Pick the machine that holds up your output, and write down three numbers each shift: scheduled minutes, good units and the ideal cycle time. Then OEE = good units × ideal cycle time ÷ scheduled time. Add a simple stop log in week 2, and by week 4 you will have one clear loss to fix.
Why start with one machine
One machine gives you a real number within a week, with no project and no budget. It also teaches you the rules, so you make your mistakes where they are cheap.
A whole-plant rollout has its place, and how to implement OEE covers it. This page is for the person who wants a first honest number on Monday.
Week 1: three numbers
Choose a machine with a queue in front of it. If it runs short, the plant ships less, so a gain here pays; bottleneck analysis helps if you are unsure.
Then settle three things before the first shift. Write them at the top of the sheet.
- Ideal cycle time. The fastest proven time per unit for this product. Not last month's average. See ideal cycle time.
- Scheduled time. The minutes the machine should run. Decide once whether breaks are in or out.
- What counts as good. Only units right the first time, with reworked units left out.
This short form cannot hide a loss. Stops, slow running and scrap all reduce good units, so they are all in the number.
Getting the count: use the machine's own counter, a tally of cases or a count of finished goods at the end of the shift. If you have none of those, part counting sensors are a small first purchase.
What week 1 looks like
Example: a bagging machine with a 3-second ideal cycle time, scheduled 450 minutes per shift, one shift a day.
| Day | Good units | Good minutes | OEE |
|---|---|---|---|
| Mon | 6,300 | 315 | 70.0% |
| Tue | 6,480 | 324 | 72.0% |
| Wed | 5,040 | 252 | 56.0% |
| Thu | 6,660 | 333 | 74.0% |
| Fri | 6,120 | 306 | 68.0% |
| Week | 30,600 | 1,530 | 68.0% |
A week of this already raises a good question: what happened on Wednesday? Go and ask while people still remember.
Do not worry about whether 68% is good yet. Your own trend matters more at this stage, and that comparison needs the same rules each week.
Week 2: add a stop log
The score says how much you lost, and now you need to know what caused it. Keep a sheet at the machine with start time, end time and a reason.
Use five or six reasons at most, with "other" kept small. The downtime log template and reason codes give a starting list.
Agree a threshold, such as logging stops of 5 minutes or more, and let shorter ones go. Nobody will write down 40 jams of 20 seconds, and the method below catches them anyway.
Weeks 3-4: read the losses
Example: the same week's 2,250 scheduled minutes minus 1,530 good minutes leaves 720 lost. The log explains 450 of them, and the scrap count gives 800 bags × 3 seconds = 40 more.
Whatever is left must be slow running and short stops, even though nobody wrote it down.
That is the benefit of starting from the count. The number you get by subtraction shows the loss that manual logs tend to miss.
You can now split it into the three factors too. Availability is 1,800 ÷ 2,250 = 80.0%, performance is 1,570 ÷ 1,800 = 87.2% and quality is 1,530 ÷ 1,570 = 97.5%.
To look into the hidden 230 minutes, stand at the machine with a tally for 30 minutes and count short stops. What are micro-stops explains what you are looking for.
Week 4 and after: fix one thing
Pick the single loss with the best mix of size and ease, and agree one action with the people who run the machine. Then keep the sheet going and see whether the number moves.
Waiting for material may be the easy win in this example. It takes 150 minutes, and it may only need a change to how material is called off; the OEE improvement action plan shows how to write that up.
Three numbers each shift. Get the score, and ask about the odd days.
Add stops with a reason, at the machine, during the shift.
Rank the lost minutes, including what you find by subtraction.
One action, one owner. Keep measuring, and only then move on.
Tools for one machine
Paper is enough for the log. A spreadsheet saves the sums, and the OEE tracking template has the layout ready.
For a quick check, the OEE calculator needs no setup. The free OEE tracker keeps shift entries in one place if you would rather not manage a sheet.
Traps with a single machine
- A machine inside a line. If it waits for the one before it, code that as starved, and do not treat it as its own fault. See line OEE vs machine OEE.
- Picking the easy machine. A high score on a machine with spare time ships nothing extra.
- Changing rules mid-way. Each change breaks the trend, so note the date if you must.
- Showing the score without the plan. Tell the crew it is for finding losses and is not a judgment of them.
- Growing too soon. Add a second machine once the first has produced a fix that worked.
When the paper log has shown its value and short stops are the main unknown, you need a live count. Fabrico takes stops, speed and counts from PLCs or, on older machines, external IoT sensors, and operators tag the cause at the line. See how Fabrico connects machines.
What people ask next
Can I calculate OEE for just one machine?
Yes, and it is a good way to start. The same formula applies, and you can test your figures in the OEE calculator.
Which machine should I start with?
The one that limits output, usually with a queue in front of it. Bottleneck analysis shows how to find it.
What if my machine has no counter?
Count finished cases or pallets at the end of the shift, and multiply by units per case. Later, a simple sensor can count for you, as retrofit sensors for OEE explains.
How long before OEE on one machine is useful?
A week shows the odd days, and about four weeks shows a ranked list of losses. For a firmer baseline, keep going until a typical product mix has passed through.
Do I need software to track OEE on one machine?
No. Paper and a spreadsheet are enough for the first number. Consider a tool when short stops become the main unknown; OEE data collection methods compares the options.
When should I add a second machine?
When the first machine has produced a fix that held. Then add the next constraint, using the same rules so you can combine the results later.
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: How to Start an OEE Program: The First 30 Days · OEE for Job Shops: High Mix and Short Runs