The Data Paradox: Why Having KPIs Isn't the Same as Having Visibility
Years of dashboards, and the same question still unanswered: where is the margin actually leaking?
Martín Suárez Yumar · Tecno-Fab
8/26/20264 min read
Almost no mid-sized manufacturing company starts from zero on digitalization today. Most already have an ERP, some kind of production dashboard, quality indicators, maybe a partial MES. The line we hear most often in a first conversation with a general manager isn't "we don't have data", it's closer to:
"We have data. We have KPIs. And I still don't know where the margin is slipping away."
That sentence captures what we call the data paradox: the more technology accumulates without an operational criterion to organize it, the more noise builds up around the same blind spot.
What's usually available vs. what's still missing
In most of the plants we've analyzed, a reasonable layer of instrumentation already exists: production panels, quality reports, some ERP with a manufacturing module, tracking spreadsheets.
What almost never exists is the next layer, the one that turns that data into an answer to very specific questions:
Which products actually generate margin, and which ones destroy it without anyone having calculated it?
What specific constraint is limiting growth, capacity, raw material, labor, sequencing?
Where exactly is productive capacity being lost: in the changeover, in the recurring micro-stop, in the bottleneck no one has identified?
Which investment should be prioritized first, and with what evidence does that priority hold up?
When these questions have no answer, decisions don't stop being made — they're made anyway, on assumptions. And capital gets allocated accordingly: sometimes well, sometimes not, with no way to know until the outcome is already locked in.
The iceberg no dashboard shows
The indicators that do get reported, OEE, plan compliance, standard cost, are the visible tip. Below the surface, uncaptured by any metric, there are usually three layers:
Hidden operational constraints. The real bottleneck is almost never the one flagged in the ERP. It's the one everyone in the plant already knows by heart, but that no one has translated into a defensible number.
Invisible margin erosion. Standard cost says one thing; real cost, with scrap, absorbed downtime, misallocated overhead, says another. The gap between the two is rarely measured systematically.
Suboptimal decisions. Without the two layers above resolved, mix, pricing, or capacity investment decisions get made with the best of intentions, but without the data that would actually support them.
The practical consequence: competitive advantage is no longer won by having more data. It's won by understanding the operation better with the data that already exists.
An illustrative case - not attributed to any real client
To keep this from staying abstract, here's an example built from patterns that recur frequently in plants with €15–50M in revenue (illustrative figures, not corresponding to any specific company):
The reported OEE wasn't the real OEE. The dashboard showed 74% efficiency. On reviewing how it was calculated, the model used an incorrect reference speed and didn't capture micro-stops below a certain threshold. Real OEE was between 61% and 65%. The gap wasn't a data problem, it was a problem of what had been chosen to measure, and how.
Standard cost was hiding references that were destroying margin. Three products considered profitable in the monthly reporting turned out, once recalculated against actual production data (not theoretical), to be selling below cost on certain small batches. No one had seen it because the standard hadn't been checked against plant reality in years.
A deviation took weeks to detect, when the infrastructure to detect it in minutes already existed. The problem wasn't technological, the systems were already in place. The problem was that the data those systems captured wasn't connected to the decision it was supposed to support.
In all three cases, the pattern is the same: the technology was already there. What was missing was the operational judgment to ask it the right question.
Why this doesn't get solved by adding another dashboard
The natural temptation, on noticing this gap, is to buy one more tool: a BI module, a new panel, an extra layer of advanced analytics. In most cases, this doesn't close the gap — it relocates it.
A new dashboard sitting on top of a poorly defined process, with unclear ownership and no routine for responding to a deviation, produces the same result as the last one: more visibility into a problem that still isn't being managed.
The sequence that actually works runs opposite to instinct:
Understand the real operation, not the one in the manual, the one that happens on the night shift.
Identify where value is being lost, with evidence, not intuition.
Quantify that impact in euros, not in loose percentages.
Only then, choose the technology, if any is even needed, that solves that specific problem.
The question that actually matters
It's not how many indicators your plant has. It's what happens exactly after one of them falls out of tolerance: is there a shared interpretation, an owner, an action with a deadline, a check that the result actually held?
If that chain isn't complete, the indicator informs, but doesn't control. And the difference between informing and controlling is, almost always, where the margin you can't see is living.
Where is your operation losing value without you seeing it clearly?
Tecno-Fab's Operational Check reviews 12 indicators across six dimensions, planning, production and capacity, inventory and service, data, quality and continuous improvement, and digitalization. In about five minutes, it helps identify relationships and potential exposures worth exploring. The result is directional , it doesn't replace a plant visit or constitute a diagnosis.
Shall we talk about your operation?
© 2025. ALL RIGHTS RESERVED. · Tecno-Fab · Independent industrial digitalization consultancy · Barcelona