A production dashboard is a single screen that shows how the floor is running right now and points to what needs attention. Build it so every tile answers one glance-and-act question, value, target, trend, status, instead of cramming in every metric you happen to collect.
Most plant dashboards fail the same way: they show everything and say nothing. Thirty numbers, no hierarchy, no targets, and nobody changes what they do after looking. A good dashboard is the opposite, a short list of questions a supervisor must answer this shift, each with the context to answer it in a glance. This post walks through designing one tile at a time, laying out the screen, and keeping it honest with live data.
What is a production dashboard for?
A production dashboard exists to drive action, not to archive data. Its job is to surface the handful of things a supervisor should react to this shift, a line behind schedule, a machine bleeding downtime, a quality slip, fast enough to still do something about them. If a number does not change a decision, it does not belong on the screen.
That is the line between a dashboard and a report. A production report documents what happened for the record; a dashboard shows what is happening so you can steer. Both matter, but they are different tools with different rules, and trying to make one screen serve both jobs is how dashboards end up unreadable. A dashboard that reads like an end-of-month report is too dense to act on; a report that only shows live tiles cannot answer "what happened last Tuesday." Build the dashboard for the shift and let manufacturing analytics handle the deep look-back.
Why does one-glance-and-act beat cramming in every metric?
Because attention is the scarce resource on a shop floor, not data. A screen with five tiles that each demand a decision gets used; a screen with thirty tiles gets ignored, because the eye cannot find the one thing that matters. Every metric you add makes the important ones harder to see.
The test for any tile is simple: what will someone do differently when this number goes the wrong way? If you cannot name the action, cut the tile. Vanity metrics, totals that only go up, numbers with no target, counts nobody owns, crowd out the few that drive behavior. A tight dashboard of real manufacturing KPIs beats a wall of numbers every time.
What makes a good dashboard tile?
A good tile carries four things: the current value, the target it is measured against, the trend direction, and a status color. Strip any one and the tile stops answering its question. A bare number tells you nothing without a target; a target tells you nothing without the trend that shows where it is heading.
Keep the color semantics ruthless and consistent, green good, yellow watch, red act, and use them sparingly so red actually means something. A dashboard where half the tiles glow red has trained everyone to ignore red. Borrow the discipline of an andon system: a signal only works if it is rare enough to demand a response. The same restraint that makes visual management work on the floor makes a dashboard work on a screen.
Which metrics belong on a production dashboard?
Pick the few metrics that tie to what this audience controls this shift. For a line supervisor that usually means output against plan, OEE or its three parts, downtime by reason, and a quality or scrap signal. That is four to six tiles, not twenty. Everything else lives one click deeper.
| Tile | Question it answers | Action it triggers |
|---|---|---|
| Output vs. plan | Are we on pace for the shift target? | Add resources, expedite, or flag the gap |
| OEE (or A / P / Q) | How well is the equipment converting time to good units? | Attack the weakest of availability, performance, quality |
| Downtime by reason | What is stopping the line most right now? | Dispatch maintenance to the top reason code |
| Scrap / first-pass yield | Are we making good product or making rework? | Stop and correct before more scrap piles up |
| Constraint status | Is the bottleneck fed and running? | Protect and expedite to the constraint |
Layer machine monitoring data underneath and the tiles update themselves instead of waiting on a clipboard.
What does a standard say about dashboard metrics?
Standards can settle the argument over what to measure and how to define it. ISO 22400-2:2014 the international standard for key performance indicators in manufacturing operations management, defines 34 KPIs, including OEE, availability, and throughput rate, with a formula and a named user group for each. Building a dashboard on those definitions means a tile labeled "OEE" means the same thing on line 1 as on line 6, and the same thing to the operator as to the plant manager.
The standard also tags each KPI with who uses it, operator, supervisor, or manager, which is a useful filter when you are deciding what goes on whose screen. That maps cleanly onto the one-audience-per-dashboard rule: pick the KPIs the standard assigns to your audience, define them the standard's way, and your numbers travel across lines, shifts, and plants without translation. The same definitions feed forward into your capacity planning metrics so the rate on the dashboard and the rate in the plan are one number, not two.
How should you lay out the screen?
Lay it out in tiers, top to bottom, from headline to detail. Put the two or three numbers that decide whether the shift is winning at the top, where the eye lands first. Put trends and exceptions in the middle. Put drill-downs and reason breakdowns at the bottom, for the person who needs to dig after the top tiles raised a flag.
- Pick one audience per screen. A supervisor's dashboard and a plant manager's dashboard are different screens. Designing one board for everyone pleases no one and doubles the tiles.
- Name the top question. Decide the single most important thing this audience must know, usually output against plan, and give it the top-left, biggest tile.
- Add only tiles with an action. For each candidate metric, write the action it triggers. No action, no tile.
- Give every tile a target. A number without a target cannot be judged at a glance. Set the target from real capability, not a wish.
- Set consistent color rules. Green, yellow, red mean the same thing on every tile, and red is reserved for "act now."
- Layer the detail underneath. Keep the top screen sparse; put the breakdowns one click or one tier down for whoever needs to investigate.
- Test the glance. Show it to someone for a few seconds and ask what they would do. If they cannot answer, it is too busy, cut tiles until they can.
What are the common dashboard mistakes?
The classic mistake is the everything-dashboard: every metric anyone ever asked for, none of them with a target, all the same size. It looks thorough and drives nothing. Close behind are stale data, inconsistent colors, and tiles nobody owns.
- No targets. Numbers with nothing to compare against cannot be judged in a glance. Give every tile a target line.
- Stale data. A dashboard fed by yesterday's spreadsheet is a report wearing a dashboard's clothes. If it is not close to live, do not call it a dashboard.
- Color inflation. Red on half the tiles means red on none of them. Reserve it for act-now.
- No owner per tile. A metric nobody is accountable for is decoration. Name who acts on each one.
- One board for all audiences. Cramming the operator, supervisor, and plant-manager views onto one screen guarantees clutter. Split them.
How do you keep the dashboard honest and current?
A dashboard is only as trustworthy as its freshest number. If the floor sees one figure on the live board and the morning meeting shows another from a spreadsheet, people stop believing both. The fix is to feed the dashboard from data captured where the work happens, in real time, so there is one number everyone shares.
That is the case for capturing production counts and downtime reasons on the floor as they occur, rather than reconstructing them at shift end. Plants like CLS replaced paper logs with real-time capture and automated daily reporting, so the tile a supervisor reacts to overnight is the same number in the morning review, no reconciliation, no argument. Related boards like a maintenance KPI dashboard or a bowling chart for month-over-month targets sit alongside the live production view, each built on the same trusted data. Want a first metric to anchor the board? Run your line through a free OEE calculator and put that number, with a target, in the top-left tile.