Common Causes of Low OEE (and Which Ones Are Actually Fixable)

“Our OEE is low because the machines are old” is the most common explanation offered, and it’s usually wrong — or at least incomplete. Old equipment can genuinely limit performance, but in most plants that investigate the actual data instead of assuming, the real losses are scattered across a handful of specific, addressable causes that have very little to do with equipment age.

Availability Loss: Where It Actually Hides

Unplanned breakdowns are the obvious one, and the one every plant already tracks — but they’re rarely the largest availability loss once measured properly.

Changeovers are the one that surprises people. Changeover time that varies by 30-40% between operators on the same machine, same product, isn’t equipment variation — it’s a process that hasn’t been standardised, and the fastest documented changeover on your floor right now is usually evidence of exactly how much time is available to recover if the method gets taught consistently.

Waiting for material looks like a production problem but is usually a planning and visibility problem — the machine is available and functional, but starved of input because nobody upstream had real-time visibility into what the line actually needed and when.

Undocumented minor stops — a five-minute adjustment here, a two-minute jam clear there — rarely get logged individually, but they accumulate. A machine that stops for two minutes fifteen times a shift has lost half an hour that a manual shift log will never capture as a single line item.

Performance Loss: The One Everyone Underestimates

Performance loss is the ratio between what a machine actually produced and what it could have produced running at its rated speed for the time it was actually available — and it’s consistently the factor plants most underestimate, because “the machine was running” feels like a win even when it was running slow.

Running below rated speed without a documented reason is the most common source, often because an operator dialed back speed months ago to manage a quality issue that’s since been resolved, and the speed setting never got revisited.

Micro-stops that don’t trigger a full downtime event — the machine pauses, restarts, pauses again — degrade the performance number without ever showing up as availability loss, because the machine technically never went fully offline long enough to count.

Quality Loss: Usually a Data Problem Before It’s a Process Problem

Startup rejects after a changeover or a stoppage are often treated as an unavoidable cost of doing business rather than a measured, trackable loss — which means nobody notices when the startup reject rate on one line is three times higher than another running the same process.

Rework counted as “good” production inflates the OEE number in a way that hides a real problem — if a unit takes two passes to get right, the OEE calculation should reflect the actual first-pass yield, not the eventual outcome after correction.

Disconnected quality and production systems mean a unit rejected in the lab three hours after it was counted as “produced” on the floor rarely makes it back into the OEE calculation at all, which is why quality loss is so often the smallest, least trusted number of the three — not because quality problems are rare, but because the data connecting them to OEE is incomplete.

Diagnosing Instead of Guessing

The plants that actually fix OEE don’t start with a capital equipment request. They start by connecting machine-level data, downtime-reason tagging, and quality records into one continuous view — the same architecture covered in Understanding OEE Beyond the Formula — and let the data show which of these specific, addressable causes is actually driving the number, instead of assuming it’s the machines. SG2’s Manufacturing & Industry 4.0 practice starts every OEE engagement with exactly that diagnostic step, before any equipment or process change gets recommended.