The shadow spreadsheet is telling you something

Walk any plant with a modern MES and you'll find a spreadsheet running alongside it. Sometimes three. Someone on second shift maintains a scrap log in Excel because the MES scrap codes don't match how the quality team actually thinks about defects. The scheduler keeps a whiteboard because the planning module can't handle the family-mold changeover rules. The maintenance lead runs a personal parts list because the CMMS parts catalog was set up by someone who left in 2019.

Those artifacts aren't a failure of discipline. They're a signal. The purchased system doesn't fit the work, so the work routed around it.

The workaround tax

Every shadow spreadsheet costs you something specific. Double entry eats operator time and introduces reconciliation errors. Reports built off the MES understate scrap because the real scrap lives in Excel. First-pass yield looks better than it is. Root-cause analysis takes a week because three people have to align three versions of the same shift.

The SaaS-bloat conversation usually frames this as a licensing problem. It's not, mostly. Vertice's write-up on lean tech stacks makes the point that overlapping tools accumulate through renewals nobody audits, and that's true. But on a plant floor the bigger cost isn't the seat fee. It's the labor and the decisions made from bad data. Procurement Magazine's summary of the Forbes SaaS bloat piece gets closer: teams end up adapting their workflow to vendor assumptions, and the adaptation itself is the expense.

What the spreadsheet is actually telling you

In our experience, the shadow spreadsheet almost always encodes one of three things:

1. A real business rule the vendor's data model can't express. Family molds, co-products, rework loops that re-enter the line three operations back. Standard software assumes a cleaner graph than most shops have. 2. A vocabulary mismatch. The quality team calls it a short shot, the MES calls it a reject code 47, and the customer complaint field wants a free-text description. Somebody has to translate, and translation lives in Excel. 3. A trust gap. The number the system produces doesn't match what the shift lead saw with their own eyes, so they keep their own tally. This one is dangerous because it means nobody upstream is using the system's numbers to make decisions either.

What to do before you buy anything else

Before adding a module, a platform, or an AI layer, spend a week collecting the shadow artifacts. Every spreadsheet, every whiteboard photo, every text thread where a supervisor asks "did we run the 40-ton at all yesterday." That pile is your real requirements document. It's more honest than any RFP.

Then you have a choice. Bend the workflow to fit the tool, retrain everyone, and enforce it — which sometimes works, when the tool is genuinely better than the habit. Or build a thin layer that matches how the work actually happens and hands clean data back to the systems that need it.

We lean toward the second, obviously. But the point isn't which you pick. The point is that the spreadsheet is not the problem. It's the diagnosis.