Power BI OEE Dashboard: Build It Without Wrong Numbers

Key takeaways

• Sum first, divide last: OEE = good count x ideal cycle time / planned production time.
• Never average OEE percentages across shifts, lines or plants.
• Four tables are enough: shift calendar, stops, counts and ideal cycle times.
• The report is only as good as the capture behind it. Power BI shows data; it does not collect it.
• Show minutes and money next to every percentage.

Power BI can draw a good OEE dashboard. The trouble is the maths and the data underneath it. Build OEE from summed minutes and counts, never from an average of percentages, and feed it from stop records with real timestamps.

Fabrico, OEE and light MES software with AI cameras, computes OEE live by line and shift from the machine signal. Building your own report instead, or next to it? This guide shows the data model and the one formula that keeps the numbers right.

What data does a Power BI OEE dashboard need?

Four tables cover it. Link them by date, shift, line and SKU.

Keep one unit per line. If the filler counts bottles and the packer counts cases, convert cases to bottles with the pack size before anything is summed. Mixed units are a quiet source of wrong totals, and nobody spots them on a chart.

The stops table should come from the machine signal, with timestamps set at the source. Our OEE data model describes these tables and fields in more detail.

How do you calculate OEE in Power BI?

Use the standard definitions, and build each one from sums:

• Availability = run time / planned production time.
• Performance = ideal cycle time x total count / run time.
• Quality = good count / total count.
• OEE = good count x ideal cycle time / planned production time.

The last line is the one to trust, because it works at any level: one shift, one week or ten plants. In Power BI terms, the OEE measure looks like this:

OEE = DIVIDE(SUMX(Counts, Counts[Good] * RELATED(IdealCycle[Minutes])), SUM(ShiftCalendar[PlannedMinutes]))

Each unit carries the ideal cycle time of its own SKU, so a mix of products is handled. Our guide to OEE calculation walks through the formula step by step.

Worked example: why averaging OEE percentages goes wrong

Sample line (illustrative numbers, not a customer). One line runs two shifts on the same day. Shift A is a full shift of 480 planned minutes. Shift B is a short shift of 120 planned minutes.

Good output time is good count x ideal cycle time. The average of 80% and 40% is 60%. But the short shift weighs four times less than the full one. Summed, the day is 432 / 600 = 72%.

The same error hides in plant and group reports. Average ten plants' OEE and a small site counts as much as the biggest. Sum their minutes instead.

Next to the percentage, show what it means. The day lost 600 minus 432 = 168 minutes of good output time. Priced at the line's margin, that is the number a plant manager acts on.

What can Power BI not do for OEE?

• It does not capture stops. If stops are typed in after the shift, short stops are missing and long ones get rounded.
• It does not ask for reasons. Someone at the line still has to say why each stop happened.
• It is not a line screen. A scheduled refresh suits the weekly review, not the operator who needs to see a stop now.
• It does not fix definitions. If two plants count planned time differently, the report compares two different things.

Our article on why your OEE number is wrong lists more of these traps.

Should you build it in Power BI or buy OEE software?

Often both. They do different jobs.

A common setup: OEE software captures and confirms the data at the line. Power BI then combines it with finance or supply data for head office. Ask any OEE vendor how you can get your data out before you sign. See our guide to OEE dashboards for what to show on each screen.

Where does Fabrico fit?

Fabrico is a computer vision enabled Manufacturing Performance Platform. It is the capture and OEE layer: a light MES with AI cameras on stops.

• Automatic downtime logging from the machine signal, with real timestamps.
• The operator confirms the reason in one tap, so each stop has a cause.
• OEE live by line and shift, summed the right way at every level.
• Losses ranked by money, and one view across plants for groups.

How Fabrico data reaches your other tools is a question for Fabrico's technical team. You start with a pilot on one line. There is no free plan and there are no tiers: each plant gets a quote.

Frequently asked questions

How do you calculate OEE in Power BI?

Build a measure that sums good count times the ideal cycle time of each SKU, then divides by the summed planned production time. That one formula gives the right OEE for a shift, a week or a whole group. Avoid measures that average the OEE of individual rows.

Can you average OEE across lines or plants?

No. A plain average gives a small line the same weight as a big one, and a short shift the same weight as a full one. Sum the good output time and the planned time across lines or plants first, then divide. The result is the true combined OEE.

What data do I need for an OEE dashboard?

You need a shift calendar with planned minutes and a stop table with start times, end times and reasons. Add counts of total and good units, and an ideal cycle time per SKU and line. The stop table should come from the machines, not from notes typed at the end of the shift.

Is there a ready Power BI OEE template?

Many templates exist online, and the visual part is the easy bit. Before you use one, check that it sums minutes and counts before it divides. Also check where its data comes from. A neat template fed by typed stop logs still shows the wrong losses.

Power BI or OEE software for the shop floor screen?

For the screen at the line, use OEE software that shows stops as they happen and asks for the reason. Power BI suits the weekly and monthly reports, where a scheduled refresh is fine and data from other systems can be added alongside.

See the numbers behind your OEE report

Bring last week's OEE report. We will show the stops, minutes and money behind it, line by line.

Request a demo · Contact us