Words That Work

Luntz is a pollster and message consultant who worked primarily for American conservative politicians and for corporations, and the book draws on focus groups and dial-testing conducted for those clients.

Stating this plainly matters for reading it. The book is a practitioner's account of techniques that were deployed in real political campaigns, and several of its most-discussed examples are terms he helped popularise to reframe contested policy debates.

It is a practitioner's account by someone whose examples are his own campaign work. Read it knowing that.

Devmark’s reading — not a quotation from the book

His central proposition: the intended meaning of a message is irrelevant compared with how it is received, and the only way to know how it is received is to test it.

His methodological point — that writers and speakers systematically overestimate how clearly their meaning arrives, and that testing regularly produces surprises — is sound and applies well beyond politics.

His principles: simplicity, brevity, credibility, consistency, novelty, sound and texture, aspiration, visualisation, asking a question, and providing context.

The most useful are consistency — repeat a formulation far past the point where you are tired of it, because audience exposure is far lower than the speaker's — and context, since a number or fact without a comparison point carries no meaning for a listener.

The book's most famous and most controversial content: pairs of terms with similar denotation and very different reception, each tested.

Several of these were consequential in public debate. His recommendations for how to describe taxation, environmental regulation and energy exploration are the ones most often cited, and they demonstrate the technique working precisely as advertised.

His treatment of how the terms of a debate determine its outcome, and of how a participant who accepts the other side's vocabulary has conceded before arguing.

The underlying linguistics is real and is treated more rigorously by George Lakoff from an opposing political position. Reading the two together is the most instructive way to approach this subject, because they agree entirely about the mechanism and disagree about everything else.

His chapters on business communication: naming, slogans, corporate apologies, and the language of customer-facing organisations.

His analysis of why most corporate statements fail — abstraction, evasion, and language written for lawyers rather than customers — overlaps with Bernoff's and is well observed. His examples of successful brand language are the least dated part of the book.

It cannot be avoided and the book largely does avoid it. Luntz's position is that he clarifies rather than deceives, and that finding the words people respond to is legitimate communication.

The objection is that several of his substitutions change the apparent meaning of a policy without changing the policy, which is a description of euphemism rather than clarity. He has spoken publicly about discomfort with some later uses of these techniques, which is worth knowing.

The objection is that some substitutions change what a policy appears to be without changing what it is.

Devmark’s reading — not a quotation from the book

Stripped of the political content, several things hold: test language rather than assuming it works, prefer concrete and short words, repeat far more than feels necessary, and give numbers a comparison so they mean something.

The observation that audiences respond to the sound and rhythm of a phrase independently of its content is also well founded and rarely acted on outside professional speechwriting.

The examples are firmly of the mid-2000s American political and corporate landscape, and the media environment it assumes — broadcast, campaign advertising, polling — has changed substantially.

It also predates social media entirely, which has altered how phrases spread and how quickly a tested formulation can be identified as such and neutralised.

As a technical manual written by an interested party, useful for understanding how political and corporate language is engineered and how to recognise it when it is being used on you.

That defensive reading is arguably its greatest value now: a reader who finishes it will notice reframing in public discourse that they previously absorbed without seeing.