When Claims Are Free

The word comes from burning. Old English brand meant a firebrand, a piece of burning wood, and by the 1550s it named the mark a hot iron left on a cask, applied to say who had made what was inside and how good it was.^{11 Oxford English Dictionary and Douglas Harper, Online Etymology Dictionary, s.v. “brand.” Old English brand, brond, “fire, flame, firebrand,” from Proto-Germanic *brandaz, “a burning”; Old Norse brandr is a cognate rather than the source. The sense “mark made by a hot iron,” especially on a cask, to identify the maker or the quality of the contents, dates from the 1550s, broadening to “a particular make of goods” by 1854.} Two and a half centuries before that, in 1300, a statute of Edward I required silver in England to be assayed and struck before it could be sold, because a buyer holding a spoon cannot tell what is in it, and a seller can say anything.^{22 The Goldsmiths’ Company Assay Office, “History of Hallmarking.” The 1300 statute of Edward I required the Guardians of the craft to assay work and apply the leopard’s head mark, with silver at the sterling standard of 92.5 per cent.}

Both practices solve the same problem. The buyer cannot judge quality at the moment of purchase, and the seller can.

Every institution we now call marketing grew out of that asymmetry. Akerlof’s 1970 paper on markets with unverifiable quality gave it the formal treatment: where buyers cannot judge, they discount everything, good sellers withdraw, and the market degrades toward its worst participants.^{33 George A. Akerlof, “The Market for ‘Lemons’: Quality Uncertainty and the Market Mechanism,” Quarterly Journal of Economics 84, no. 3 (1970): 488–500.} Nelson distinguished goods you can assess before buying from those you cannot,^{44 Phillip Nelson, “Information and Consumer Behavior,” Journal of Political Economy 78, no. 2 (1970): 311–329.} and Darby and Karni added the ones you cannot assess even after.^{55 Michael R. Darby and Edi Karni, “Free Competition and the Optimal Amount of Fraud,” Journal of Law and Economics 16, no. 1 (1973): 67–88.} Spence explained why a signal is believed: not because it is loud, but because it is expensive enough that a low-quality seller could not afford to send it.^{66 Michael Spence, “Job Market Signaling,” Quarterly Journal of Economics 87, no. 3 (1973): 355–374.}

A brand, on this reading, is not decoration and not a story. It is a technology for making quality legible across a gap, and it works to the extent that faking it is costly.

I want to ask what happens to that technology when the cost of sending a signal falls to approximately zero, which is the situation we are now entering. This essay is an attempt to reason about it carefully rather than dramatically.

Three rules I am holding myself to

Avoid the two available clichés. One says that brand is finished because discovery is algorithmic and attention is bought. The other says nothing has changed and brand simply matters more than ever, which is what people in my industry say when they would like it to be true. Both are positions of convenience, and I will try to take neither.

Acknowledge uncertainty. I do not know how fast this happens, whether buyers adapt faster than I expect, or whether new verification institutions emerge and make the problem look small in hindsight. Several of the arguments below could be wrong in ways I would not currently predict. I have marked the places where I think that is most likely.

Prefer the boring explanation. Where a shift can be explained by a change in relative costs, I will explain it that way before reaching for a cultural one. Culture moves, but prices move first and more legibly.

What I mean by abundance

By abundance I do not mean that everything becomes free. I mean something narrower and more specific: the marginal cost of producing a plausible claim, in any language, in any medium, at any volume, is approaching zero, while the cost of verifying one is not falling at the same rate.

A company can now produce a campaign, a site in fifteen languages, a hundred case studies, a research report, a documentary-grade film and a year of thought leadership for a fraction of what any one of those cost a decade ago. The quality floor of that output is rising and will keep rising.

This is a change in the cost of symbols, not in the cost of substance. Building the product, staffing the support desk, honouring the warranty, being present in a market: none of these have become cheaper. That gap between the price of saying and the price of doing is the subject of this essay.

The two positions, and why I take neither

The first position: brand collapses into distribution. If anyone can produce excellent-looking material, then nothing distinguishes companies except reach and price. Buyers will rely on algorithmic recommendation, aggregated ratings and whatever their assistant tells them, and the accumulated goodwill of a name will matter less each year.

The problem with this position is that it assumes verification gets easier as production gets cheaper. It does not. When the volume of plausible claims rises and the cost of checking them stays flat, buyers do not become better informed. They become more reliant on heuristics, and a brand is the oldest heuristic available. Akerlof’s argument runs in the opposite direction to this position: the harder quality is to verify, the more the market depends on institutions that stand in for verification.

The second position: nothing changes, and brand simply matters more. Since differentiation is harder, the argument runs, a strong brand becomes the decisive asset, and the correct response is to invest more in exactly what we were already doing.

The problem here is that it ignores what made brands work. A brand functions as a signal because it is expensive to build and expensive to damage. If the visible components of it, the language, the imagery, the apparent sophistication, the volume of evidence, can be produced by anyone at near-zero cost, then those components stop telling anyone anything. They are still necessary. They have simply stopped being sufficient, and continuing to invest in the cheap layer while calling it brand-building is how a company ends up well-dressed and unbelieved.

The position I actually hold is narrower than either. The function of brands is becoming more important and the usual instruments for building them are becoming less effective, because those instruments were valuable in proportion to their cost. The gap between the two is where the next decade of this discipline will be decided.

Five consequences

1. Signal inflation. When everyone can produce the markers of quality, the markers stop marking. Professional-looking is now free. Extensively-documented is now free. Multilingual is now free. Each of these was, until recently, evidence that a company had resources and had made a choice to spend them here. None of them is evidence of anything now, and a buyer who has not yet updated on this will be wrong more often.

2. The return of the lemons problem, at the level of categories. If claims are free and verification is not, the rational buyer discounts all claims. This is already visible in how people evaluate software: demos are distrusted, reference calls carry more weight than any material a vendor produces, and the most valued input is somebody who has actually used the thing. Buyers have routed around the vendor’s own material.

3. Convergence, and then price. Porter’s argument about operational effectiveness applies exactly.^{77 Michael E. Porter, “What Is Strategy?” Harvard Business Review 74, no. 6 (November–December 1996): 61–78.} When everyone can reach the same frontier of execution, competition collapses onto the one axis left. Categories where every company can produce excellent material and none can differentiate will compete on price, and this will be experienced by the people inside them as a mysterious loss of pricing power rather than as a predictable consequence.

4. Fluent absence. What a company says and shows can now be adapted for any market instantly. What it charges, where it sells, how it delivers and who answers the phone cannot. The predictable result is companies that sound native and behave foreign, which is worse than being uniformly foreign, because fluency raises an expectation the rest of the company then fails to meet.

5. Attention moves to what cannot be produced. This is the most interesting consequence and the least discussed. If symbols are abundant, the informative things are the ones that remain costly: physical presence, duration, staking something that can be lost, refusing a market, employing people whose names are attached to the work, being wrong in public and correcting it. These are not aesthetic choices. They are the remaining expensive signals, and they will be read as such whether or not companies intend them.

What people are actually short of

Separate what is abundant from what is scarce, because marketing’s attention has not followed the shift.

Baumol’s observation about cost disease is the useful one here.^{88 William J. Baumol and William G. Bowen, Performing Arts: The Economic Dilemma (New York: Twentieth Century Fund, 1966); and Baumol, “Macroeconomics of Unbalanced Growth: The Anatomy of Urban Crisis,” American Economic Review 57, no. 3 (1967): 415–426.} In an economy where productivity rises fast in some sectors and barely at all in others, the sectors that cannot be automated take an ever-larger share of income and anxiety. Manufactured goods and digital content get cheaper every year. Care, housing, health, education, energy and food do not, and they are where households now feel the pressure.

The discipline’s centre of gravity has not moved with this. Marketing’s best people, its best thinking and its most admired work are concentrated in consumer technology and consumer goods: categories where supply is abundant, novelty is cheap, and the marginal social value of persuading someone to switch is close to zero.

Meanwhile the categories where a buyer cannot assess quality, and where being wrong is expensive, are the ones that most resemble the conditions brands were invented for. A clinic, a school, an energy supplier, an insurer, a bank, a care provider. These are credence goods in the technical sense. The institution that solved this problem for silversmiths in 1300 is exactly the institution they need, and they are served, on the whole, by the least interesting work in the field.

I think this is where the next decade of useful marketing practice is, and I hold that view with moderate confidence. The counterargument, which I take seriously, is that these categories are regulated, slow and often publicly funded in ways that limit what brand work can do, and that the gap persists because of how those markets are built rather than because nobody has noticed it.

Hope, fear, or restraint

The question I get asked in different forms is what a company should sound like now. The available registers are hope, which dominated the last decade of technology marketing, and fear, which is dominating the current one.

I think both are wrong, and the reason is Spence’s, not a matter of taste.^{99 Spence, “Job Market Signaling.” The argument that a signal tells a buyer something in proportion to what it cost the sender is the idea this section rests on.}

Hope is cheap to produce. Any company can promise a transformed future, and now every company can produce that promise in fifteen languages with a film attached. Because it costs nothing, it tells a buyer nothing, and audiences have already discounted it. The same is becoming true of fear: warning about a danger that your product happens to solve is a free move, and it is being made so often that it has stopped landing.

Restraint is the only register that is still expensive. Saying what your product does not do, which customers you are not for, what you have decided not to sell, what you got wrong: each of these costs something real and immediate. A company that publishes its limits has forgone revenue to do it, which is what makes the statement credible under Spence’s conditions. A refusal cannot be faked cheaply, because a company that would say anything has no reason to refuse anything.

This is not a recommendation to be gloomy or modest as a style. Style is free, and a performed humility is as cheap as a performed vision. It is an argument that the informative content of any message is now roughly equal to what it cost the company to say it, and that in an abundance of claims the only statements that tell a buyer anything are the ones with something forgone inside them.

What a leader should do

Audit which of your signals are still costly. Take the last twenty things the company published and ask, for each, whether a competitor could have produced it in an afternoon. Everything that could be is now table stakes, and should be produced efficiently and not celebrated.

Move investment from production to verification. Third-party proof, named customers who will speak, published limitations, independent testing, open methodology. These are expensive, which is the point.

Do not let what you say run ahead of what you have built. In any market where the price, the channel and the support are not local, deliberately plainer communication is the cheaper and more credible position.

Say what you will not do, and mean it. A refusal list is a strategic document, not a communications one. Porter’s account of strategy as trade-off is the same argument arriving from a different direction: a position that costs you nothing is not a position.

Staff for discrimination, not production. The scarce input is the judgment about whether a thing is any good and right for this market. That capacity does not increase because production got cheap, and the organisations that will do well are the ones that noticed where the value moved.

What would change my mind

If credible, cheap verification infrastructure emerges, something like assay offices for claims, then the asymmetry I have built this on narrows and brands matter less rather than more. I think that is possible and I would welcome it.

If buyers adapt faster than I expect, developing reliable heuristics for distinguishing costly signals from cheap ones, then the transition is shorter and less damaging than described.

And if the cost of doing, rather than saying, falls as fast as the cost of saying, then the gap I have spent this essay on closes and the problem dissolves. I see no sign of that, but it is the assumption most worth watching.

What I am confident about is narrower. Marking a thing to say who made it and what it is worth is one of the oldest technologies we have, and it was invented for exactly the conditions we are re-entering at a scale its inventors could not have imagined. It would be strange to abandon it now. It would be stranger still to keep operating it as though the mark were the expensive part.