Case Study: Packaged goods manufacturer improves OEE by 30% in 6 months and saves $12m with data modernization

Executives and operational managers at a leading dairy packaged goods enterprise were facing a critical challenge in their production process. Despite being a key player in the industry, they had limited visibility into their Overall Equipment Effectiveness (OEE), a crucial metric for understanding and optimizing machinery performance.

The existing analytics tools provided slow reporting, lacked flexibility, and didn't provide a granular enough view (i.e. visibility on individual machine states at 1 minute intervals). This made it difficult to access and interpret vital data in a timely manner. Teams would circumvent their existing analytics tools and generate manual reports to get the data they needed. This was time consuming and the reliability of the output was always under question.

This lack of real-time detailed insights into line and machine OEE metrics limited their ability to closely manage their machine utilization leading to substandard OEE rates of around 40%.

The team realized they needed the ability to quickly drill down to individual machines in near real-time to properly assess their line performance and identify and correct issues before they had a knock-on effect on the rest of production. Recognizing the need for a more robust and responsive analytics and insights capability, they turned to BigLittleRobots for support.

To help the client achieve their objective, we started by deploying the modern data infrastructure that would support this type of flexible on-demand analytics. The first step required centralizing their data, i.e., ingesting data from diverse and previously disconnected data sources such as SAP, Zarpac, and openDoc, and consolidating them within Snowflake.

We then crafted a robust data modeling and metrics layer to achieve consistent quality, reliability, and readiness of the data for analytics purposes. Utilizing tools like dbt, we structured multiple databases to address distinct aspects of the solution's implementation. First, a staging database was established as a landing zone for raw data, allowing for extraction and loading without immediate transformations. Next, a derived layer transformed this raw data into a format aligned with business concepts, facilitating its reuse in various data products. Finally, a reporting layer was developed for end users, enabling direct data consumption and analysis.

We also introduced the concept of data mesh to ensure this design was scalable, secure, and optimized. Data mesh is an innovative architectural approach that decentralizes data ownership and management. By doing so, we enabled the client to start taking advantage of their data more effectively, fostering a more collaborative and agile data environment within their organization.

Finally, together with the client, we designed a custom data product: an interactive analytics dashboard they would use to visualize, interact with, and share key OEE metrics across the organization.

The OEE data product was quickly adopted by the CEO, executive, and operational teams, becoming a vital tool in the organization. In just a matter of months, what was once a tedious and unreliable process to obtain visibility on machine utilization turned into a near real-time source of actionable insight giving them the ability to act fast and implement changes spurring machine utilization optimization.

And, as a result, within 2 months our client started increasing their existing line OEE, and within 6 months it had rose by 30% leading to savings of $12 million just in that period.