What is hard to measure may disappear from management’s view without becoming any less important.
The McNamara Fallacy describes a style of governance in which what is measurable gains increasing attention and is increasingly treated as reality, while relevant but hard-to-quantify aspects are ignored. The term comes from Daniel Yankelovich, not McNamara himself.
What is hard to measure may disappear from management’s view without becoming any less important.
Yankelovich described a progression: measure what is easily measurable; ignore what cannot easily be measured or assign it an arbitrary value; then assume that it is unimportant; and eventually that it does not exist.
Yankelovich set out the four steps in “The New Odds” (15 October 1971), pp. 4–5, explicitly connecting them to Vietnam. This is a historical warning, not an experimentally established behavioural law.
Measurability and data availability direct attention. An omitted dimension may therefore effectively disappear from the decision model and later be treated as irrelevant.
AI transfer to investigate: AI can structure text, calculate counts and summarise information. Whether relevant unmeasured dimensions are retained is a separate research question.
Proposed test, not yet conducted: Give decision-makers structured quantitative evidence alongside equally relevant qualitative evidence, but vary machine-legibility and AI summarisation. Measure attention, choice and predictive performance.
The governance question is always: what matters but has not made it into this measurement architecture?
Specify the processing AI performs and the relevant dimension omitted from measurement. The historical formulation does not validate the proposed AI transfer.
Yankelovich (1971), The New Odds — printed pp. 4–5; PDF pp. 5–6
The four-step passage appears in a shared archival scan of Yankelovich’s text. The scan’s durable preservation has not been independently established. The historical formulation is not empirical evidence for the proposed AI application.
Help improve this case
We review your contribution before updating the text or sources.