How to Calculate OEE on the Floor Without Buying Software
OEE is three numbers multiplied together. Here is the floor-level math, a paper tally sheet you can run on one machine tomorrow, and a worked example.
The first time a vendor showed me an OEE dashboard, the number on the screen was 87 percent and the plant manager next to me was beaming. I asked him one question: how many parts did line 4 actually run last shift versus how many it could have run. He had no idea. The dashboard had eaten the answer and handed back a percentage that made everyone feel good. That is the trap with OEE. It is sold as a screen you buy. It is actually a piece of arithmetic you already have the data for, and the moment you do the arithmetic by hand on one machine, you learn more than any dashboard will tell you.
So put the quote-request form down. You do not need software to start. You need a clipboard, a stopwatch on your phone, and one shift of honest attention.
OEE Is Just Three Fractions Multiplied
OEE stands for Overall Equipment Effectiveness, and people make it sound like a statistics problem. It is not. It is three simple fractions multiplied together:
Availability times Performance times Quality.
Each one is a number between 0 and 1, and each one answers a plain question about the same shift:
- Availability: Of the time the machine was supposed to run, how much did it actually run? (Breakdowns and changeovers eat this.)
- Performance: While it was running, did it run at the speed it is supposed to? (Slow cycles and small stops eat this.)
- Quality: Of the parts it made, how many were good the first time? (Scrap and rework eat this.)
Multiply the three and you get OEE. That is the whole formula. The reason it is useful is not the final percentage. It is that the three losses live in three different departments and three different fixes. A bad availability number is a maintenance and changeover problem. A bad performance number is a speed and small-stops problem. A bad quality number is a defect problem. One number, three doors. The dashboard hides which door. The hand calculation points right at it.
The Numbers You Need (And Where to Get Them)
You need six raw numbers for one machine over one shift. Every one of them is sitting in plain sight on your floor right now.
- Planned production time. The shift minus planned, scheduled stops you do not count against the machine (lunch, scheduled breaks, planned no-demand). An 8-hour shift with a 30-minute lunch and two 10-minute breaks is 480 minus 50, so 430 minutes planned.
- Downtime. Every minute the machine was down when it was supposed to run. Breakdowns, changeovers, waiting on material, waiting on a forklift. Write the reason next to each one. The reasons are gold.
- Run time. Planned time minus downtime. The minutes it was actually making parts.
- Ideal cycle time. The fastest sustainable rate the machine is rated for. Seconds per part, or parts per minute. Use the engineering spec or the best clean run you have ever seen, not a wish.
- Total count. Every part that came off the machine this shift, good and bad together.
- Good count. The parts that passed first time, no rework.
That is it. If you already keep a production tracking sheet with a loss-reason column, you are most of the way there. If you do not, the production tracking sheet I laid out with a “why we missed” column is the exact data feed OEE wants. OEE is downstream of that sheet.
The Three Calculations, Step by Step
Take the six numbers and turn them into three fractions.
Availability = Run Time / Planned Production Time. If you planned 430 minutes and lost 70 to a breakdown and a changeover, run time is 360. Availability is 360 / 430 = 0.837.
Performance = (Ideal Cycle Time x Total Count) / Run Time. This asks: at the rated speed, how long should those parts have taken, versus how long the machine actually had? If ideal cycle is 30 seconds (0.5 min) per part and you made 600 parts, ideal time is 300 minutes. You had 360 run minutes. Performance is 300 / 360 = 0.833. The machine ran slower than rated.
Quality = Good Count / Total Count. If 600 came off and 570 were good first time, quality is 570 / 600 = 0.95.
OEE = 0.837 x 0.833 x 0.95 = 0.662, or 66 percent.
Now look at what that 66 percent told you that a dashboard would have buried. The biggest losses are availability and performance, almost tied, and both are worse than the quality loss. So the dashboard says “you are at 66.” The hand math says “go fix the breakdown and the slow cycle before you touch the scrap, because that is where the minutes are.” That is the difference between a metric and a tool.
A Paper Tally Sheet You Can Run Tomorrow
Here is the sheet. One machine, one shift. Print it, clip it to the machine, have the operator or the lead fill it as the shift runs. No app.
MACHINE: ________ SHIFT: ______ DATE: ________ OPERATOR: ________
PLANNED TIME
Shift length (min) ............... [ 480 ]
Minus planned stops (min) ........ [ 50 ] (lunch + breaks)
= PLANNED PRODUCTION TIME ........ [ 430 ]
DOWNTIME LOG (when machine should run but isn't)
Time Min Reason
______ ___ _________________________
______ ___ _________________________
______ ___ _________________________
= TOTAL DOWNTIME ................. [ ___ ]
RUN TIME = Planned - Downtime .... [ ___ ]
COUNTS
Ideal cycle time (sec/part) ...... [ ___ ]
Total parts run (good + bad) ..... [ ___ ]
Good parts (passed first time) ... [ ___ ]
CALCULATE
Availability = RunTime / Planned ................. ____
Performance = (IdealSec x TotalParts / 60) / RunTime ____
Quality = Good / Total ...................... ____
OEE = A x P x Q .................................. ____ %
BIGGEST LOSS THIS SHIFT (circle one): AVAIL PERF QUALITY
ONE THING TO FIX NEXT SHIFT: _______________________________The two lines that matter most are the last two. The percentage is bookkeeping. “Biggest loss” and “one thing to fix” are the reason you ran the sheet. If you fill those two lines honestly every shift for two weeks, you will out-improve a plant that spent forty grand on a system and never reads the screen.
Stop Chasing the Number, Start Chasing the Loss
A few rules that keep OEE honest instead of turning it into another wall poster:
- Do not benchmark against 85 percent. That “world-class OEE is 85” line is repeated everywhere and it wrecks more programs than it helps. Your baseline is your machine last month. Beat that. The only comparison that means anything is you versus you.
- Never average OEE across machines into one plant number. A plant-wide OEE is a number you cannot act on. Run it per machine, per bottleneck. Fix the constraint, not the average.
- If you cannot name the biggest loss, you measured wrong. OEE that does not point at a door is just a feeling with a decimal point.
- Count rework as a quality loss, not a good part. “We fixed it so it counts” is how quality numbers lie. Good-first-time only.
- The downtime reasons are the real product. The percentage fades. The list of why the machine stopped is what you bring to the daily huddle and the tier meeting tomorrow.
Run the sheet on your worst machine first. Not your best, your worst, the one everybody complains about. You will have a real availability, performance, and quality number by the end of one shift, and you will know which of the three to attack. That is a better day’s work than any demo.
If you want this tally sheet as a clean, fillable page along with a one-machine OEE worksheet that does the arithmetic for you, it is part of my Production Tracking series, where I walk the same paper-first approach across downtime tracking, loss-reason coding, and bottleneck analysis. The sheet will not run the machine and it will not promise you a number. It just turns the data you already have into the three doors you can actually walk through.
Measure one machine. Find the biggest loss. Fix that. Then do it again.
Liked this? Get the free SOP Writing Cheat Sheet.
The 6-step SOP format on one page. Free.