An Elegant Puzzle

by Will Larson

An Elegant Puzzle
My Impression
★ ★ ★ ★ ☆ 4/5

Some things are valuable, some are obvious, and some are simplified too far to hold up in an actual organization.


Notes

Strategy Has Three Parts

Larson borrows Rumelt’s kernel from Good Strategy, Bad Strategy. A strategy describes the problem and how to solve it:

  • Diagnosis: a claim about what is actually going on. It names the one or two things that matter, discards the rest, and can be wrong.
  • Guiding policy: the approach and its tradeoffs. A policy is a constraint that says what we will not do. No visible tradeoff means it’s an aspiration.
  • Coherent actions: concrete steps that reinforce each other. Following from the policy individually isn’t enough.

His addition: write strategy bottom-up. Write down how you actually decided the last few similar cases, then extract the policy that was already operating.

Vision Dissolves Tradeoffs

A vision describes the future in which the strategy succeeded, specifically one where today’s tradeoffs are no longer mutually exclusive. A vision that doesn’t dissolve a real tradeoff is just optimism.

Its parts:

  • Vision statement: one or two sentences, specific enough that someone could disagree
  • Value proposition: the value the org creates for users and the company
  • Capabilities: what the org will be able to do. The gap to today is the strategy backlog.
  • Solved constraints: the limits binding us today that will be gone
  • Future constraints: what becomes scarce next. Naming it keeps the vision honest.
  • Reference material: the data the vision rests on
  • Narrative: prose memorable enough that people repeat it without the document

Anatomy of a Goal

A goal has four components: a target (where we want to be), a baseline (where we are), a trend (current velocity), and a time frame.

Goals come in two types. Investments increase a metric, baselines keep one steady. Good goals pair them: push one number while holding a controlling one in check.

They also live at two altitudes. A directional goal sits above the actions and derives from diagnosis and policy. Missing it after doing the actions refutes the diagnosis. An execution goal sits below a single action and instruments it. Missing it refutes that action only, so you pick another one under the same policy.