Artificial Intelligence
The OEE Dashboard Problem Nobody's Software Fixes
Most OEE dashboards fail at data, not math. Why the handoffs between machine, ERP, and spreadsheet break first, and what connecting them actually looks like.
Walk onto most shop floors and ask to see the OEE dashboard, and you'll usually get a half-apology before anyone pulls it up. The number on screen rarely matches what the operator wrote on a clipboard an hour ago, and both disagree with whatever showed up in last month's report. An OEE dashboard is supposed to collapse availability, performance, and quality into one number you can trust. Most don't, not because the arithmetic is hard, but because the inputs cross three or four systems that were never built to agree with each other.
What a Working OEE Dashboard Actually Needs
A working OEE dashboard shows live availability, performance, and quality for each machine or line, ranks the downtime and loss reasons behind the score, and ties every number back to a timestamped event on the floor. When a number can't be traced to its source, don't build a decision on it.
The three factors multiply. Availability is run time against planned production time. Performance is the ideal cycle time times total count, divided by run time. Quality is good parts divided by total parts. Vendor guides often call 85% "world-class" and something near 60% a typical plant average, though the benchmark shifts depending on whose guide you're reading and what you're making. Treat any published target as a starting point for your own conversation, not as a grade you're handed.
A dashboard that fails rarely fails at this math. The real break happens one step earlier, at the handoff where a machine's cycle counts, an operator's log, and a supervisor's spreadsheet are supposed to become one row of data and instead become three.
Why the Data Behind an OEE Dashboard Falls Apart
Most plants have plenty of data; what they lack is agreement about which copy of it is correct. OEE readings live in one system, order and routing data live in the ERP, and maintenance events live in a CMMS, each kept by a different team on its own clock. Plants that have already connected their ERP and MES data still run into this, because wiring two systems together moves numbers faster. It doesn't decide which number wins once they disagree.
The same gap shows up downstream. A line that logs every micro-stop can still bury the signal that matters under ones that don't, which is how plants end up drowning in notifications nobody acts on instead of fixing root causes. A dashboard that reports quality as one percentage also hides which defect is actually worth chasing, something we've covered in more detail in our look at scrap and rework patterns.
Packaged OEE Dashboard Tools vs. Connecting What You Already Run
Most of what ranks for "OEE dashboard" is a platform pitch: install sensors or a gateway, pipe the data into a new system, and view OEE inside that system's own screens. For one line in one plant, that can be the right call, and we'll say so plainly even though it's not what we sell. A packaged tool that already reads your PLC and prices by the line is often faster and cheaper to stand up than anything built custom.
The packaged route gets harder once a second plant, an older PLC with no modern protocol, or an ERP that already owns the order and routing data enters the picture. At that point you're choosing between forcing every site onto one vendor's hardware, or building a layer that pulls from the paper logs, spreadsheets, and ERP already running on the floor. We work on the second kind of problem: wiring together the systems a plant already has instead of asking the plant to replace them.
What This Looked Like at One Manufacturer
We saw this pattern end to end at a manufacturer with revenue in the billions. Production data started on paper registers on the floor and got typed into Excel by hand. From there it became charts rebuilt from scratch in PowerPoint, and it ended with a person walking a room through slides on a call. The paper register itself wasn't where things went wrong. Every handoff after it, the typing, the rebuilding, the live explanation, was where the chain actually broke: each team did its own part correctly, but nobody owned what happened to the number in between. We've written about this case in more detail.
The problem was never the budget. Each handoff belonged to a different team, every part worked in isolation, and the failure lived only in the joins between them, which meant fixing it was nobody's actual job. What we built replaced the chain with validated entry forms, structured storage with required fields, visualization that builds itself from that data, and an AI layer that answers direct questions and flags revenue-leakage patterns as they appear. The system went live in September 2026, which is too recent to put a number on what it's changed.
Frequently Asked Questions
What's a good OEE score?
Vendor guides commonly cite 85% as "world-class" and something near 60% as a typical plant average, but the right target depends on your equipment, your product mix, and how strictly you count planned downtime. Build your own trend line before you borrow someone else's benchmark.
How do you calculate OEE?
Multiply three ratios together. Availability is run time divided by planned production time. Performance is ideal cycle time times total count, divided by run time. Quality is good count divided by total count. The result is a single percentage.
Can you build an OEE dashboard without new hardware?
Often, yes, if your machines already report counts or states to a PLC or controller. The harder part is usually getting that signal into the same place as your order and maintenance data, not capturing the signal itself.
Does an OEE dashboard replace an ERP or MES?
No. An OEE dashboard reports what happened on the floor. Your ERP still owns orders and routings, and your MES, where you have one, still owns work execution. The dashboard only earns its place if it can pull from both without becoming a third system to maintain. Our note on connecting ERP and MES data directly covers where that usually goes wrong.
Is a packaged OEE tool enough for a single plant?
Often, yes, for one line in one plant running equipment the tool already supports. It gets harder once you add a second site, older equipment, or systems that already hold the order and maintenance data your dashboard needs. An audit of what you're already running is a reasonable first step either way.