Management Is Prediction: The Process Control Worldview

Summary

Cedric Chin’s thread on Statistical Process Control (SPC) applied to business operations. Core thesis from W. Edwards Deming: management is prediction — to be a good operator, you need to predict (within limits) the business outcomes of your actions. Most operators run on “superstition” — they believe their actions cause certain outcomes but never verify the causal relationship. The path to operational excellence is an org-wide pursuit of “knowledge,” defined as models or theories that enable better prediction.

The Process Control Worldview

1. Management is prediction — good operators can predict the outcomes of their actions
2. Most operators run on superstition — assumed causality without verification
3. “Knowledge” = theories or models that let you predict better (Deming’s definition)
4. Becoming data-driven = the pursuit of knowledge in whatever form fits your business

Understanding Variation

SPC’s foundational insight: “Understanding variation is the beginning of knowledge.”

• When looking at business data, distinguish normal variation from exceptional variation
• Normal variation: ignore it
• Exceptional variation: investigate it — it leads to knowledge
• Exceptional variation means unknown underlying causes are affecting outcomes
• Identifying those causes lets you remove or manipulate them, leading to better prediction

Practical Tools

Amazon’s Approach

• Weekly Business Review (WBR) — forces the org to confront data regularly and improve prediction
• FCF Forecasting Tool — if free cash flow is your primary financial measure, build tools to predict it
• Amazon doesn’t use SPC charts but achieves the same outcome (org understanding of variation) through different processes

Process Behaviour Charts

SPC’s recommended tool for identifying variation patterns. Not a panacea — takes skill to use properly. The charts themselves matter less than the worldview: constantly designing processes and tools to improve prediction.

The Framework Applied

If your business is a process: figure out which input metrics you can manipulate to change the outputs. Rinse and repeat for every sub-process in every sub-system of your org.

Connections

This is the theoretical foundation for how Ray Data Co should think about operations. The “management is prediction” frame directly justifies building dashboards, tracking metrics, and running experiments — not as busywork but as the pursuit of knowledge that enables better operational prediction.

Connects to why analytics efforts fail — most analytics teams produce reports without the process control worldview, so the data never converts to operational knowledge.