The McNamara Fallacy: How People Analytics Loses the War It's Winning on the Dashboard

60-Second Summary

Robert McNamara (US Secretary of Defense, 1961–1968) ran the Vietnam War with quantitative rigour — body counts, kill ratios, weapons expended — and lost the war he could not measure (legitimacy, popular support, morale).
The McNamara Fallacy, named by sociologist Daniel Yankelovich, has four stages: (1) measure what can be measured, (2) disregard what can't, (3) assume what can't be measured isn't important, (4) claim what can't be measured doesn't exist.
Every people-analytics function moves through these stages predictably. The question is which stage you are in — not whether you are in one.
The stage-1-to-stage-2 transition is quiet and dangerous: things stop being measured and drift out of the conversation without anyone deciding they don't matter.
Counter-design: maintain a permanent 'unmeasured but important' list, protect qualitative reporting alongside quantitative, and audit annually whether decision quality has improved as measurement has expanded.

Robert McNamara arrived at the Pentagon in 1961 from the presidency of Ford Motor Company. He was the finest quantitative manager of his generation. He had rebuilt Ford's business by relentless measurement. When he took over Defense, he applied the same discipline to war. The metrics were rigorous, the reporting was disciplined, the dashboards were state of the art. He ran the Vietnam War on the numbers he could produce: body counts, weapons expended, sortie rates, hamlet pacification indices. And he lost — because the war was decided by variables he had systematically failed to measure. He acknowledged this himself, decades later, in a documentary that is required watching for anyone who runs a dashboard for a living. The pattern he embodied now bears his name, and it is the single most useful diagnostic for people-analytics functions.

McNamara in Vietnam

“We were wrong, terribly wrong. We owe it to future generations to explain why.”
— Robert McNamara, In Retrospect: The Tragedy and Lessons of Vietnam (1995)

McNamara's Pentagon produced weekly quantitative reports of extraordinary rigor. Body counts were tracked to the individual. Bomb tonnage was recorded by target category. Sortie rates were graphed by squadron. What was not tracked, because it was hard to track: Vietnamese popular support for the National Liberation Front; the legitimacy of the South Vietnamese government among rural populations; the effect of civilian casualties on recruitment for the opposing side; the difference between a village 'pacified' on the pacification index and a village actually loyal to the government. All of these turned out to be the deciding variables of the war.

In The Fog of War (2003), Errol Morris's Oscar-winning documentary, McNamara — then in his late 80s — walks through eleven lessons. Lesson 8: 'Be prepared to reexamine your reasoning.' Lesson 11: 'You can't change human nature.' Between them sits the implicit lesson that got lost: the measurement discipline that had built Ford's business had, in a different domain, systematically excluded the variables that determined the outcome. This is the fallacy that bears his name.

The four stages, in order

“The first step is to measure whatever can be easily measured. This is OK as far as it goes. The second step is to disregard that which can't be easily measured or to give it an arbitrary quantitative value. This is artificial and misleading. The third step is to presume that what can't be measured easily really isn't important. This is blindness. The fourth step is to say that what can't be easily measured really doesn't exist. This is suicide.”
— Daniel Yankelovich, sociologist, articulating the McNamara Fallacy

The four stages, applied to any measurement function

1
Stage 1 — Measure what you can
Healthy. New instrumentation, new visibility. Decisions improve. This is where every analytics function starts and where the value case gets built.
2
Stage 2 — Disregard what you can't
Quiet. Unmeasured variables stop appearing in reports. Nobody decides they don't matter — they simply stop being discussed because they can't be graphed. The transition is invisible and rarely resisted.
3
Stage 3 — Assume what you can't measure isn't important
Structural. The function's culture now treats measurability as a proxy for importance. New unmeasured variables are dismissed with 'we can't quantify that so we can't act on it'. This is where most mature analytics functions sit.
4
Stage 4 — Claim what you can't measure doesn't exist
Terminal. The function has replaced the territory with the map. Practitioners who raise unmeasured variables are treated as unscientific. The organisation cannot see what its instruments cannot see, and defends the blindness as rigour.

Mapping the fallacy onto people analytics

What each stage looks like in a modern people-analytics function

1
Stage 1 signals
Attrition tracked by team, cohort, and manager. Time-to-productive measured. Engagement scores collected and reported. Regretted vs unregretted attrition classified. Decisions improve visibly.
2
Stage 2 signals
The team no longer discusses variables that don't have dashboards. Manager quality, decision-making capacity of teams, and cultural drift are still known to matter but stop appearing in reviews. Nobody decided to drop them — they just didn't fit the reporting.
3
Stage 3 signals
New proposals require quantitative justification to be considered. 'We can't measure that' becomes 'we can't act on that'. Qualitative signals from senior operators are discounted as anecdotal. The analytics function starts to feel like a gatekeeper rather than a service.
4
Stage 4 signals
The organisation's model of itself is its dashboard. Executives cite metrics that everyone knows are noisy, because those are the metrics that exist. Practitioners who raise unmeasured concerns are met with methodological pushback rather than curiosity. The function is producing rigour and losing the war it was hired to fight.

Which stage is your analytics function in?

Ask five senior operators outside HR: 'What are the three most important variables in this org's people system?' If more than one of the three variables they name is not currently on a dashboard, you have unmeasured but important variables. That is normal. The question is what happens next.
Follow up: 'What happens when you raise those variables in a meeting where analytics is present?' If the answer is 'nothing changes because we can't measure it', you are at Stage 3.
Diagnostic question for the analytics function itself: 'What variable have you added to the reporting in the last 12 months because a senior operator said it mattered, even though it was hard to measure?' If the answer is 'none', you are at Stage 3 minimum.
Terminal signal (Stage 4): the analytics function has become the arbiter of what the organisation is allowed to discuss. Concerns without dashboards are treated as invalid. This is the point at which the function has replaced the territory with the map.

The stage-2 trap

The transition from stage 1 to stage 2 is the most dangerous because it is silent. Nobody decides that manager quality, cultural drift, or founder relationships don't matter. They simply stop appearing in reviews because they don't fit the reporting. Six months later, everyone talks about them in hallways and nobody raises them in the meetings where decisions are made. This is why the McNamara Fallacy is a systems failure, not an individual failure.

Counter-design that keeps the unmeasurable alive

Maintain a permanent 'unmeasured but important' list. Publish it alongside the dashboards. Review quarterly. Missing items are added; items measurement has caught up with are graduated to the dashboards.
Protect qualitative reporting alongside quantitative. Not 'anecdotes' — structured qualitative reporting with named senior owners. The best HRBP in each business unit reports monthly on things that don't fit the dashboards.
Annual audit: has decision quality improved as measurement has expanded? If the answer is no, more measurement is not the fix. This audit is uncomfortable and rarely done.
Reward the analyst who raises the unmeasured variable. Career consequence for the person who says 'the dashboard is missing the thing that matters'. This is culturally hard because it rewards contradiction of the function's central claim.
Publish, in every quarterly analytics review, the top three questions the function cannot currently answer with data. This forces the function to acknowledge its own limits and gives senior operators standing to bring unmeasured concerns.
Time-limit new metrics. Every new metric added to the standard reporting has a 12-month sunset unless it can be demonstrated to have driven a decision that improved an outcome. Prevents metric accumulation from becoming metric substitution.

The mature analytics function's admission

The mature people-analytics function does not claim comprehensive measurement. It claims useful measurement of a defined subset, structured qualitative reporting on the rest, and honest acknowledgement of what it cannot see. This is a smaller claim than most functions make. It is also more defensible, more useful, and less likely to be the reason the organisation loses the war it thought it was winning.

FAQ

Frequently asked questions

Isn't more measurement always better?

No. More measurement is better up to the point where measurable variables start to crowd out unmeasurable variables in decision-making. Beyond that point, more measurement produces worse decisions, not better ones, because it strengthens the fallacy.

How does this relate to Goodhart's Law?

Goodhart's Law: when a measure becomes a target, it ceases to be a good measure. McNamara Fallacy: what can't be measured drops out of consideration. Goodhart is about the corruption of measured variables; McNamara is about the invisibility of unmeasured ones. Both are properties of measurement in human systems and both compound.

Does this argue against people analytics as a function?

No. It argues for people analytics as a function with humility, honest limits, and structured protection of the unmeasured. The best people-analytics leaders read Fog of War and revise their own operating model.

Takeaways

The McNamara Fallacy has four stages. Every measurement function moves through them predictably. The question is which stage you're in.
The most dangerous transition is stage 1 to stage 2 — silent, and rarely resisted. Unmeasured variables drop out of the conversation without a decision being made.
The counter-design is a permanent 'unmeasured but important' list, protected qualitative reporting, and an annual audit of whether decision quality has improved as measurement has expanded.
The mature analytics function does not claim comprehensive measurement. It claims useful measurement, structured qualitative reporting on the rest, and honest acknowledgement of what it cannot see.
McNamara built the finest dashboards of his generation and lost the war they weren't looking at. Every people-analytics leader should re-read that sentence quarterly.