The McNamara Fallacy: Why B2B Marketing Is Winning Every Metric and Losing the War

In 1965, Robert McNamara had a problem. As US Secretary of Defense, he needed to prove that America's strategy in Vietnam was working. The war was expensive, politically fractious, and increasingly difficult to justify. He needed numbers.

So he picked the metric he could count: enemy dead.

Body counts became the currency of progress. Field commanders were pressured to report them. Analysts compiled them into charts. The numbers climbed. Briefings to the President showed steady, measurable improvement. By every metric the Pentagon tracked, the war was being won.

It wasn't. And everyone on the ground knew it.

The things that actually determined the outcome couldn't be reduced to a number. The political resolve of the North Vietnamese. The loyalty of the South Vietnamese population. The morale of American troops. Whether the strategy was achieving its stated aims. All of it was complex, qualitative, and deeply human. So it was left out of the reports. Not because McNamara believed it was unimportant, but because it couldn't fit in a column.

The philosopher Charles Handy later codified this into what he called the McNamara Fallacy, set out in four devastating steps:

1. Measure what can be easily measured.

2. Disregard what can't be easily measured.

3. Presume what can't be measured isn't important.

4. Presume what can't be measured doesn't exist.

Decades later, McNamara himself admitted the mistake in the documentary The Fog of War: "We were wrong, terribly wrong." Not because the data was inaccurate. The body counts were real. The error was that the data was irrelevant to the question it was supposed to answer.

The body counts of modern marketing

I think about those four steps every time I see a B2B marketing dashboard.

MQLs. Click-through rates. Cost per lead. Email open rates. Form fills. Webinar registrations. These are the body counts of modern marketing. They go up. The quarterly business review looks healthy. CMOs present slides showing steady, measurable improvement.

And underneath it all, a question nobody asks: does any of this mean someone actually trusts us enough to buy?

Brand salience can't be put in a cell. Neither can buyer sentiment. Nor the moment your company gets mentioned unprompted in a planning meeting you'll never see. The one where a mid-level manager says, "What about those people? They seem to get what we're dealing with." The feeling a prospect has when a competitor's name comes up and they think, instinctively, not them.

None of this fits in a pivot table. So it gets deprioritised. Then underfunded. Then forgotten. Step by step, the McNamara Fallacy plays out in marketing departments everywhere.

The inverse correlation

Here's the part that should genuinely unsettle anyone running a B2B marketing function: the things we can easily measure are often inversely correlated with the things that actually matter.

A gated whitepaper generates a trackable MQL. It also trains your buyer that your relationship is transactional. Their personal data exchanged for a PDF they'll skim for ninety seconds and never open again. The metric went up. The trust went down. The dashboard can't tell the difference.

A high-frequency email sequence generates measurable engagement data. Opens, clicks, replies. It also teaches your buyer to associate your brand with interruption. Every touchpoint is technically trackable. The cumulative effect on how they feel about you is not.

A retargeting campaign generates impressions and click-throughs that look impressive in a report. It also follows your buyer around the internet in a way that feels, at a human level, like being stalked by a company that can't take a hint.

In each case, the measurable metric improves while the unmeasurable reality deteriorates. Trust. Goodwill. The buyer's willingness to pick up the phone. The dashboard shows green. The pipeline tells a different story.

AI is accelerating the fallacy

This is where it gets genuinely dangerous. Because AI hasn't corrected the McNamara Fallacy in B2B marketing. It has supercharged it.

You can now generate, target, distribute, and measure content at a speed and scale that would have seemed absurd three years ago. The production cost of a campaign has collapsed. The measurement infrastructure has never been more sophisticated. And the temptation to optimise for what's measurable has never been stronger.

But if the measurement framework is broken, velocity just means you arrive at the wrong conclusion faster. More dashboards full of more metrics tracking more of the wrong thing, refreshed in real time.

Generic AI makes this worse in a specific, insidious way. A general-purpose model optimises for what it can predict. Which is, by definition, the statistical average of what already exists. Ask it to write a nurture sequence and it will produce something grammatically flawless, structurally familiar, and psychologically empty. It reads well. It measures well. It means nothing to the person receiving it.

The output looks like progress. The metric confirms it. And the buyer deletes it without a second thought, because nothing in it demonstrated that you understand their world.

The antidote isn't better measurement. It's relevance.

If the McNamara Fallacy tells us what to stop doing, mistaking countable outputs for meaningful outcomes, the harder question is what to do instead.

The answer, uncomfortable as it is, is to optimise for something most dashboards don't have a column for: relevance.

Not relevance in the marketing-automation sense. "They visited the pricing page, so trigger the sales sequence." That's behavioural surveillance dressed up as personalisation. Real relevance is the buyer feeling, when they encounter your brand, that you understand something true about their situation. Their pressures. Their constraints. What their Tuesday actually looks like.

When content is genuinely relevant, it builds trust. Not because it asked for trust, but because it demonstrated something that earns it: I see your world clearly enough to say something worth your time.

Trust, as the behavioural psychologist Dr Paul Marsden frames it, sits on two dimensions: competence and care. Competence is whether you can do what you claim. Care is whether you've done the work to understand what the buyer actually needs. Most B2B marketing leads with competence, features, capabilities, benchmark data, and skips care entirely. The result is content that is technically impressive and emotionally irrelevant.

The brands building real pipeline right now are the ones that have flipped this. They invest in understanding buyer psychology. What motivates the person. What they're afraid of. What "success" looks like in their specific context. And then they create content that speaks to that reality. Not at scale for scale's sake. At scale with precision.

This is where AI should be pointed. Not at producing more content faster, but at understanding the buyer deeply enough that what you produce is worth their attention. The tensions. The emotional drivers. The real reasons people choose and act. AI directed at human motivation doesn't just create more. It creates work that resonates. And resonance is the thing no dashboard tracks but every pipeline reflects.

The question for every CMO

McNamara spent the rest of his life haunted by what the numbers missed. The data was never wrong. It was just irrelevant to the question that mattered.

Your MQL count went up this quarter. Your email engagement is healthy. Your content calendar is full. The dashboard is green.

Now ask the question the dashboard can't answer: when your buyer sits in a room with three vendors' logos on a slide, do they trust you? Not because you've measured something at them. Because you've understood something about them.

That's the metric that moves markets. And it starts with being relevant enough to earn it.