OEE Metrics Every Plant Should Track

OEE as a single percentage answers one question: how did the shift go, overall. It’s a useful headline number and a poor diagnostic tool on its own, because a 65% OEE caused by long changeovers needs a completely different response than a 65% OEE caused by chronic micro-stops or a quality problem on one specific station. The metrics underneath OEE are what actually point at which lever to pull.

Availability Metrics

Planned downtime vs. unplanned downtime — separating scheduled maintenance and changeovers from genuine breakdowns is the first split that matters, because they need entirely different responses.

MTBF (Mean Time Between Failures) — how long equipment runs, on average, before an unplanned failure. This is a reliability metric, and a declining MTBF on a specific asset is often the earliest signal that predictive maintenance would catch before a full breakdown does.

MTTR (Mean Time To Repair) — how long it takes to get equipment back online once it fails. A high MTTR often points to a spares, documentation, or troubleshooting-process problem rather than an equipment problem, and is frequently the faster, cheaper thing to fix.

Changeover time — measured per changeover, per operator, per product combination, this is one of the highest-leverage metrics on this list because the variation between the fastest and slowest documented changeover on most floors is a direct, no-capital-required improvement opportunity.

Performance Metrics

Actual cycle time vs. ideal cycle time, tracked continuously rather than sampled — this is what surfaces the “running, but slow” pattern that a simple up/down availability metric completely misses.

Micro-stop frequency and duration — individually too small to matter, collectively often the largest, least-visible performance loss on the floor. Tracking frequency by station, not just total time lost, points directly at which specific point in the process needs attention.

Speed loss by shift or operator — when the same machine, same product, runs measurably faster on one shift than another, that gap is a training and standardisation opportunity hiding in plain sight.

Quality Metrics

First-pass yield — the percentage of units that make it through the full process correctly on the first attempt, without rework. This is the metric that most directly connects to customer-facing quality and warranty cost.

Scrap rate by cause, not just total scrap — a scrap number without a cause breakdown tells you there’s a problem; a scrap number broken down by cause (material, setup, machine, operator) tells you which problem to fix first.

Startup reject rate — rejects in the period immediately following a changeover or restart, tracked separately from steady-state quality, because startup loss is often disproportionately large and disproportionately fixable through a standardised startup procedure.

Where to Actually Start

Not all fourteen at once. Pick the three or four that map to the biggest known pain point — usually changeover time and MTTR if availability is the suspected problem, micro-stop frequency if it’s performance, first-pass yield and scrap-by-cause if it’s quality — get those reliably measured and connected to a dashboard someone actually looks at, and expand from there. SG2’s Manufacturing & Industry 4.0 practice scopes exactly this kind of phased metric rollout as part of the OEE improvement engagement, rather than instrumenting everything on day one of a plant that currently tracks none of it.