Real-time production analytics replacing end-of-month surprises

The VP of Operations described it bluntly: "We run a $52 million operation and we find out how we did last month three weeks after it's over." Production data lived in a custom MES system that nobody on the finance team could access. Cost data came from SAP Business One, but it was allocated at the facility level rather than the production line level. Quality metrics were tracked in Excel spreadsheets maintained by shift supervisors.

The result was that leadership operated on lagging indicators. By the time a yield issue showed up in the monthly report, it had been running for weeks. Waste was measured in aggregate, making it impossible to identify which production lines, shifts, or products were driving the problem. Cost variances were discovered during monthly close, when it was too late to course-correct.

The company had tried to solve this before. A previous BI vendor had built dashboards on top of SAP, but they only showed financial data — they couldn't integrate the MES or quality data that production managers actually needed.

We began by connecting the three core systems — SAP Business One, the custom MES, and the Excel-based quality tracking — into a unified data model on Microsoft Fabric. The MES integration was the most complex piece, requiring a custom connector to extract production run data, machine utilization, and downtime logs.

The key innovation was building a production cost model that allocated costs at the production line level rather than the facility level. This required mapping the bill of materials from SAP to the production run data from the MES, then layering in labor allocation from Kronos based on shift assignments. For the first time, the company could see the true cost and margin of each product line at each facility.

We delivered the first working dashboard — a facility-level production overview — in three weeks. Over the following five weeks, we added line-level detail, shift-level quality metrics, and anomaly detection that automatically flagged yield drops exceeding historical variance.

The impact was immediate and measurable. Daily visibility into production metrics enabled operators and managers to catch and correct issues in real time rather than discovering them weeks later.

In the first year, waste reduction across the three facilities totaled $620K — driven primarily by faster identification of yield drops, material quality issues, and machine calibration drift. Yield improved by 14% on average across the three facilities, with the largest gains at the facility that had previously been measured least rigorously.

Beyond the direct financial impact, the cultural shift was significant. Shift supervisors who had been maintaining Excel spreadsheets adopted the dashboards within weeks. The monthly operations review meeting, which had previously been a three-hour exercise in reconciling conflicting numbers, now focuses on forward-looking decisions because everyone starts from the same data.