Analytics Glossary

A real-time, unified view across planning, supply and logistics data that surfaces disruptions and exceptions early enough for planners to act — not just a dashboard, but a decision layer.

The policies, roles and controls that determine who can access, change and trust a piece of data — the difference between "a number" and "a number you can defend."

A cloud data architecture that combines the low-cost, flexible storage of a data lake with the structure and performance of a data warehouse, so raw and curated data can live in one governed platform.

An organizational and architectural approach where domain teams own and publish their own data as a product, rather than funnelling everything through one central data team.

A live, data-fed virtual model of a physical asset, process or network used to simulate scenarios and predict behaviour before committing to a real-world action.

Extract-Transform-Load (or Extract-Load-Transform) — the process of moving data from source systems into a platform where it's cleaned, reshaped and made usable for analysis.

A central, governed repository of the engineered variables ("features") that feed machine-learning models, so teams reuse the same trusted inputs instead of rebuilding them per project.

Generative AI — models that produce new text, images or other content rather than just classifying or scoring existing data; in industrial settings, most often used for copilots, summarization and document Q&A.

Large Language Model — a model trained on vast amounts of text that can understand and generate natural language, underlying most GenAI copilots and chat-style tools.

The practices and tooling that take a machine-learning model from a data scientist's notebook into reliable production use — deployment, monitoring, retraining and version control.

Overall Equipment Effectiveness — a manufacturing metric combining availability, performance and quality into one score that shows how much of an asset's full potential is actually being used.

The integration of Operational Technology (PLCs, SCADA, sensors) with Information Technology (cloud, analytics, enterprise systems) so plant-floor data flows into the same governed platform as business data.

Microsoft's business-intelligence platform for building interactive dashboards and reports on top of governed data — one of the most common ways decision-makers actually see the output of an analytics program.

Using sensor, maintenance-log and operating data to forecast when an asset is likely to fail, so maintenance happens just before failure rather than on a fixed calendar or after breakdown.

The application of connected sensors, computer vision and analytics to quality management — catching defects in-line and tracing them to root cause, instead of relying solely on end-of-line inspection.

Retrieval-Augmented Generation — a technique where an LLM's answer is grounded in specific, retrieved documents (manuals, logs, reports) rather than relying only on what it learned in training, which reduces fabricated answers.

A model's estimate of how much operating time or how many cycles an asset has left before it's likely to fail, used to plan maintenance, spares and replacement timing.

A business-friendly translation layer sitting between raw data tables and reporting tools, so "revenue" or "downtime" means the same governed thing everywhere it's used.

A model that infers a hard-to-measure variable (like a chemical property or flow rate) from other, easier-to-measure signals, standing in for an expensive or slow physical sensor or lab test.

A multi-dimensional view of procurement spend — by supplier, category, business unit and time — that's the starting point for, but not the same thing as, a tracked savings pipeline.

Predicting future values of a metric (demand, price, generation, failure risk) from its historical pattern over time, often expressed as a range of likely outcomes rather than one number.

A single, real-time data layer (often built on MQTT/Sparkplug) that gives every system on the plant floor and in the cloud a consistent, contextualized view of equipment and process data.

Value Analysis / Value Engineering — a structured method for reducing product cost or improving performance without sacrificing function, increasingly informed by data on cost drivers and design trade-offs.