Uninformative advertising

Why does an advertisement that tells you nothing about a product still make it more believable?

A famous actor walks through a field at sunset. There is a bottle. Nothing has been claimed — not an ingredient, not a price, not a comparison. And yet something has shifted: the brand feels more solid than the one you have never seen advertised. That reaction is easy to dismiss as gullibility, but it is worth taking seriously first, because it is remarkably consistent, and because dismissing it explains nothing about why firms spend so heavily on messages that say nothing.

Begin with why an informative advertisement is hard to build in the first place. If a claim can be checked before you buy — the size of a screen, the price of a flight — then advertising it is straightforward. But for a great many products, the qualities you care about only reveal themselves in use: whether the shampoo suits your hair, whether the restaurant is any good. Phillip Nelson called these experience goods, and the trouble with them is that any claim the firm makes about quality is exactly as easy for a bad firm to make as a good one. Words are free. So words carry nothing.

If the message cannot carry the information, what could? Ask instead what the advertisement is, apart from its content: it is a large, visible, unrecoverable expenditure. And here the question becomes interesting — who can afford to make that expenditure?

Follow the money for each kind of firm. Suppose you sell something people try once and never buy again if it disappoints. Your expensive campaign brings in a wave of first-time buyers, they try it, they do not return, and the spend is not recovered. Now suppose your product is genuinely good. The same wave arrives, most of them come back, and the campaign is paid for many times over by the stream of repeat purchases. The identical advertisement is a bad investment for the bad firm and a good one for the good firm.

Notice what that means. A firm that spends heavily is telling you something it cannot fake: that it is expecting repeat business. Not claiming it — staking money on it. Nelson's argument, later formalised by Paul Milgrom and John Roberts, is that the expenditure is the message, and the apparent emptiness of the content is beside the point. Any content would do, as long as you can see that it was expensive.

Does the viewer have to work this out? No, and this is why the effect survives in people who would laugh at the argument. All that is needed is the learned association — heavily advertised brands have usually turned out to be acceptable — which is what a lifetime of purchases teaches whether or not anyone reasons about it.

Two refinements keep this honest. The first is that signalling is not the only explanation on offer, and probably not the whole of any case. Mere exposure — the well-replicated finding that repeated encounters with something make people rate it more favourably — would produce a similar effect with no inference at all. So would the brand-salience account, on which advertising mostly works by making a brand come to mind at the moment of choosing, which is a memory effect rather than an argument about quality. These are not rivals to be adjudicated once; they plausibly run together.

The second is that the signalling story is elegant and hard to test. The prediction is a correlation between advertising intensity and quality among experience goods, and the measured correlations have been weak and inconsistent — partly because product quality is difficult to measure independently, partly because the theory does not say how large the effect should be. It also has a known leak: where there is a steady supply of new customers, a poor product can advertise profitably on first-time sales alone, which is roughly the economics of tourist-district restaurants.

• How warranties and free trials work as signals of the same kind, and why a firm might prefer one over the other.
• Why an advertisement's placement — a national broadcast slot rather than a cheap one — may carry more of the signal than the film itself.

Every term the collection defines is gathered in the glossary.