Both methods give the same OEE, because run time and total units cancel when you multiply A × P × Q, which leaves good units × ideal cycle time ÷ planned time. If your two answers differ by more than rounding, your data has an error.
Two routes to one number
The time-based method multiplies three factors: availability, performance and quality. It tells you where the loss is.
The count-based method needs only the good units, the ideal cycle time and the planned time. It tells you the result in about 20 seconds, and Overall Equipment Effectiveness (R. C. Hansen, 2001) calls this simple route Method 3.
Plants tend to treat them as rival formulas. They are the same formula, written two ways.
The proof, in three lines
Write each factor out in full and multiply them. Each term that appears on top and on the bottom cancels.
That has a useful meaning in plain words. OEE is the time the good units needed at full speed, as a share of the time you planned to run.
It also shows something surprising: run time and total units never reach the final result. They only decide how the loss is shared between the factors, a point that matters for stop thresholds and changeover bookings.
Worked example: one CNC lathe
Example: a lathe has 960 planned minutes across two shifts, with 150 minutes of stops. Its ideal cycle time is 1.5 minutes per part, and it makes 470 parts, of which 452 are good.
| Route | Calculation | Result |
|---|---|---|
| Availability | 810 ÷ 960 | 84.38% |
| Performance | (1.5 × 470) ÷ 810 = 705 ÷ 810 | 87.04% |
| Quality | 452 ÷ 470 | 96.17% |
| Time-based OEE | 0.8438 × 0.8704 × 0.9617 | 70.6% |
| Count-based OEE | (1.5 × 452) ÷ 960 = 678 ÷ 960 | 70.6% |
Then add the losses in minutes as a third check: stops 150, slow running 105 (810 − 705), bad parts 27 (18 × 1.5) and good output 678. They add up to 960, exactly the planned time.
The minutes add up for the same reason the two methods agree, and the OEE loss tree names each box.
Why bother with both?
Use the count method to check. You can do it on a phone at the line, and it needs no stop log at all.
Use the time method to act. A 70.6% gives no clue what to fix, while the factors say that availability is the first place to look.
Together they catch errors that neither one can catch alone. The OEE calculator and a simple OEE tracking template both work well for running the two side by side.
When they disagree: what each gap means
A gap is good news, in a way, because it shows exactly where to look. These are the usual causes.
| What you see | Cause | Fix |
|---|---|---|
| Gap under half a point | Rounded factors | Keep decimals until the end |
| Time-based is lower | Performance capped at 100%, or good units used in P | Use total units in P, then look at the raw reading |
| Time-based is higher | Factors averaged across shifts or machines | Add up minutes and units first |
| Gap on one product only | One rate used for a mixed run | Add ideal time by product |
| Gap jumps from day to day | Counts from different points or periods | Take total and good from one point and one window |
Neither route wins by default, so do not report the number you prefer. Find the input that differs, because the two examples below show that either route can be the wrong one.
Gap example: good units in performance
Say someone calculates performance on the lathe with good parts: (1.5 × 452) ÷ 810 = 83.7%. Then 0.8438 × 0.837 × 0.9617 = 67.9%, against a true 70.6%.
The bad parts were charged twice, once in performance and once in quality. See how to calculate OEE performance for the correct input.
Gap example: averaged factors
This one is the most common in monthly reports. Here are two days on one machine, each with 480 planned minutes and 100% quality, and an extreme split on purpose.
- Day 1: runs 240 minutes at full speed. A = 50%, P = 100%, OEE = 50%.
- Day 2: runs all 480 minutes at half speed. A = 100%, P = 50%, OEE = 50%.
Average the factors and you get A = 75% and P = 75%, which multiply to 56.3%. The count method gives 480 ideal minutes ÷ 960 = 50%, which is right, since each day was 50%.
The right performance for the two days is 480 ÷ 720 = 66.7%, because day 2 had twice the run time. How to average OEE covers this in full.
Build the check into your sheet
Add one column to your report, and let it flag any shift where the two routes split.
Watch the units when you build it. A cycle time in seconds with a planned time in minutes gives an answer 60 times too big, a classic route to OEE over 100 percent.
Limits of the count method
- It gives no causes. It cannot tell a slow line from a stopped one.
- It trusts the good count. If bad units are counted as good, the error goes straight into OEE. See good count.
- It trusts the ideal rate. A soft ideal cycle time lifts both methods equally, so they will agree and still be wrong.
- Mixed runs need a sum. Add good units × ideal cycle time for each product before you divide.
That third point is worth remembering: agreement proves the arithmetic, not the inputs. OEE calculation mistakes lists the input errors to rule out next.
For the factor definitions themselves, see availability, performance and quality.
The two methods agree only when stops and counts describe the same time window. Fabrico is a light MES for OEE: it takes stops, speed and counts live from PLCs or IoT sensors, with add-on AI cameras counting output where a machine has no usable signal, and shows OEE per line and shift. See how Fabrico works as a light MES.
What people ask next
What is the simplest OEE formula?
Good units × ideal cycle time ÷ planned production time. It needs just three numbers and gives the same result as A × P × Q. Try it in the OEE calculator.
Why do I need availability, performance and quality if one formula gives OEE?
The single formula gives the result but not the cause. The three factors show whether to fix stops, speed or scrap, and that is what you act on. See how to improve OEE.
Which method should I report?
Report the factors, and use the count method as a check on them. If they differ by more than rounding, fix the data before the number goes on a board.
Does the count method work with several products?
Yes, as long as you work out each product's ideal time and add them up. Good A × ideal cycle time A, plus good B × ideal cycle time B, all divided by planned time. See OEE by product.
Why does OEE not depend on run time in the count method?
Because run time appears on top of availability and on the bottom of performance, so it cancels. It only decides how the loss is shared between those two factors.
Fabrico is a light MES for OEE. It takes stops, speed and counts from PLCs or IoT sensors, with add-on AI cameras counting output where a machine has no usable signal, and operators tag the cause at the line.
Related guides: Performance Over 100 Percent in OEE: Fix Guide · OEE Formula in Excel: Paste-Ready Formulas