Alchemy

by Rory Sutherland

  Yet in 1925, Sears opened its first bricks-and-mortar shop. By 1929, the companies had opened a further 800 between them – perhaps Amazon’s purchase of Whole Foods Market is history repeating itself.* I could go on endlessly about the psychological factors at play here, but let’s go back to the lazy assumption that 1 x 10 is the same as 10 x 1, which is also relevant.* Online shopping is a very good way for ten people to buy one thing, but it is not a good way for one person to buy ten things. Try and buy ten different things simultaneously online* and it turns chaotic. Items arrive on four separate days, vans appear at your house at different times and one delivery always fails.* By contrast, the great thing about Walmart, which investors tend to overlook, is that people turn up, buy 47 different things and then transport them home at their own expense. Amazon can be a very big business selling one thing to 47 people, but if it can’t sell 47 things to one person, there’s a ceiling to how large it can be.

  Many other mathematical models involving humans make the mistake of assuming that 10 x 1 = 1 x 10. For instance, our tax system assumes that ten people who earn £70,000 for one year of their life should be taxed the same amount as one person who earns £70,000 for ten consecutive years, yet I have never heard anyone question this – is it another example of bad maths?

  Recently I was involved in a discussion of train overcrowding.* Again, metrics do not distinguish between ten people who have to stand 10 per cent of the time and one person who has to stand all the time, but these two things aren’t the same at all. If I am an occasional traveller and find myself standing once a month, so be it, but if I had paid £3,000 for a season ticket and didn’t ever get a seat, I would feel robbed. Reframed this way, the problem becomes easier to solve. Why not run two trains in each direction each day exclusively for season ticket holders, or give them the right to sit in first class when standard class is full?* Or better still, expand first class and allow all season ticket holders to sit in it. You haven’t solved the problem of train overcrowding, but you have solved it for those who are worst affected, which is what really matters.

  1.10: Recruitment and Bad Maths

  Okay. So it might work for trains. But if I told you that you could use a similar insight to increase employment diversity, you probably wouldn’t believe me. But again, 10 x 1 does not equal 1 x 10. Imagine you have ten roles to fill, and you ask ten colleagues to each hire one person. Obviously each person will try to recruit the best person they can find – that’s the same as asking one person to choose the best ten hires he can find, right? Wrong. Anyone choosing a group of ten people will instinctively deploy a much wider variance than someone hiring one person. The reason for this is that with one person we look for conformity, but with ten people we look for complementarity.

  If you were only allowed to eat one food, you might choose the potato. Barring a few vitamins and trace minerals, it contains all the essential amino acids you need to build proteins, repair cells and fight diseases – eating just five a day would support you for weeks. However, if you were told you could only eat ten foods for the rest of your life, you would not choose ten different types of potato. In fact, you may not choose potatoes at all – you would probably choose something more varied.

  The same applies to hiring – we are much more likely to take risks when hiring ten people than when hiring one. If you hire ten people, you might expect one or two of them not to work out: you won’t risk your reputation if a couple of them leave after two years, or if one starts stealing staplers and photocopying his bottom at the Christmas party. But if you hire one person and they go rogue, you have visibly failed. So individuals who are hiring individuals may be needlessly risk averse; they are hiring potatoes.

  When hiring, we should understand that unconscious motivation and rational good sense overlap, but they do not completely coincide. A person engaged in recruitment may think they are trying to hire the best person for the job, but their subconscious motivation is subtly different. Yes, they want to hire a candidate who is likely to be good, but they are also frightened of hiring someone who might turn out to be bad – a low variance will be as appealing to them as a high average performance. If you want low variance, it pays to hire conventionally and adhere to the status quo, while people hiring a group of employees are much more likely to take a risk on some less conventional candidates.

  We can see this diversity mechanism clearly in house hunting. If I were to give you a budget to choose your perfect house, you would have a clear idea of what to buy, but it would typically be a bit boring. That’s because when you have one house, it cannot be too weak in any one dimension: it cannot be too small, too far from work, too noisy or too weird, so you’ll opt for a conventional house. On the other hand, if I were to double your budget and tell you to buy two houses, your pattern of decision-making would change. You would now be looking to buy two significantly different properties with complementary strengths – perhaps a flat in the city and a house in the countryside.

  If you are choosing a parliamentary candidate, the safe option is to pick a vapid-but-presentable PPE graduate, yet no one choosing ten candidates would pick ten of those – they’d throw in a few wild cards.* Cecil ‘Bertie’ Blatch clearly understood this when, as President of the Finchley and Friern Barnet Conservative Association, he decided to ‘lose’ a couple of a more conventional candidate’s votes to give Margaret Thatcher her first winnable seat. He wasn’t cheating; he was correcting a mental bias. Once you understand this, the potential to increase the level of diversity in employment, education or politics, without imposing quotas, rises as people are chosen in batches.*

  Everyone worries about declining social mobility, rising inequality and the hideous homogeneity of politicians, yet it is possible these have arisen from well-meaning attempts to make the world fairer. The quandary is that you can either create a fairer, more equitable society, with opportunities for all but where luck plays a significant role, or you can create a society which maintains the illusion of complete and non-random fairness, yet where opportunities are open to only a few – the problem is that when ‘the rules are the same for everyone’ the same boring bastards win every time. The idea that you should therefore try making your recruitment system less fair outrages people when I suggest it, but it is worth remembering that there is an inevitable trade-off between fairness and variety. By applying identical criteria to everyone in the name of fairness, you end up recruiting identical people.*

  At Ogilvy we now recruit creative talent through an internship scheme called ‘The Pipe’. Applicants don’t have to be graduates; they don’t have to be young; they don’t have to have any qualifications at all – in fact, we recruit them blind for the first few stages. It is too early to make any definitive judgement about the programme’s success, but the recruits seem every bit as interesting to chat to as the Oxford graduates – and more interesting, in some cases.* Within months of our first intake joining, several of them had won an award at the Cannes Festival of Creativity for an advertising idea for Hellmann’s Mayonnaise – something some people spend a lifetime in the industry without achieving.

  Remember, anyone can easily build a career on a single eccentric talent, if it is cunningly deployed. As I always advise young people, ‘Find one or two things your boss is rubbish at and be quite good at them.’ Complementary talent is far more valuable than conformist talent.

  1.11: Beware of Averages

  When Lieutenant Gilbert S. Daniels, a physical anthropologist, was hired by the US military to design a better cockpit for high-speed aircraft in the early 1950s, the assumption he had to challenge was that you should design a cockpit for ‘the average man’. The idea was that if you took an average of many pilots’ bodily dimensions, you would have a template around which you could design a cockpit – with the instruments visible to most people, and with the controls within easy reach of all but the most unusual physical specimens.

  However, Daniels already knew from his measurements of human hands
that an average human hand is not a typical human hand, and in the same way he found that an average human body – that is one which is average in a range of dimensions – is surprisingly rare. When you designed a cockpit for an average man you were designing a cockpit not for everyone, but for a surprisingly rare, or even non-existent, body-type. Not a single pilot of the 4,000 measured was within the average range on all ten bodily measures.*

  Don’t design for average.

  Metrics, and especially averages, encourage you to focus on the middle of a market, but innovation happens at the extremes. You are more likely to come up with a good idea focusing on one outlier than on ten average users. We were discussing this recently in a meeting when a round of sandwiches arrived. ‘This proves my point exactly,’ I said, pointing at the food. The sandwich was not invented by an average eater. The Earl of Sandwich was an obsessive gambler, and demanded food in a form that would not require him to leave the card table while he ate.

  Weird consumers drive more innovation than normal ones. By contrast, it is perfectly possible that conventional market research has, over the past fifty years, killed more good ideas than it has spawned, by obsessing with a false idea of representativeness.

  1.12: What Gets Mismeasured Gets Mismanaged

  The above findings have important implications for metrics. Businesses love them, because they make it easy to compare and manage things. It’s true that ‘what gets measured gets managed’, but the concomitant truth is ‘what gets mismeasured gets mismanaged’. One great problem with metrics is that they destroy diversity because they force everybody to pursue the same narrow goal, often in the same narrow way, or to make choices using the exact same criteria.

  I have never seen any evidence that academic success accurately predicts workplace success.* The eminent obstetrician and fertility specialist Professor Lord Winston does not seek out academic high-flyers to work with in medicine, even though he could have the pick of the crop. However, it is now common practice in British firms only to interview people with an upper second-class degree or above, a criterion that is applied with no evidence but simply because it seems logical. If you need to select from a list of graduate applicants, using their undergraduate performance as a filter seems to make sense, but without more evidence it is absurd – and when practised on a wide scale, it’s a disgraceful waste of talent.

  It wouldn’t matter much if only Goldman Sachs, say, or a few elite institutions used this criterion, but when everyone else copies the same approach, it is ludicrous. Since almost half of graduates should by definition fall below this hurdle, it will either result in thousands of people spending three years at university for no benefit or to grade inflation in universities, with degree classes becoming meaningless.* This is another example of people not using reason to make better decisions, but simply for the appearance of being reasonable.

  As any game theorist knows, there is a virtue to making slightly random decisions that do not conform to established rules. In a competitive setting such as recruitment, an unconventional rule for spotting talent that nobody else uses may be far better than a ‘better’ rule which is in common use, because it will allow you to find talent that is undervalued by everyone else.

  One of the other problems with a logically consistent system for hiring people is that ambitious middle-class people can exploit it by ‘gaming the system’. Violin lessons – check, work placement at uncle’s bank – check, charity work with the disadvantaged – check,* high GPA – check. By contrast, if you hire that brilliant backgammon player, you do know one thing. He’s genuinely talented at something – and it’s unlikely that his parents have spent a fortune on private backgammon lessons.

  Real excellence can come in odd packaging. Nassim Nicholas Taleb applies this rule to choosing a doctor: you don’t want the smooth, silver-haired patrician who looks straight out of central casting – you want his slightly overweight, less patrician but equally senior colleague in the ill-fitting suit. The former has become successful partly as a result of his appearance, the latter despite it.

  1.13: Biased about Bias

  Is some of what we think to be racial or gender bias merely bias in favour of the status quo? After all, the less eccentric your hire, the less blame you are exposed to if something goes wrong. The advertising industry at present is obsessed with gender ratios and ethnic composition; it is perfectly reasonable to look at these figures, but the industry seems completely blind to another bias, which is that it is extraordinarily prejudiced in favour of hiring physically attractive people.

  Understanding what is going on subconsciously is vital here – if we assume that racial bias is the only subconscious bias, we may introduce misdirected measures. If, however, we consider the possibility that part of the issue regarding a lack of diversity arises from status quo bias, the solutions will be very different. I am not denying for a moment that unconscious racial bias exists, but I am suggesting that people may have an unhealthy predisposition to attribute all disparities of outcome and opportunity between different ethnic groups to it, whereas in reality many other forces may be at work. If you want to solve the problem, you have to understand ‘the real why’.

  Some evolutionary psychologists, most notably Robert Kurzban, believe that racial prejudice is a relatively weak force in human psychology since for most of evolutionary history we wouldn’t have encountered people of a different ethnicity. We might, rather, be heavily predisposed to attaching ‘outsider status’ to people who speak with a different accent to us, since such distinctions would have been experienced more frequently. Coming from a country that historically was obsessed with accents,* this does seem worth investigating. Would a privately educated Nigerian be at a disadvantage in seeking a job in London in competition with a white Liverpudlian with a strong Scouse accent? I rather doubt it. Kurzban’s work suggests that prejudice depends on context; we may be more prone to ascribe ‘outsider status’ to someone of a different ethnic background, but this does not mean that we cannot easily adopt anyone of any ethnic background into an ‘in group’ in a different setting.

  I am slightly sceptical about using gender and ethnic quotas as a way of tackling prejudice, since they define diversity in narrow ways. For instance, it would be perfectly possible to improve your racial diversity figures by employing ten highly talented Nigerians. You might congratulate yourself for the admirable diversity shown by your firm, but what if you were to then find out that all ten came from the Igbo tribe and none come from the almost equally populous Yoruba tribe – would you bury this fact and remain smug about your newly diverse workforce, or would you ask whether your definition of diversity is a little heavy on weighting skin colour and a bit light on everything else?

  A recent article in Harvard Business Review showed that apparent gender or racial bias may not only arise for the assumed reasons; other unconscious mechanisms may also be involved, depending on the context. It reported on three studies which examined what happens when you change the finalists for a job. The research revealed that: ‘When there is only one woman, she does not stand a chance of being hired, but that changes dramatically when there is more than one. Each added woman in the pool does not increase the probability of hiring a woman, however – the difference between having one and two women seems to be what matters. There were similar results for race when we looked at a pool of four candidates.’

  Recreated by Greg Stevenson, with permission of Stephanie K Johnson

  Are people biased against minorities – or are they biased against anyone in a minority of one?

  This suggests that the prejudice we apply against a lone black candidate or a lone female candidate might also apply to a lone ‘anything’ candidate.*

  1.14: We Don’t Make Choices as Rationally as We Think

  The context and order of choosing affects things in ways we would not consciously expect – not just in corporate decision-making and recruitment, but also in personal decisions. The psychologist and behavioural
economist Dan Ariely was one of the first people to highlight the famous decoy effect* in the decision process – the phenomenon whereby consumers tend to have a specific change in preference between two options when also presented with a third option that is more desirable than one, but less desirable than the other.

  A seminal example explored by Ariely is the Economist magazine’s subscription offer. The middle option – where you are offered a print only subscription for $125 – is known as a decoy. No one – except perhaps someone who deeply loathes technology – would ever choose it, since for the same price you could get the full print and online subscription, but it does have a huge effect on behaviour. By creating a very easy ‘no-brainer’ decision, it encourages more people to take up the higher-value full subscription. In one experiment carried out by Ariely, it led 84 per cent of putative subscribers to choose the full-price, all-in subscription. However, when you remove the dummy option and only have the two sensible ones, preferences are reversed: 68 per cent chose the lower-priced, online-only subscription.

  Recreated by Greg Stevenson, based on study referred to in Dan Ariely’s Predictably Irrational

  Context is everything: strangely, the attractiveness of what we choose is affected by comparisons with what we reject. As one friend remarked, ‘Everyone likes to go to a nightclub in the company of a friend who’s slightly less attractive than them.’

  Estate agents sometimes exploit this effect by showing you a decoy house, to make it easier for you to choose one of the two houses they really want to sell you. They will typically show you a totally inappropriate house and then two comparable houses, of which one is clearly better value than the other. The better value house is the one they want to sell you, while the other one is shown to you for the purpose of making the final house seem really good.