By Stephen Gnidovec, Founder, Great Lakes Analytics
I have spent the last decade in food and beverage manufacturing, eight years leading operations analytics at a large-scale global manufacturer and the last three advising mid-market manufacturers through Great Lakes Analytics.
Across the industry, the pattern is the same. Leadership approves the investment. Implementation takes a year, sometimes longer. The ERP goes live. The MES starts collecting. A visualization layer gets built on top. And then the project closes.
Except the floor never really uses it.
Supervisors keep their own spreadsheets. Managers look at the dashboard, then call someone to verify the number before making a call. Cross-functional meetings open with ten minutes of reconciliation before anyone gets to the decision. The people closest to the operation have quietly decided they don’t trust what the system is showing them.
The technology doesn’t fail. The data environment underneath it is never finished.
The operational layer your digital transformation left unfinished
The scope of a digital transformation almost always covers the systems, the integrations, the data flows, the dashboards. What doesn’t get addressed is the operational layer that makes all of it usable. Who owns a given KPI? What is the authoritative source when the MES says one thing and the ERP says another? Does the reporting cadence match the speed at which operations actually needs to move?
Those questions don’t have a line item in the implementation budget. So they don’t get answered. The result is a data environment that is technically complete and operationally broken.
Poor data quality drains millions in lost productivity
In my experience, operational teams routinely spend a significant portion of their day on manual data verification and reconciliation. Not analysis. Not decision-making. Just confirming that the number on the screen matches what they believe to be true.
The research backs this up. Gartner estimates that poor data quality costs organizations an average of $12.9 million per year, based on a survey of 154 enterprise reference customers across their 2020 Magic Quadrant for Data Quality Solutions.
And Gartner predicts that by 2027, 80% of data and analytics governance initiatives will fail, largely because organizations treat governance as a reactive, data-only exercise rather than tying it to business outcomes. That tracks with what I see on the ground. The governance layer that connects the data to the decision is the part that keeps getting skipped.
I was talking to a quality director at a food manufacturer recently who described exactly this split. Her team had done the foundational work. They validated every input, resolved measurement errors, and built a decision process on top of data they could actually stand behind. They presented monthly to executive leadership. Plants were held accountable to one version of the number, and it worked.
Every other department in the same facility had skipped that work. They assumed their data was solid. The floor had stopped trusting it. Nobody said anything because working around the data had become the default.
Same company. Same systems. One team was making decisions. The rest were still holding meetings to figure out which number to use.
Why adding more technology won’t fix a structural data problem
The instinct is always to fix this with more technology. More dashboards. More training. A bigger BI team. None of that addresses the actual problem.
The problem is structural. It sits in the space between what the system produces and what the operating team is willing to stake a decision on. If the inputs were never validated from the floor up, the metric definitions were never reconciled across departments, and nobody formally owns the KPIs, then the visualization layer is just a window into data nobody acts on.
You can build the most sophisticated dashboard in the industry. If the supervisor on the floor doesn’t trust the number behind it, the spreadsheet wins every time. And it should. That spreadsheet is the most honest signal in your operation. It tells you exactly where data trust broke down.
How to close the data gap without buying new systems
Closing this gap doesn’t require new systems. It requires finishing the work the original implementation left incomplete.
That means mapping the data environment as it actually operates today. Where are the shadow spreadsheets? Where do two systems report different numbers for the same metric? Which KPIs have a clear owner and which are floating between departments? From there, validate the inputs the team relies on, reconcile the metric definitions, assign ownership, and align the decision cadence to the speed the operation actually runs.
None of that is glamorous. But it is the difference between a digital transformation that runs and one that returns.
Stephen Gnidovec is the founder of Great Lakes Analytics, a data trust advisory firm for mid-market food and beverage manufacturers. He spent eight years leading operations analytics at a large-scale global manufacturer and brings a decade of industry experience. He is the author of The Data Culture Handbook and teaches data analytics and MBA courses at Elmhurst University and SNHU.
