An MES dashboard shows the live state of orders and lines: what is running, how far each order has got, when it will really finish, what stopped, and what is on quality hold. OEE is one tile among those. A good tile also names who acts on it, and at what threshold.
Orders are what make it an MES dashboard
An OEE dashboard asks how well the equipment ran, and OEE dashboard examples shows the layouts for that. An MES dashboard adds the question a supervisor gets asked all day: will this order ship on time?
That needs data a machine view does not hold. It joins counts with the order, the due time, the quality status and the material.
The six tiles worth having
| Tile | Data from | Refresh | Example threshold | Owner |
|---|---|---|---|---|
| Line state and stop | Machine signal, operator cause | Seconds | Stop over 5 min | Supervisor |
| Order progress | Count, order quantity | 1 min | 10% behind time used | Supervisor |
| Finish estimate | Recent rate, due time | 5 min | Later than due | Planner |
| Quality holds | Check results | On change | Any open hold | Quality |
| Material cover | Order, stock at line | 15 min | Under 30 min left | Logistics |
| OEE and factors | Time, counts, ideal rate | 15 min | Reviewed daily | Supervisor |
If nobody can name the owner of a tile, remove it. A tile that asks nothing of anyone is decoration.
The finish estimate most dashboards get wrong
Many screens divide the remaining quantity by the ideal rate. That hides trouble until it is too late to act.
Use the good rate from the last 30-60 minutes, leaving out a long stop that has been fixed. Show the answer as a clock time, since people act on "13:21" faster than on "85%".
What the planner does when that tile turns red
A late estimate is only useful if it leads to a choice. The same numbers tell the planner how hard the choice is.
Example: to finish by 13:00 from 11:20, the line has 100 minutes for 4,600 units, so it needs 46 good a minute. It is making 38, and the ideal rate is 50.
- Can the line catch up? It would need 92% of ideal rate (46 / 50) for the rest of the order, which it has not done all morning.
- Can the due time move? A 21-minute slip may not matter if the truck leaves at 15:00.
- What is next in the queue? The order after this one starts 21 minutes late too, so warn that line's crew now.
Decide at 11:20, with 100 minutes left to act. The same news at 12:55 leaves you only the phone call.
Compare progress with time used
A progress bar alone says little, because 62% done is fine at 11:00 and bad at 12:30. Put time used right beside it.
That single pair of bars replaces a lot of phone calls. For hour-level detail use an hour-by-hour production board on the line.
One data set, three screens
- Line board: state, this order, count against the hour. Readable from a few metres, with large type.
- Supervisor screen: all lines, top stops, holds and orders at risk.
- Planner and manager view: finish estimates for today, and KPI trends for the week.
Build them on one data set, so the numbers on each screen agree. Two sources mean two arguments.
Show when the data is old
A dashboard that froze looks exactly like a line that is running fine. Print the time of the last update on each screen, and turn it red if it passes a limit you choose.
Show unknown time as unknown as well. Filling a gap with "running" flatters the shift, and unknown downtime explains how to treat it.
The test for each tile
Stand in front of the screen with a supervisor, point at each tile and ask what they would do if it turned red. If the answer is "nothing" or "it depends", redesign it.
Then check it on the floor. Can they read it from where they stand, on a bright day, with gloves on if it is a touch screen?
What spoils a dashboard
- OEE as the only big number. A line can have a fine OEE and still miss the order.
- Averages across lines. They hide the one line in trouble, and OEE averages need weighting anyway.
- Colours with no agreed limits. Red should mean the same thing to everyone.
- Too many tiles. Six on a line screen is already plenty.
- Stops without causes. The screen can show a stop, but a person must supply the reason, so make logging fast.
Built in, or built on top
Most MES products ship with screens. If yours does not fit, you can chart its data in another tool, as long as it exposes orders, events and counts.
Grafana suits live time-based views, and this Grafana guide shows how. Power BI suits daily and weekly reporting, covered in OEE dashboard in Power BI.
The state and OEE tiles need live data with a cause attached. Fabrico is a light MES for OEE that 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 shows OEE per line and shift with causes tagged at the line. See how Fabrico shows OEE per line.
What people ask next
What should an MES dashboard include?
Line state with the current stop, order progress against time used, a finish estimate, open quality holds, material cover, and OEE with its factors. Add only what someone will act on. Shop floor KPIs helps you choose.
What is the difference between an MES dashboard and an OEE dashboard?
An OEE dashboard scores how well the equipment ran. An MES dashboard adds orders, due times, holds and material, so it also says what will ship. OEE dashboard examples shows that side.
How often should an MES dashboard refresh?
State needs seconds, order progress about a minute, and OEE can update every 15 minutes or so. Faster refresh on slow numbers only makes them jump around. Print the last update time on the screen.
Can I build an MES dashboard in Grafana or Power BI?
Yes, if your system exposes the data, and the chart tool then only draws it. The orders, events and rules still have to live somewhere. See Grafana and Power BI guides.
How do I calculate order finish time on a dashboard?
Divide the remaining good quantity by the good rate of the last 30-60 minutes, then add it to the current time. Avoid the ideal rate, as it flatters the estimate. Plan attainment covers the daily view of the same question.
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.