The 93% Myth: What Mehrabian Actually Measured

You have heard it: 7% of what we communicate is words, 38% is tone of voice, 55% is body language. The rule shows up in sales training, in public speaking courses, in leadership manuals, and it has become the justification for the broader claim that "93% of communication is nonverbal."

The origin is two studies published in 1967 by Albert Mehrabian and colleagues.¹ What they measured was far narrower than the rule implies: how people infer attitude — liking versus disliking — when they receive a contradictory signal, using stimuli that were, in one of the experiments, single recorded words delivered in different tones.²

Three limitations follow directly, and none of them is subtle:

1. The object was affective attitude, not communication in general. The studies say nothing about transmitting a technical instruction, a price or a deadline.
2. The condition was deliberate inconsistency between channels. When channels agree — the normal case — the proportion does not apply.
3. The stimuli were minimal: no syntax, no context, no subject matter. A single word is not a conversation.

The author spent decades saying this

This is what makes the myth interesting: it is not a case of bad science later corrected. It is a valid result, in a narrow scope, generalised by third parties far beyond what it supports — while the original author repeatedly tried to contain the generalisation, without success.³

And the elementary arithmetic dispenses with the literature entirely. If 93% of communication were nonverbal, you would understand 93% of a conversation held in a language you do not speak. Nobody who has been in a foreign country believes that for two seconds. The rule survives because it is easy to quote, not because anyone tested it against their own experience.

The most direct way to see this is to read the primary source. Mehrabian's Silent Messages (1971) is more careful than the rule extracted from it: the qualifications are on the page, in plain language. They were simply discarded somewhere between the book and the training slide.

Why the myth is so durable

Because it is useful to whoever is selling. A specific number, with implied decimal places and an academic name attached, solves the rhetorical problem of justifying why someone should pay for presence training. "Nonverbal communication matters" is true and sells nothing. "93%" sells.

It is worth recording that this is not the pattern's only appearance. Power-posing research followed a similar trajectory: a high-profile initial finding,⁴ a significant replication failure for the hormonal effects,⁵ and a partial retraction of the effect by one of the original authors.⁶ The subjective component — feeling more confident — remains more defensible than the physiological one. But the version that travelled through conference stages and corporate training was the stronger claim, not the one that survived.

What to put in its place

Discarding the myth does not mean concluding that the nonverbal channel is irrelevant. It means trading a false proportion for three claims that hold:

• The nonverbal channel dominates when the subject is affect and the channels contradict each other. That is exactly what Mehrabian measured, and within that scope the finding stands.
• No isolated signal is diagnostic. What carries information is deviation from an individual baseline — and even then as an indication, not a conclusion. We covered the method in Baseline First.
• Precision matters more than the strength of the claim. A publication that quotes "93%" loses authority with every reader who knows the literature, and gains reach with those who do not. That is a poor long-term trade, and it is the reason this desk exists.