Prof. Warren Powell - Better Decisions

Prof. Warren Powell is a Princeton emeritus professor and a pioneer of modern decision science, best known for translating decision theory into large-scale, real-world systems across logistics, energy, and transportation. He founded Optimal Dynamics and has spent four decades studying what actually works or fails when organizations make high-stakes decisions under uncertainty.

A decision is information you control, not outcomes or forecasts. Organizations often fail because they optimize results without clearly defining what choices they actually control. Most impactful decisions are indirect: leaders set metrics, constraints, and evaluation rules that guide behavior throughout the system. Metrics are not just measures, they are levers of control.

Decisions and uncertainty come in multiple forms. Powell identifies six types of decisions — including resource allocation, discrete actions, experimentation, communication, function design, and parameter tuning — and emphasizes that uncertainty must first be articulated in plain language. What is not named cannot be managed. Risk is defined by outcomes outside your primary metrics, not by ordinary variability.

Simplicity plus tuning often outperforms complexity. Deterministic models, heuristics, and buffers are practical because they are adaptable and interpretable. All decision policies fall into four structural forms: rules or heuristics, local optimization, approximate future valuation, or full lookahead. The key insight: better-framed decisions beat better models.

Source: https://castle.princeton.edu/makingdecisions/

87% of readers share our post with someone who will benefit from it.

Subscribe here to to be first to know when the next episode drops: https://www.youtube.com/@SLASOG

For more of my thoughts, follow me on LinkedIn.

Get my book Data Impact for a pragmatic take on data-driven value creation for business.

Thanks for reading SLASOG: Leaders are Readers! Subscribe for free to receive new posts and support my work.