MTTR, and OEE matters in practice. MTBF and MTTR are leading indicators; OEE Availability is a lagging outcome. If MTBF is falling on a critical asset, the OEE Availability hit is coming — the question is whether you see it before or after the breakdown.

For guidance on setting the PM intervals that protect Availability, see the preventive maintenance interval and cost guide.

---

Putting it together: a quick-reference summary

Factor	Formula	World-class threshold	Example result
Availability	Run Time ÷ Planned Production Time	≥ 90%	90.0%
Performance	(Ideal Cycle Time × Total Pieces) ÷ Run Time	≥ 95%	87.5%
Quality	Good Pieces ÷ Total Pieces Produced	≥ 99.9%	98.0%
OEE	Availability × Performance × Quality	≥ 85%	77.1%

World-class thresholds: Tractian citing Nakajima/TPM, 2026. Industry average OEE ≈ 60%: InfluxData / LeanProduction / Fabrico, 2024.

---

What to do with your OEE number

Calculating OEE once is an exercise. Tracking it over time across an asset — or a fleet — is where it becomes a management tool.

A few practical next steps:

Identify which factor is the constraint. In the worked example above, Performance is the weakest link at 87.5%. That points to a speed-loss investigation — micro-stops, tooling wear, or a process condition — rather than a maintenance-driven breakdown. If Availability were the constraint, the answer would look different: review PM intervals, check MTBF trends, assess whether MTTR can be reduced with better parts staging.

Track trends, not snapshots. A single OEE reading is a diagnostic. A rolling 13-week OEE trend on a critical asset tells you whether your maintenance program is holding the line or gradually losing ground.

Connect OEE to cost. An Availability drop of five percentage points on a machine running two shifts is a meaningful chunk of lost throughput — and it shows up in maintenance cost as reactive repair spend, not in OEE directly. Relating OEE losses back to their cost (labor, parts, downtime consequence) is the conversation that gets PM programs funded.

---

Calculate it in a structured workbook — not a one-off formula

The OEE formula is simple to state and surprisingly easy to miscalculate in practice — particularly the Planned Production Time boundary, the Ideal Cycle Time baseline, and whether rework counts in Good Pieces or Total Pieces (it belongs in Total Pieces, reducing Quality, not Availability).

If you're tracking OEE across multiple assets, or want a structured place to log the inputs and watch the factors over time, the MTBF / MTTR / OEE Calculator Workbook gives you a pre-built calculation layer with labeled inputs for all three factors, the three-way multiplication already wired, and a comparison against the world-class benchmarks. Download it, enter your shift data, and the number is there — without rebuilding the formula from scratch each time.

It's a practical starting point for any maintenance manager who wants to move OEE from a one-time calculation into a standing metric.