Smart Factories: From Sensors to OEE in 90 Days

Overall Equipment Effectiveness (OEE) is the metric that manufacturing leaders lose sleep over. It multiplies availability, performance, and quality into a single number that tells you how much of your production capacity you are actually using. World-class OEE is 85%. Most factories operate at 50-60%, often without knowing where the losses are.

The problem is visibility. Without real-time data from the shop floor, losses are invisible — attributed to generic causes like "machine downtime" or "quality issues" rather than specific, actionable root causes. Industrial IoT changes this by instrumenting every machine, every line, and every process with sensors that stream data to analytics platforms in real time.

The OEE problem

We deploy smart factory solutions in three 30-day phases, each delivering measurable value. The first 30 days are about instrumentation — installing IoT sensors on critical machines, deploying edge gateways, and establishing data pipelines. We do not rip and replace existing equipment; we instrument it. Legacy CNC machines get retrofitted with vibration and current sensors. Manual processes get tablet-based check-in stations.

The second 30 days focus on analytics — transforming raw sensor data into OEE calculations, downtime reasons, and quality trends. This is where the invisible becomes visible. Production managers see dashboards that show exactly which machines are underperforming, why, and by how much — in real time, not in end-of-month reports.

The third 30 days introduce predictive maintenance — using machine learning models trained on the first 60 days of data to predict failures before they happen. This is where OEE moves from monitored to improved. Unplanned downtime shifts to planned maintenance, and first-pass yield climbs as quality issues are detected earlier in the process.

Key takeaways

• Instrument before you optimise — you cannot improve what you cannot measure.
• Retrofit, do not rip-and-replace — sensors work with legacy equipment.
• Start with critical machines, not the entire factory — prove value, then scale.
• Predictive maintenance needs 60 days of baseline data before models are reliable.

The 90-day phased approach

A production-grade smart factory architecture has three tiers. At the edge, sensors and PLCs feed data to industrial gateways that perform local processing — filtering, aggregation, and protocol translation. The edge tier is critical because it reduces latency for real-time alerts and ensures the factory keeps running even if cloud connectivity drops.

In the middle, a time-series database ingests high-frequency sensor data and makes it queryable for analytics. The cloud tier runs the OEE dashboards, predictive models, and reporting. This three-tier architecture balances real-time responsiveness with analytical depth — and it scales from a single line to an entire facility without re-architecting.

Architecture: edge to cloud

On a recent engagement with an industrial electronics manufacturer, this 90-day approach reduced unplanned downtime by 33% and improved first-pass yield by 9% across 22 production lines. The instrumentation phase alone uncovered that 15% of downtime was caused by a single recurring changeover issue that had been invisible in manual logs.

The lesson is consistent across every smart factory we have deployed: the biggest gains come not from the predictive models but from the visibility itself. When production managers can see real-time OEE on a dashboard, they make better decisions — faster. The technology is the enabler; the transformation is operational.