Weighted OEE rolls the Overall Equipment Effectiveness of several machines or lines into one number by weighting each asset by its planned production time, so an asset that ran all day counts more than one that ran an hour. Weighting by time, not a plain average of the percentages, is the only rollup that keeps the plant number honest, because a small machine's bad morning cannot outvote a big line's whole shift.
The moment you have more than one asset, someone asks for "the plant OEE," and the easy answer is to average the numbers on the board. That answer is almost always wrong. This post shows why the simple average lies, how to weight the rollup correctly, and when machines in a line should be handled differently from independent machines running in parallel. If you are new to the base metric, start with the OEE calculation first; this piece assumes you can already compute a single asset's OEE.
What is weighted OEE?
Weighted OEE is a plant- or area-level score that combines individual OEE figures in proportion to how much production time each asset represents. The cleanest definition skips the percentages entirely: add up the fully productive time across all assets and divide by the total planned production time across all assets. Fully productive time is the time an asset spent making good parts at its ideal rate, the numerator hiding inside every single-asset OEE.
Written the other way, weighted OEE equals the sum of each asset's OEE times its planned production time, divided by the total planned production time. The two forms give the same answer, because each asset's OEE is already its fully productive time over its planned time. The percentage form is easier to explain in a meeting; the time form is easier to trust, because it is just good-output-time over available-time for the whole area.
Why not just average the OEE numbers?
Because a plain average gives every asset an equal vote regardless of how much it ran, and your assets do not run equal amounts. Take two machines: A runs one hour at 50% OEE, B runs eight hours at 80%. The simple average is 65%. But the plant spent nine machine-hours of planned time, and eight of them were on the good machine. The time-weighted answer is (0.50 × 1 + 0.80 × 8) ÷ 9 = 77%. The plant really performed near 77%; the 65% figure would have you chasing a problem that barely touched output.
The distortion runs both ways. Average the numbers and a tiny cell having a terrible day can drag the plant score down into a crisis meeting, while a flagship line quietly losing a few points, worth far more units, gets ignored because its percentage looks fine. Weighting by time puts the attention where the lost hours actually are. That is the whole reason to weight: the plant number should move when real output moves.
How do you calculate weighted OEE?
Gather each asset's planned production time and its OEE for the same window, then combine. The steps are short, and the only discipline is using the same clock and the same window for every asset.
- Fix one time window for all assets. Same shift, same day, same week. Mixing windows is the fastest way to a meaningless rollup.
- Get each asset's planned production time. This is the scheduled time it was supposed to run in the window, after planned stops. It is your weight.
- Get each asset's OEE for that window. Availability times performance times quality, computed per asset the normal way.
- Multiply each asset's OEE by its planned production time. This converts a percentage back into fully productive time, real minutes of good output.
- Sum those products, and separately sum the planned production times. Now you have total fully productive time and total planned time for the area.
- Divide the first sum by the second. That ratio is your weighted OEE. Equivalently, it is total good-output time over total available time.
- Publish the weights alongside the number. A rollup without its weights hides which assets moved it. Show planned time per asset so the number is auditable.
A worked illustration with invented numbers: a filling line has 600 planned minutes at 65% OEE; the case packer beside it has 1,200 planned minutes at 80%. Weighted OEE is (0.65 × 600 + 0.80 × 1,200) ÷ (600 + 1,200) = (390 + 960) ÷ 1,800 = 1,350 ÷ 1,800 = 75%. A simple average would have said 72.5%, close here, but only because the numbers are gentle. Widen the gap in run times and the two answers diverge fast.
Parallel or in a line: when do you multiply instead?
Time-weighting is the right rollup for assets that run independently, two filling lines, a bank of CNC machines, cells making separate output. Their losses do not chain together, so summing their productive time and dividing by their planned time is exactly correct.
Assets arranged in a line, where every part passes through each in sequence, are a different animal. A stop on one station starves the next and blocks the one before it, so the losses interact. For a true serial line the honest number is the line's own OEE, good units out of the line, divided by planned time times the line's ideal rate, measured at the pace-setting constraint not a blend of the stations. Some references multiply the stations' individual OEE figures to model the compounding, and that gives a rougher estimate, but it assumes the losses are independent, which serial lines violate. When in doubt, measure the line as one system at its slowest step. Distinguishing line OEE from cell OEE is a whole topic on its own; the rule of thumb is to weight what runs in parallel and to measure-as-one what runs in series.
What should you weight by?
Planned production time is the default and the most defensible weight, because OEE is fundamentally a ratio of time. Two alternatives show up, and both have caveats.
| Weight by | When it works | Caveat |
|---|---|---|
| Planned production time | Always; the native unit of OEE | None, this is the default |
| Good units produced | Assets making the same part at the same rate | Distorts when ideal rates differ between assets |
| Constraint hours | When one asset sets plant throughput | Ignores non-constraint assets that may still matter |
Weighting by good units is tempting because the counts are handy, but it only equals time-weighting when every asset runs the same part at the same ideal rate. The instant a fast machine and a slow machine are in the mix, unit-weighting overcredits the fast one. Stick with time unless you can prove the rates match. Constraint-hour weighting has its place when a single asset governs plant throughput, but it deliberately ignores everything else, so reserve it for capital and scheduling decisions rather than the routine scoreboard.
What are the pitfalls of a rolled-up number?
A weighted plant OEE is useful for trend and terrible for diagnosis. It tells you the area got better or worse; it never tells you which machine or which loss. Treat it as a headline, and always keep the per-asset numbers and the six big losses underneath it, because that is where the action is. A plant that manages only the rollup will optimize the average and miss the one line bleeding availability.
The second pitfall is comparing rolled-up numbers between plants or areas with different product mixes and schedules. Two plants at "72% OEE" can be running completely different games. Weighted OEE compares an area to its own past, on the same assets and mix. Cross-plant comparison needs the losses broken out, not the headline. Fold the metric into your broader manufacturing KPIs as one signal among several, not the scoreboard.
By the numbers
Rolled-up effectiveness has a national analogue: capacity utilization. Per the Federal Reserve's G.17 Industrial Production and Capacity Utilization release U.S. total-industry capacity utilization ran 76.2% in May 2026, a production-weighted measure of how much of installed capacity is actually in use, roughly three points under its long-run average. It is the same idea as weighted OEE one level up: you do not learn much from an unweighted average of plants any more than from an unweighted average of machines. The weight is what makes the number mean anything. And like OEE, a utilization figure in the mid-70s says most operations have real room to raise output on equipment they already own.
How do you track weighted OEE?
You need the same window, the same definitions, and honest planned-time and count data from every asset, which is exactly where hand-tallied OEE falls apart. When one line logs downtime on paper and another estimates it at the end of the shift, the rollup inherits both errors and multiplies them by time. Plants like CLS replaced paper production logging with real-time capture, so every asset feeds the same clock and the plant number is the same one the floor saw overnight. To see the losses behind a single asset before you roll anything up, put its numbers through a free OEE calculator then combine assets by planned time. For a companion view of the flow metrics that sit alongside OEE, see WIP metrics.