Rory Sutherland: Why the Decisions That Look Most Rational in Business Are Often the Worst

In the 1940s, a US Air Force scientist set out to solve a problem. Fighter pilots were struggling with their planes, controls were awkward, accidents were happening, and nobody could agree on why. His solution seemed obvious: design the cockpit around the average pilot. Measure the forearm length, thigh length and shoulder width of thousands of airmen, take the mean of each, and build accordingly.

Nobody fit.

Virtually no pilot found the seat comfortable. And when researchers looked more carefully, they discovered something that should have been obvious but wasn't: the more dimensions you average across, the smaller the group you're actually designing for. Some pilots had average forearm lengths. A much smaller number had average forearm and thigh lengths. Average across eight or ten variables and you've created a specification that describes almost no one at all. You have designed, as Rory Sutherland puts it, "for a customer that doesn't exist."

Sutherland, vice chairman of Ogilvy Group and one of the most original thinkers in British marketing, tells this story not as a historical curiosity but as a diagnosis of something that runs through modern business. The drive to quantify, to aggregate, to make decisions that can be defended with data, it isn't just limiting. It is actively misleading. It strips out exactly the information that would make a decision good and replaces it with a number that feels safe but points in the wrong direction. Financial rationalism has quietly colonised the boardroom. And the costs are everywhere, even when nobody is connecting them.

Why Driving by the Rear-View Mirror Feels Safer Than It Is

Part of the appeal is understandable. Data from the past is rich. You know a great deal about what happened, what sold, what worked. The future, by contrast, is blurry and then invisible. Sutherland compares it to weather forecasting: computational power has made three or four-day forecasts remarkably accurate, but beyond seven or eight days, you reach what he describes as "the realm of the unknowable." You genuinely cannot know. And the right response at that point is to stop pretending that the data you have will tell you.

There is a thought experiment he likes. Imagine you are trapped by a psychopath in a basement, and your only route out is to correctly describe the weather outside. What is the single best thing you can say? "The same as yesterday." That is the most accurate answer available. It is also the logic that quietly governs most large businesses: do what worked before, do it slightly more efficiently, and don't stray too far from what you can measure.

Roger L. Martin, the Canadian business writer and strategist who has been a significant influence on Sutherland, puts a number on this. Boards of directors are comfortable with roughly 70% of what a business does, because it falls within their direct control, operations, finance, procurement. Marketing and innovation sit outside that zone. They interface with real human beings who cannot be controlled, and who don't behave the way a spreadsheet would prefer. So companies shy away from them. They feel weirdly probabilistic. A bit random. "That," Sutherland says, "is simply the real world."

Marketing Is a Casino and at Some Point You Have to Walk In

Sutherland describes marketing as probabilistic rather than deterministic. On any single decision, it isn't clear what you'll win. But across a body of work, the odds are good. The problem is that people from finance are made deeply uncomfortable by this framing. They want certainty before they commit. They want to know in advance what they're getting.

The solution most businesses settle on is to only commit to what they can already prove. If you can justify it with data, you do it. If you can't, you don't. Sutherland quotes Martin again: if you can only do what you can prove works, you are exploring a very small percentage of the available solution space. There are plenty of logical people around. What businesses consistently lack is the willingness to back an imaginative idea, to walk into the casino, as he puts it, and accept that you cannot know the outcome in advance. "If you want to get lucky," he says, "at some point you've just got to walk into a casino. You've just got to choose the right one."

The alternative he describes is what some media agencies call 70-20-10. Seventy percent of what a business does is established activity, what worked last year, do it again. Twenty percent aims for incremental improvement. And the remaining ten percent is total blue-sky experimentation. In that ten percent, the goal isn't to minimise failure. It's to aim for it. If you're never failing, you're not testing things that are genuinely risky. You're finding slightly different versions of what you already knew worked, and calling it innovation.

The evidence for this comes from performance marketing research that Sutherland finds particularly striking. Analysts discovered that businesses testing eight things weren't simply more successful than those testing four, they were disproportionately more successful. The only plausible explanation: when you commit to testing eight things, you inevitably include a couple of genuinely silly ones alongside the logical choices. And the real breakthroughs, consistently, come from the silly ones. "If there were a logical solution to your problem," Sutherland says, "someone would already have found it." The scarcity isn't logic. It's imagination.

The Variable Nobody Measured and Why It Explains More Than You'd Think

The most persistently overlooked variable in business is psychology. Not consumer psychology as a department or a research programme, but the psychological reality that sits beneath every buying decision and is almost entirely absent from company data.

Sutherland gives several examples, each of them precise. Timotei was a disproportionate success in markets that had assumed it would struggle. Packaged in plain green and white, with no female imagery, it was one of the few shampoos that men would actually buy without embarrassment. The advertising was oblique, a blonde woman washing her hair in a mountain stream, but the packaging had quietly removed a friction point that nobody had thought to measure. Male reluctance to be seen buying products coded as feminine wasn't in any dataset. It was real, and Timotei stumbled into solving it.

The iMac tells a similar story. Before Jony Ive redesigned what a computer could look like, a machine in your living room was a work device, beige, utilitarian, indistinguishable in emotional register from a filing cabinet. The iMac came in colours you might want to lick. It also had a handle. Users weren't expected to carry it. The handle was there because a computer that felt touchable felt like something that belonged in a home rather than an office. It was almost entirely psychological, it cost real money to include, and it changed everything. "They wanted it to feel like something you could comfortably touch," Sutherland says. "The handle was there almost as much for emotional as for functional reasons."

Company data captures none of this. It records transactions. It records what was bought and when. It is gathered, as Sutherland puts it, "for the convenience of the finance function", which means it is automatically a record of behaviour with the emotion stripped out, and almost useless for understanding why any of it happened. Most company data, he argues, tells you a great deal about what customers did and almost nothing about how they felt. And how they felt is where the real decision was made.

Soviet Capitalism: How Narrow Fitness Functions Kill Discovery

The deeper problem isn't that individual businesses fail to understand psychology. It is that the structures governing modern business actively punish the conditions under which good decisions can be made at all.

Sutherland describes a conversation with Stephen Wolfram, the mathematician behind Wolfram Alpha, that he found unexpectedly illuminating. Wolfram observed that evolution generates extraordinary diversity because it operates with a loose fitness function. The definition of success is minimal: stay alive long enough to reproduce. Within that constraint, a patch of moss and a great white shark are equally valid. The ecosystem is inventive and resilient precisely because so many different solutions are allowed to qualify.

What modern finance does, Sutherland argues, is impose a narrow fitness function on business. Quarterly reporting. Shareholder value. Short-term return on investment. If evolution had worked this way, if only creatures with highly efficient gills survived, and efficiency was assessed on a 90-day cycle, the overall ecosystem would be tiny, fragile, and ultimately unworkable. The same logic applies in commerce.

He is not gentle about the parallel. "That's what Soviet Russia did. Business success was defined in advance." The shareholder value movement, for all its apparent commitment to market forces, produces the same result as central planning: you determine in advance what success looks like, and everything that doesn't fit gets cut. The 50% of business success that comes from being able to take advantage of lucky accidents, from stumbling into a market you didn't anticipate, from a product that worked for reasons you never planned, is exactly what a narrow fitness function eliminates.

Martin Sorrell's approach was the deliberate opposite. His guiding principle, "we don't let our strategy interfere with our tactics", sounds like wilful disorder, but it is actually a description of a loose fitness function in practice. Broad direction, yes. Predefined outcomes, no. If something arises that wasn't in the five-year plan, you don't reject it because it wasn't in the five-year plan.

The 180-Degree Flip That Changes Everything

Sutherland's sharpest observation comes near the end of the conversation, provoked by a simple comment. Neil Martin tells him that what he values most about Sutherland's thinking is that it changes how he sees things, not just what he knows, but how he looks.

That, Sutherland says, is exactly the point. And it is exactly what marketing is, or ought to be. Not a function that justifies its existence in spreadsheets. Not a department that produces campaigns and tracks impressions. A state of mind. A particular way of looking at a business, not from the inside out, but from the outside in. What Mark Ritson calls the 180-degree flip: seeing the whole operation as a customer sees it.

Marketers have spent decades defending their function by translating it into financial language, measuring reach, calculating return on ad spend, making the case in terms that finance people can accept. Sutherland thinks this has been a catastrophic mistake. "In seeking to defend the function," he says, "we've destroyed the state of mind." And the state of mind is where the value actually lives.

A board of directors without that state of mind makes decisions that are entirely rational from the inside and invisible disasters from the customer's perspective. His example is self-checkout. The decision to install it reduces labour costs, measurably, directly, defensibly. The finance team claims the credit. But it also offloads work onto customers, removes the human interaction that made theft socially awkward, and has contributed, Sutherland believes, to a significant rise in shoplifting. The problem lands somewhere else in the business, or in the wider system, and nobody who made the original decision is held responsible for it.

A board with a marketer on it, someone capable of making the 180-degree flip, sees that view before the decision is made rather than after the consequences arrive. That is not a soft skill or a creative flourish. It is the capacity to see what a purely internal perspective will consistently miss: that the customer is not within your direct control, does not behave the way your data suggests, and is making decisions for reasons that never appear in a transaction record. Ignore that, and you can build a perfectly defensible strategy for a customer who doesn't exist.