Industrial IoT platform funded as overhead
A manufacturer runs 12 plants and has deployed 4,000 sensors. A predictive-maintenance pilot at one plant cut unplanned downtime by 18%. The platform team of eight engineers reports to IT operations and is funded as overhead. Its backlog holds 140 requests from plants, and prioritization is first-come, first-served. Plants have built shadow spreadsheets and local dashboards because central reports arrive 24 hours late. OT engineers distrust IT and insist that safety-critical alerts must run locally, even during network outages. Corporate leadership wants the platform scaled to all 12 plants in nine months and has asked for a plan. The platform team says it cannot support more than three plants without doubling headcount. No service catalog, SLOs, or chargeback model exists. Data is copied nightly to a central warehouse; plant managers say they cannot see current production status. The pilot's success has not been codified into reusable services.
1. Why has the platform team failed to scale beyond the pilot, even though the pilot produced an 18% downtime reduction? Focus on operating model, not technology.
Read the worked answer
The pilot was a project success, not a product capability. The team is funded as overhead, reports to IT operations, and lacks a product manager, service catalog, and SLOs. Demand arrives as 140 first-come, first-served requests, so capacity is allocated by arrival, not value. Plants build shadow dashboards because the platform does not meet their need for current data; that is a product gap, not user resistance. OT distrust and local safety requirements are legitimate constraints the central team has not designed for. The 18% result is not reusable because the work was not codified into self-service onboarding, reference designs, or supported services. Doubling headcount would add capacity to a broken demand and funding model. The binding constraint is the operating model: no product ownership, no service management, and no funding link to plant outcomes.
2. Recommend a funding and organizational model to scale to all plants. Specify what you would change first and what you would not centralize.
Read the worked answer
Change funding from overhead to an investment with showback or chargeback tied to plant outcomes, and add a platform product manager. Create a platform team with a service catalog, SLOs, and a repeatable onboarding playbook. Pair it with stream-aligned plant teams that own local outcomes and adoption. Do not centralize safety-critical alerting: keep edge processing for alerts, with cloud for analytics and history. Do not centralize plant-specific dashboards; expose APIs and self-service so plants can build views without shadow spreadsheets. First 90 days: onboard two more plants in waves, publish latency and availability SLOs, and replace first-come intake with value-based prioritization. Form a federated governance group with OT and IT. Measure time to onboard a plant, adoption of platform services, unplanned downtime, and reduction in shadow IT. Re-baseline the nine-month target after two waves instead of promising all 12 plants now.