Last reviewed: June 20, 2026
Low OEE most commonly comes from Availability losses, specifically unplanned downtime and informal time losses that do not get captured in logs. Before changing equipment, processes, or staffing, run a one-week structured data collection to find where time is actually going. The answer is almost never where the team assumes.
Key takeaways:
Availability losses are the most common cause of low OEE and also the most undercounted, because unscheduled breaks, unofficial extended breaks, and shift-start delays are rarely logged as downtime even though they consume planned production time.
Per Guidewheel's 2026 benchmark data, no-business or no-orders was the single largest loss driver for 37.6% of machines tracked. These losses sit entirely outside OEE and are only captured by TEEP. If your plant has significant idle time from lack of orders, OEE improvement alone will not show the full picture.
In plants we have worked with, 20 to 35% of scrap occurs in the first 15 minutes of a production run. If your Quality component is low, start with setup and first-off inspection, not in-run variation.
Performance losses are the most underdetected component because micro-stops under 5 minutes are rarely logged. When a short stop goes unrecorded, the lost output surfaces as a gap between theoretical and actual output, which is a Performance loss, not Availability. A machine that stops and restarts 30 times per shift for 2 minutes each has lost an hour of production that appears nowhere in the downtime log.
START HERE BEFORE YOU CHANGE ANYTHING
Low OEE has three possible sources: Availability (the machine was not running when it should have been), Performance (the machine was running slower than it should have been), or Quality (the machine produced parts that did not pass inspection on the first attempt). Each component fails for different reasons and requires different countermeasures. Treating the wrong component wastes time and money.
The first step is not to fix anything. It is to read the components.
If you do not know your current Availability, Performance, and Quality figures separately, you cannot know which component is driving the number. The formulas for each component are covered in How to Calculate OEE, Scrap Rate, and First-Pass Yield. The short version: OEE = Availability x Performance x Quality. A 75% OEE from 0.90 x 0.85 x 0.98 has a completely different root cause than a 75% OEE from 0.75 x 1.00 x 1.00.
| OEE PROFILE | PRIMARY DRIVER | WHERE TO LOOK FIRST |
|---|---|---|
| Low Availability, normal Performance and Quality | Unplanned downtime: breakdowns, changeovers, shortages | Downtime log by cause; changeover time records |
| Normal Availability, low Performance | Speed losses: micro-stops, reduced speed, degraded tooling | Actual vs. theoretical output; cycle time observations |
| Normal Availability and Performance, low Quality | Defects: setup errors, tooling wear, incoming material | First-off pass rate; scrap by reason code |
| All three components low | Systemic: unreliable data, or multiple overlapping losses | Start with data collection, not countermeasures |
AVAILABILITY LOSSES: THE MOST COMMON CAUSE
Availability is the component that suffers most in most mid-market plants. Unplanned downtime from breakdowns, unplanned changeovers, tooling failures, and material shortages takes more production time than Performance and Quality losses combined in most operations we have assessed.
Availability is also the most commonly undercounted component. Two reasons.
First, informal time losses that do not get a downtime ticket do not reduce Availability at all on paper. When a short stop goes unrecorded, the machine was nominally running during that time. The lost output surfaces instead as a gap between theoretical output and actual output, which lands in the Performance component. The total production loss is the same, but it gets attributed to the wrong place. The Performance section below covers this in detail. The practical consequence is that Availability looks cleaner than it is, Performance looks worse, and the team investigates speed losses when the real issue is untracked stops.
Second, unscheduled breaks, unofficial extended breaks, and shift-start delays consume planned production time but are rarely logged as downtime events. An operator who starts the line 6 minutes late or takes a 12-minute break in a 10-minute window has not done anything that generates a ticket. Over a week on a multi-shift operation, that loss is material and invisible.
A reliable Availability number requires capturing all planned-time losses, not just the ones that generate a formal ticket.
| AVAILABILITY LOSS TYPE | TYPICAL CAUSE | HOW TO CAPTURE IT |
|---|---|---|
| Unplanned breakdown | Equipment or tooling failure | Downtime ticket at every stop, regardless of duration |
| Changeover overrun | No standard time, no SMED work done | Log planned vs. actual changeover time per run |
| Material shortage | Supply chain delay or internal logistics gap | Log wait time separately from equipment downtime |
| Short stops under 5 minutes | Jams, adjustments, minor quality checks | Tap-to-record at machine; recording these moves them into Availability rather than leaving them as a Performance gap |
| Shift start delay | No standard startup sequence, late arrival | Log actual first-part time vs. scheduled start |
| Unofficial breaks | Extended or unscheduled breaks not tracked | Supervisor sign-off on break start and end times |
PERFORMANCE AND MICRO-STOP LOSSES: THE MOST UNDERDETECTED
Performance losses fall into two categories: speed losses (the machine runs slower than its best-demonstrated rate) and micro-stops (the machine stops briefly and restarts without a formal event being logged).
Speed losses are common when the ideal cycle time used in the OEE calculation is the average rather than the best-demonstrated rate, which masks the loss entirely because the baseline is already depressed. Operators also run at reduced speed to manage a recurring jam, accommodate worn tooling, or work around a quality issue they have learned to live with.
Micro-stops are the most underdetected loss in most plants because they do not get logged. A stop under 2 to 5 minutes does not feel like a downtime event to an operator. Because the stop is unrecorded, the machine is counted as running during that time. The lost output shows up as a gap between theoretical output (run time x ideal cycle time) and actual output, which is a Performance loss. If it happens 20 to 40 times per shift, it can represent 40 to 80 minutes of lost production that appears nowhere in the downtime log and nowhere in Availability.
The way to find micro-stop losses: compare theoretical output (run time x ideal cycle time) to actual output. If Availability looks reasonable (above 0.88) but OEE is still well below 80%, the gap is almost certainly in Performance. A video observation or a simple tally sheet at the machine for one shift will show where the micro-stops are coming from and how frequently. Once they are recorded, they can be moved into Availability tracking; until then, they distort Performance.
QUALITY LOSSES AND WHERE SCRAP ACTUALLY COMES FROM
Quality losses in OEE are defined as parts that did not pass on the first attempt. Scrap and rework both count against the Quality component.
In plants we have worked with, 20 to 35% of scrap occurs in the first 15 minutes of a production run. Setup errors, cold tooling, and first-off parts that fail initial inspection are disproportionate contributors to the Quality loss total. If your scrap data shows high Quality losses overall, check the first-off pass rate before assuming the problem is in-run variation.
A second commonly missed source: incoming material variation. When a new lot of material arrives with slightly different dimensional or chemical characteristics, scrap rates can spike and the cause gets attributed to equipment or operator when it is actually upstream. In plants we have worked with, incoming material accounts for a meaningful share of unexplained Quality variation that only becomes visible when scrap is tracked by material lot, not just by shift.
| QUALITY LOSS SOURCE | WHEN IT IS LIKELY | FIRST CHECK |
|---|---|---|
| Setup and first-off errors | High scrap rate in first 15 minutes of run | First-off pass rate by setup |
| Tooling wear | Scrap rate increases as the run continues | Track scrap rate by hour within a run |
| Incoming material variation | Scrap spikes when a new lot starts | Cross-reference scrap data with material lot numbers |
| Operator variation | Scrap rate differs significantly across operators on the same machine | Compare FPY by operator |
HOW TO FIND YOUR SINGLE BIGGEST LOSS IN ONE WEEK
Run this sequence. One question per day. The first step where the data does not exist is where to start.
Day 1: Read your components. Pull the last 30 days of OEE data broken down by Availability, Performance, and Quality separately. If you cannot do this because the data does not exist, that is the finding: structured data collection comes before root cause analysis.
Day 2: Rank your Availability losses. List every recorded downtime event from the last 30 days by cause and total minutes. Pareto the list. The top two causes typically account for 60 to 70% of total downtime. Also ask: are shift-start delays and informal breaks being captured anywhere?
Day 3: Check your Performance baseline. Compare actual output per shift to theoretical output (run time x ideal cycle time). A gap above 10% is a Performance problem worth investigating. A gap under 5% means Availability and Quality are the higher priorities.
Day 4: Segment your Quality losses. Break scrap out by time within the run (first 15 minutes vs. in-run), by reason code, and by shift. If reason codes do not exist yet, that is the next action before any other quality work.
Day 5: Pick one. The largest loss category in dollar terms is where to start. Do not try to address Availability, Performance, and Quality losses simultaneously. Fix the single largest loss, confirm it held for 30 days, then move to the next one.
FREQUENTLY ASKED QUESTIONS
WHAT IS THE MOST COMMON CAUSE OF LOW OEE?
Availability losses from unplanned downtime are the most common driver in mid-market discrete manufacturing. However, they are also the most likely to be undercounted because unscheduled breaks and shift-start delays rarely get logged as downtime. Before assuming a root cause, verify that all planned-time losses are being captured.
HOW DO I IDENTIFY MY BIGGEST EQUIPMENT LOSS?
Separate your OEE into its three components for the last 30 days. The lowest component identifies the loss category. Within that category, Pareto the specific causes by total minutes or total cost. The top one or two causes typically account for 60 to 70% of the total.
IS LOW OEE A PEOPLE PROBLEM OR A PROCESS PROBLEM?
In most cases it is a process problem that looks like a people problem. Operators run slowly or inconsistently because equipment has recurring jams, tooling is worn, or setup instructions are unclear. Before attributing an OEE gap to operator behavior, rule out equipment, tooling, and process causes. A 5-Why analysis run at the machine with the operator almost always reveals a system cause underneath what appears to be a behavioral one.
WHAT ARE MICRO-STOP LOSSES AND HOW DO I COUNT THEM?
Micro-stops are brief equipment stops, typically under 2 to 5 minutes, that restart without a formal downtime ticket being logged. Because they go unrecorded, they surface as a Performance loss rather than an Availability loss: the machine was nominally running, so the lost output shows up as a gap between theoretical and actual output. You count them by comparing theoretical output (run time x ideal cycle time) to actual output. To identify where they are coming from, use a tally sheet or video observation for one shift and log every stop regardless of duration.