Incrementality is a mindset.

In 1919 John Wanamaker, an American retailer, complained that half his advertising budget was wasted; the trouble was that he did not know which half. A century on, his complaint is usually repeated to show how far marketing has come. It should be read instead as a criticism. In today's fragmented media landscape, where a consumer may glance at a dozen screens before breakfast, the wasted share is almost certainly larger than half. What has changed is not the waste, but the excuse. The tools to measure advertising's true effect now exist, and are cheap. Wanamaker did not know better. His successors choose not to.

The remedy travels under an awkward name: incrementality. Remove the jargon and the idea is simple. It asks what an advertising campaign caused, meaning the sales that would not have happened anyway, rather than what merely coincided with it. A shampoo advertisement that runs while shampoo sales rise has proven nothing; loyal customers, seasonal habit and a rival's stumble may deserve the credit. Incrementality separates the extra from the inevitable. Better still, done properly it turns diagnosis into prophecy: not "what did that campaign do?" but "what will the next one do?" That question is best asked before the money leaves the building, not after.

Two techniques dominate. The first is the controlled experiment, the gold standard of causal proof, borrowed from clinical medicine. Privacy regulation has made tracking individual consumers harder, so the fashionable variant is the geographic test: pick two statistically matched cities, advertise in one, hold back in the other, and measure the gap. The second is market-mix modelling, a statistical exercise that infers, from years of spending and sales data, which channels earn their keep. Purists argue that models infer causation rather than prove it. Practitioners, who must plan next quarter's budget with or without complete certainty, find the inference good enough. The wise firm uses both: experiments to establish the truth, models to apply it more widely.

The price of not knowing

If the methods are well developed and easy to access, why does Wanamaker's problem still trouble so many boardrooms? The usual objection is cost. Measurement budgets are notoriously hard to defend in a planning meeting; they produce no advertisements, win no awards and please no one. This is a common mistake: saving a small amount now while losing a much larger amount later. Consider a brand with revenues of 2bn rupees ($23m). If disciplined measurement improves the productivity of its marketing by a mere 1%, that is 20m rupees of extra revenue, typically many multiples of what the measurement cost, before counting the compounding value of knowing what works next year too. The expense is not the study. The expense is not knowing where the money goes.

The deeper obstacle is psychological. Measurement is a mirror, and mirrors can be unflattering. A careful test will occasionally reveal that the chief marketing officer's favourite campaign added nothing, or that the agency's award-winning advertisement won trophies but not customers. In a firm that does not measure, every decision is exactly as good as the decision-maker claims it is. Ignorance feels comfortable, until the sales figures prove otherwise.

Firms that overcome these obstacles discover that incrementality is less a technique than a mindset. It is the institutional habit of demanding evidence: every initiative measured, every result (flattering or not) investigated, and answers prized above outcomes. That last inversion matters. A campaign that fails but explains why is worth more than one that succeeds mysteriously, because only the former improves the next decision. Firms with this disposition compound their learning; firms without it compound their guesswork.

Building the habit is unexciting work. Bad data usually ruins good intentions first: scattered, inconsistent records will defeat even the best statistical methods, so the basics must be fixed first. The change must also start at the top, since middle managers will rarely volunteer to have their own budgets examined closely. Some firms appoint a measurement chief; others set up task forces or hire consultants. The route matters less than the destination: measurement built into every investment from the start, rather than commissioned occasionally, after the fact and, tellingly, only when the news is expected to be good. Most companies have tried this occasionally, running the odd model or experiment. That occasional effort is exactly the problem. What is optional will remain rare. Start small, prove the value, then make it standard practice.

None of this is easy, and that is rather the point. A firm that refuses to measure is not showing confidence in its marketing. It is showing a preference for comfortable ignorance over uncomfortable arithmetic: laziness disguised as conviction. Wanamaker at least had the excuse of living in 1919. Today the difference between the casual advertiser and the causal one is a single letter, and an entire way of working. Choose the letter carefully.