Key Insights: Alchemy: The Dark Art and Curious Science of Creating Magic
**Rory Sutherland • 5 Minutes • Behavioral Science and Marketing**
Sutherland, vice chairman of Ogilvy UK, argues that business has over-trained itself in logic and under-trained itself in psychology. Logic is excellent at destroying bad ideas and nearly useless at generating good ones. The ideas that produce outsized returns look irrational until they work. He calls the reasoning behind them psycho-logic: the internal logic of a mind that cares about meaning, status, effort and fairness more than efficiency.
5 Pillars of Success
1. The opposite of a good idea can also be a good idea Logical thinking converges. Hand a rational team a problem and it produces roughly the same answer as every other rational team, which guarantees competition on cost and speed. Logic, Sutherland says, is a sieve: it filters ideas and never generates them. His illustration: the London to Paris rail link. The engineering answer to a slow journey was billions for new track to cut about forty minutes. He proposed spending a fraction of that on making the journey pleasant — attentive service, champagne — because passengers do not experience duration, they experience how the time felt. Both answer the same complaint. One costs a thousand times more.
2. Costly signalling: value is inferred from what it costs to fake People rarely evaluate quality directly. They read proxies, and the most trusted proxy is expense, which is hard to fake. Red Bull built a category on a flavour many drinkers call unpleasant; the medicinal taste is evidence that something is happening, a costly signal a delicious drink could not send. The London black-cab driver's Knowledge — years spent memorising central London's streets — is the same mechanism: the sunk cost is the trust. Price behaves identically; the same wine is rated more enjoyable when the drinker believes it is expensive. Cost is not what you pay for value; it is often where value comes from.
3. Solve the perception, not the physics Waiting is an experience, not a measurement, and the experience is far cheaper to redesign. Houston airport suffered persistent baggage-delay complaints. More staff and conveyor capacity was expensive; what worked was moving arriving gates further from the carousel so passengers walked longer and stood still less. Total time rose and complaints collapsed. Hotel lifts get the same treatment: mirrors beside the doors make the wait feel shorter than any engineering upgrade. The moving car on a ride-hailing map does work no reduction in pick-up time can match. Occupation, visible progress and a sense of agency beat speed on price.
4. The doorman fallacy: never define a role by its literal function Define a doorman as a mechanism for opening a door and the automatic door obviously wins: cheaper, always on duty, never ill. Hotels that made that substitution saved money and lost the feeling of arrival the doorman was really producing. Sutherland calls this the doorman fallacy, mistaking a role for its most measurable component. The corollary is dangerous: what cannot be measured gets cut first, and once a metric becomes a target it stops measuring what mattered. Nespresso is the mirror image: a commodity bean turned into a luxury ritual by story, design and ceremony rather than better coffee.
5. Test the paradox: behaviour beats stated preference Ask people why they buy something and they hand you a tidy, flattering, largely invented account. Watch what they do and you learn something else. The jam experiment is the standard demonstration: a stall offering twenty-four varieties attracted more browsers, while a six-variety stall sold more. Abundance draws attention and paralyses purchase. This is why Sutherland trusts experiments over surveys, but not incremental ones. Optimisation testing reliably finds the best version of the idea you already had; it will never propose champagne instead of a railway. The counterintuitive hypothesis is where the asymmetric upside lives, because no logical competitor will bother to try it.
Key Takeaway
“The opposite of a good idea can also be a good idea.” — Rory Sutherland
Reason is a good servant and a poor master: where the spreadsheet sees time or money, the customer is usually experiencing something else.
Practical Task
1. Pick one problem you have framed as an engineering problem: a wait, a price, a conversion rate, a churn number. Write down the logical fix and what it would cost. 2. Write three fixes that change perception instead of physics: occupy the waiting, show visible progress, add a ritual, attach a costly signal, or put a named human in the loop. 3. Price each one and pick the cheapest reversible option. Alchemy passes the test when it costs a fraction of the engineering answer. 4. Ship it as a live experiment this week and judge it on behaviour, not on what people say they prefer in a survey. 5. Then find the doorman in the system: the line item that shows no measurable return and quietly carries the meaning. Protect it before the next cost-cutting round.