Richard P. Rumelt

Performance evidence came before the practitioner framework

Rumelt's professional work spans the 1960s through the 2020s. He received electrical-engineering degrees from the University of California, Berkeley, in 1963 and 1965, worked as a systems engineer at the Jet Propulsion Laboratory, and completed a Harvard Business School doctorate in 1972. He joined UCLA in 1976, taught at INSEAD while on leave from 1993 through 1996, and is listed by UCLA as professor emeritus.¹

Strategy, Structure, and Economic Performance, published in 1974 from his doctoral research, classified corporate diversification and organizational structure and compared those categories with financial performance. Its record describes a historical sample of 500 large U.S. firms; the book examined change over a period ending in 1969. The design established associations within a particular population and period, not a randomized test showing that a diversification category caused later returns.²

Subsequent evidence both extended and bounded that result. A 2000 meta-analysis of 55 studies found that moderate diversification was associated with higher performance than limited or extensive diversification, while also finding that results changed materially with the operational definitions of diversification and performance. Kathryn Harrigan's 2022 historical reappraisal argues that later evidence on conglomerates, vertical integration, and mature industries requires revisiting expectations drawn from Rumelt's original period.³

Business effects were large, but industry did not disappear

Rumelt's 1991 “How Much Does Industry Matter?” partitioned variation in returns among Federal Trade Commission Line of Business reporting units into time, industry, corporate-parent, and business-specific components. The study's abstract reports negligible corporate effects, small stable industry effects, and very large stable business-unit effects. Those are estimates from one U.S. dataset and model, not proof that industry conditions or corporate ownership are unimportant for every firm.⁴

Bart Vanneste's 2017 meta-analysis found the same broad ordering while revising the magnitudes. Across 18 samples from 16 studies with 225,183 observations, the variance effect sizes were 0.08 for industry, 0.14 for corporate parent, and 0.36 for business. The paper also shows that results depend on whether studies use sums of squares, variance, or standard deviation and cautions that effect size is not identical to practical importance. Many samples reused overlapping U.S. databases, further limiting independence.⁵

The defensible conclusion is narrower than “industry does not matter.” Business-specific differences explain more performance variation in this research stream, while industry and corporate effects remain detectable and measurement choices change their apparent size. Variance decomposition also does not identify which resource, decision, or management practice produced a business effect.⁵

The kernel linked a challenge to coordinated action

Good Strategy/Bad Strategy, published in 2011, is Rumelt's participant synthesis for practitioners. Its kernel has three elements: a diagnosis that identifies critical features of a challenge, a guiding policy that states an overall approach, and coherent actions designed to carry out that policy. Rumelt's author-written article adapted from the book establishes the framework; it does not test whether organizations using the kernel outperform otherwise comparable organizations.⁶

Rumelt's four hallmarks of bad strategy are failure to face the challenge, mistaking goals for strategy, bad strategic objectives, and “fluff.” His point is not that goals, budgets, vision, or values have no use. It is that they do not by themselves explain a path through an obstacle, and an undifferentiated list can avoid the choices required by limited resources.⁷

The book then presents sources of strategic power, including leverage, proximate objectives, chain-link systems, design, focus, growth, advantage, dynamics, and inertia. These are proposed ways to reason about concentration and coordination, not measured treatment components. Rumelt later calls strategy a hypothesis whose logic should be tested rather than accepted as deduction; that claim makes revision integral to the framework without supplying an outcome study of the framework itself.⁸

Rumelt also separates strategy work from recurring planning. In a 2007 interview, he characterizes rolling plans as resource budgets and argues that strategy work is episodic and opportunity-driven. He describes strategic judgments as speculative positions under uncertainty and says the dynamics of industry change were underresearched. These are his participant judgments from research and advising, not survey evidence about all planning processes.⁹

The crux further narrowed attention

The Crux, published in 2022, defines the crux as the most important part of a set of challenges that is addressable through coherent action. Rumelt separates three judgments: which issues are important, how difficult they are, and where resources can be focused without trying to do everything at once. The authorized publisher excerpt describes this as choosing the path that promises the greatest achievable progress.¹⁰

The same introduction states that strategy is not magic: some situations are irretrievable, organizations cannot change direction instantly, and political interests can lock a problem beyond available executive power. It also tells strategists to consider organizational health and balance purposes, values, and beliefs held by stakeholders. These qualifications narrow the claim; they do not provide a procedure for resolving conflict among stakeholders or measuring whose welfare a strategy advances.¹¹

Strategic coherence does not settle legitimacy

Brian Harney's independent academic review welcomes the book's problem-driven approach but calls the good/bad contrast potentially simplistic, finds the sources-of-power section itself insufficiently focused, and asks “good strategy for whom?” His answer is that top management and the corporation dominate the book's account. The review is critical commentary, not an empirical study, but it identifies a real boundary in who receives standing to diagnose and choose.¹²

“Good” in the kernel primarily evaluates the connection among diagnosis, guiding policy, and action. The later stakeholder language in The Crux makes the account less exclusively managerial, yet it still leaves the strategist to judge which purposes and constraints govern. Strategic coherence can serve a public benefit, an extractive campaign, or a mixture of both; the framework alone does not establish moral legitimacy or net effects.¹²¹¹

The cited research does not include a comparative or causal evaluation of teams trained in the kernel or crux, nor a general estimate of effects on workers, communities, public institutions, or ecosystems. The historical performance studies concern accounting outcomes and organizational categories; the practitioner books use cases and argument. No worker-controlled, community-controlled, or ecological-impact source is included here. Those are limits on the claims that can be made, not findings of zero effect.⁵⁶¹⁰

Connections are editorial reading coordinates

The work link Good Strategy/Bad Strategy identifies Rumelt's participant work. The related links organizational intelligence and benefit for all life are comparative lenses, not terms attributed to him. The idea links are likewise reading coordinates: strategy and adaptation, purpose and legitimacy, decision and judgment, measurement and control, innovation and renewal, and executive attention and sensing. The emphasis scores enable comparison; they are not measurements or claims of historical influence.

This record declares no structured causal impacts, typed relations, or reading dependencies. A disciplined use of Rumelt's method therefore begins with a provisional diagnosis, records the evidence and exclusions behind it, states the guiding policy, and checks whether the proposed actions actually reinforce one another. It must separately decide who can challenge the diagnosis, whose interests count, what would trigger revision, and which human or ecological constraints are not obstacles to be overcome.