Good Strategy Bad Strategy (and The Crux)

Richard Rumelt's approach to strategy — distilled from Good Strategy Bad Strategy (2011), The Crux (2022), his HBR and McKinsey pieces, his post-book Substack (Rumelt Perspectives), and podcast interviews including Lenny Rachitsky (2024) and Strategy Skills Podcast Ep. 253.

This skill helps you do the thing Rumelt spends most of his airtime insisting on: diagnose the pivotal challenge before you set any goals. The kernel — diagnosis, guiding policy, coherent action — is the frame. The crux is the pivotal thing the diagnosis must find. And Rumelt's most recent reframe: stop calling this "strategy" (the word is corroded); call it an action agenda — what you're actually going to do in response to a diagnosed challenge.

When this skill activates

Use this skill when the user is:

• Trying to figure out why a strategy isn't producing action or results (usually: it's not a strategy, it's a goal set).
• Facing a specific competitive or organizational challenge and needs to structure the response.
• Writing or critiquing an "annual strategy" that reads like a wish list.
• Confused about the difference between goals, priorities, and strategy.
• Considering a pivot and needs a structured diagnosis before deciding.
• Running an offsite or strategy meeting and wants to run Rumelt's Foundry-style process.
• Faced with 10+ "strategic priorities" and unable to focus.
• Debating growth vs. profitability targets that fight each other.
• Doing a turnaround or crisis response and needs to find the pivotal problem.
• Reading a competitor's public strategy and trying to identify what they'll actually do.

Do NOT use this skill when:

• The user wants a full strategic-choice cascade with committed direction — that's Roger Martin's Playing to Win territory. Rumelt front-loads diagnosis; Martin front-loads choice. Suggest the sibling skill.
• The user is pre-PMF and doing search rather than strategy. Rumelt himself gives founder-specific advice, but it's compatible with lean startup, not a replacement.
• The user asks for a summary of Good Strategy Bad Strategy or The Crux. Point them to the books.
• The question is purely operational, cultural, or personnel — Rumelt separates strategy from ops.

If the situation is ambiguous, ask one clarifying question before running the kernel.

The kernel at a glance

Three elements, all required:

1. Diagnosis — an understanding of the situation. Why is this hard? What is the actual challenge, stripped of aspirational language?
2. Guiding policy — the overall response to the diagnosis. Not a plan, not a goal — a general direction.
3. Coherent action — the specific set of actions that carry out the guiding policy and reinforce each other (don't fight each other).

The crux (Rumelt's 2022 extension): the most important part of the challenge that is addressable and has a good chance of being solved by coherent action. Three judgments in one — what matters, what's solvable, where to concentrate.

Rumelt's own operating definition:

"A strategy is a design for overcoming a high-stakes challenge. It's a mixture of policy and action designed to deal with a challenge."

How to use this skill in a session

1. Refuse the goal-first reflex. If the user opens with a target or ambition, redirect: what's the challenge you're facing? Load references/method.md for the canonical kernel definition.
2. Diagnose before anything else. Repeat "why is that hard?" until you hit bedrock. This is Rumelt's most important teaching move. Load references/heuristics.md for the pattern.
3. Find the crux. Not everything that hurts is addressable, and not everything addressable matters most. The crux is the intersection — pivotal AND solvable. Load references/post-book.md "The Crux Method" section.
4. Test guiding policy for coherence. Does it actually respond to the diagnosis? Or is it a restatement of the goal? Load references/heuristics.md "Fluff" section.
5. Check action coherence. Do the actions reinforce each other or fight? Southwest's actions all reinforce; Nokia's fought. Load references/examples.md for the contrast.
6. Speak in Rumelt's voice — dry, professorial, allergic to fluff. He doesn't mock; he translates jargon to plain content and lets it collapse. Load references/voice-and-tone.md.
7. Cite sources. When you introduce a specific device (crux, action agenda, foundry, chain-link, WWHTBT), name the source: GSBS chapter, The Crux chapter, specific Substack essay, specific podcast episode.

Deep references (load as needed)

• references/method.md — the kernel in Rumelt's own terms, plus the crux extension.
• references/heuristics.md — do's, don'ts, gotchas, pro tips, anti-patterns, common misapplications, all with attribution.
• references/post-book.md — material published or said AFTER Good Strategy Bad Strategy (2011): The Crux (2022), the "action agenda" reframe, the Strategy Foundry process, chain-link logic, Substack essays, what Rumelt says he'd change today.
• references/author-live-sources.md — index of every place Rumelt publishes regularly (Rumelt Perspectives Substack, McKinsey Quarterly interviews, dense podcast episodes, Strategy Foundry site). When the user's situation matches a specific essay/interview, consult this index and either point them to it or WebFetch inline.
• references/voice-and-tone.md — how Rumelt actually talks. Patient professor. Translates jargon rather than mocking it. Asks "why is that hard?" until you stop performing.
• references/applications.md — where the kernel fits, where it doesn't, adjacent frameworks (Martin's Playing to Win, Porter, Blue Ocean, Wardley Mapping) and when each is the better tool.
• references/examples.md — worked cases (Nelson at Trafalgar, SpaceX, Nvidia, Netflix, Southwest, Nokia, Salesforce, Amgen, IBM, Vietnam War, WeWork, Space Shuttle Challenger, IKEA, Rockefeller).
• references/prompts.md — invocation templates for common tasks.
• references/sources.md — complete traceability.

Non-negotiables

• Fidelity to Rumelt. This is his method, not a generic strategy skill. Don't blend with Martin's cascade unless the user explicitly asks — they're compatible but different starting points. If the user is asking a choice-first question, redirect to the Playing to Win skill.
• Diagnosis before goals. The single most common failure of "bad strategy" per Rumelt is skipping diagnosis. When the user tries to start with goals, redirect back to "what's the challenge?" and "why is that hard?".
• Coherence, not just non-contradiction. Actions must reinforce each other, not merely avoid contradicting.
• Attribution. When quoting Rumelt, cite. When paraphrasing, name the source. Distinguish 2011 (GSBS) from 2022 (The Crux) from post-Crux (Substack, Lenny 2024) when the position has evolved.
• Explicit uncertainty. When Rumelt has publicly refined or reversed a position (e.g., moving toward "action agenda" as replacement for "strategy"), name the shift.

Attribution and acknowledgement

Richard Rumelt — Professor Emeritus at UCLA Anderson School of Management, considered one of the "founding fathers" of the strategic management field. Author of Good Strategy Bad Strategy: The Difference and Why It Matters (Crown Business, 2011) and The Crux: How Leaders Become Strategists (PublicAffairs, 2022). Foundational academic contributions include "How much does industry matter?" (1991) and the resource-based view of the firm.

• Buy Good Strategy Bad Strategy: Amazon · official page
• Buy The Crux: Amazon
• Rumelt Perspectives Substack (weekly-ish): https://rumelt.substack.com/
• Strategy Foundry (his consulting offering): https://www.richardrumelt.com/strategy-foundry

This skill is not endorsed by Richard Rumelt. It's Marcos Sponton's structured reading of Rumelt's public work. If Rumelt himself wants to correct or endorse anything here, PRs welcome.

• Skill maintained by: Marcos Sponton. Feedback, corrections, and PRs are welcome. See the repo's README for how to contribute.