Читать книгу The Behaviour Business - Richard Chataway - Страница 11

Chapter 3: Test-Tube Behaviours – How to Deliver Marginal Gains Using Behavioural Science

‘Sciencing the shit’ out of problems

In the Oscar-nominated movie The Martian, Matt Damon plays a NASA botanist stuck on Mars. His crew have departed after they have (reasonably) assumed he has been killed by an accident during their mission.

Damon’s character survives. It will be several years before a rescue mission can reach him – but he only has sufficient supplies to last a few months. As a highly qualified scientist, he does not panic. He decides to solve the problem in the most effective way possible.

In his words, he decides to “science the shit” out of the problem.

He consults the notes left behind by his colleagues, experiments using the limited resources at his disposal – including a highly creative way of growing potatoes with the help of his own faeces – and he keeps himself alive.

Not only is this a great movie, but it is also a powerful allegory for the way science has solved mankind’s problems. The houses we live in, the food we eat, the transport we use: all of these contain innovations that were developed by scientists using the scientific method.

A hypothesis based on existing evidence, followed by a deduction, and tested through observation. Then repeat.

And yet, in business, little work is scientifically based. In fact, most of it involves no experimentation and an awful lot is based on outdated assumptions. Isn’t it time we removed the guesswork?

One of the characteristics of governmental behavioural teams like the BIT is their use of best practice, scientific methods. Often, this means using randomised controlled trials (RCTs). David Halpern, CEO of the BIT, calls them the “gold standard of evidence-based policy”.28

RCTs are a methodology developed by medical science, where the efficacy of a medicine (for example) is tested by evaluating that medicine against a control condition, that is, one where no medicine (or an existing medicine or placebo) is given. Patients are allocated at random (hence the name), with the aim of eliminating bias. Data is then analysed to look for statistically significant differences between the two groups’ outcomes to determine not only effectiveness, but also potential side effects.

Social (including behavioural) science experiments generally involve testing a behavioural intervention (nudge) against a control condition – which is usually no change. In one BIT tax experiment for HM Revenue and Customs, they tested a social proof letter saying, “Most people pay their tax on time”. This was then evaluated against the existing HMRC letter, with taxpayers receiving one of the two letters at random. The test letter resulted in a 15% increase in the number of people paying before the deadline.

Having seen the overall success of this new letter in getting people to pay their tax on time, the BIT team interrogated the data a bit further. They found that amongst the top 5% of taxpayers, the social proof message had actually reduced the likelihood of paying on time by 25%.29

Amongst this group – it failed.

To find out what did work for this group, the BIT ran another RCT in 2015 to see what message worked best amongst the highest taxpayers. This turned out to be a loss aversion30 message based on the effect on public services of not paying tax – this increased payment rates by 8% amongst the highest taxpayers.

And, thus, progress was made. Hypothesis, deduction, observation.

Or, to put it another way: test, learn, adapt. In this case, it was only possible to understand how behaviour was heavily influenced by this context (i.e. the choice architecture) by sciencing the shit out of the problem.

Loss aversion

You may have experienced FOMO (fear of missing out), or the realisation that you want to do something (e.g. going to a party or seeing a movie) not because you especially desire it, but because you fear regretting not doing so. This is a manifestation of a particularly prevalent behavioural bias: loss aversion.

Put simply, loss aversion is the tendency for our behaviour to be more influenced by the risk of a negative outcome (loss), than the chance of gain. We generally feel the loss of £5 more keenly than gaining £5. It explains related biases like the endowment effect (that we value something we own more highly than an identical item we don’t) and scarcity bias (our increased desire for things we think are in short supply).

Evolutionary psychologists explain this in terms of our survival instincts. In a world of scarce resources, we needed to ensure that we harness and keep as much as we can, while we can, for fear it might be gone tomorrow. Even in a world of relative abundance, this still guides our behaviour.

Consequently, we also have a present bias: we value things more today than in the future, largely because we can better visualise what we can do with resources (e.g. money) now. It explains why we are generally so bad at saving for retirement, and often reach the end of the month with less money in the bank than we expected.

This bias is used frequently by businesses, particularly in marketing, through limited time offers, closing down sales and the like. In the digital world, when buying tickets or hotel rooms, for example, we often see it through nudges like ‘only 5 left at this price!’ Instinctively, this makes us want the product more and we become more likely to purchase immediately. But, note the clever phrasing here: ‘at this price’, means the price could just as well go down as up! Our inherent bias makes us assume that if we don’t grab it now, it will cost us.

When working with the call centre described later in this chapter, we experimented with using loss aversion for customer benefit. When talking about the benefits of moving to online banking – specifically that it is safer, quicker, and better for the environment (because it removes the need to send out paper statements) – we found we could significantly increase the likelihood of customers taking up the service by simply suggesting that if they didn’t, they would miss out. Previously, customer service representatives (CSRs) had quite rationally talked through these benefits, but found customers tended to ignore the information or cut them off before they finished.

Simply saying that a customer would miss out by not changing worked much better – often without the need to even say what the benefits are (suggesting they already knew what they were, they were just too cognitively lazy/uninterested to change). If a customer queried this, the CSR could then talk them through the advantages, but, in most cases, the decision had already been made.

This is yet another example of how much faster our instinctive system-1 processes are than our more conscious system-2 ones – and more influential.

Learning from failure

In his book Black Box Thinking, the journalist Matthew Syed contrasts two industries – medicine and aviation – in terms of how they encourage best practice and avoid the negative outcomes of failure (which is often death). In contrast to how medicines are tested using RCTs, he found that the medical industry does not generally adopt a scientific approach in its response to failure. Medical mistakes are often covered up, ignored, or worse – usually for fear of reprisal and legal action.

But as Syed puts it: “science [is] a discipline where learning from failure is part of the method.” This is why the introduction of black boxes in the 1960s – that retain all data pertinent to an air crash (which is then shared globally amongst the entire industry)31 – has had such an impact on aviation safety.

Indeed, 2017 was the first year on record where there was not a single fatality as a result of a commercial airline crash. A total of 399 people died globally solely in freight and military crashes that year – by contrast, in 1972, a total of 3,346 people were killed.32

Syed categorises this scientific approach to failure as characteristic of a ‘growth mindset’ (as opposed to a ‘fixed mindset’) for organisations, and the best way to deliver incremental improvements through marginal gains. He quotes the philosopher and scientist Karl Popper: “The history of science, like the history of all human ideas, is a history of … error. But science is one of the very few human activities – perhaps the only one – in which errors are systematically criticised and fairly often, in time, corrected. This is why we can say that, in science, we learn from our mistakes and why we can speak clearly and sensibly about making progress.”

If a business wishes to progress, grow and succeed, then understanding the merits of a scientific approach, and the value of testing, is critical. It must also recognise that this process is an iterative one, where we can learn as much from failure as from success, as with the BIT HMRC experiment. In this way, behavioural science – by requiring testing to establish what works in a realistic context – can increase the effectiveness of established management techniques based on continuous improvement and marginal gains, such as kaizen, lean thinking, and agile processes.

Applying a growth mindset to business challenges

As an example of this, in 2017/8 I worked on a project with partners at OEE Consulting,33 a leading services and operations management consultancy. The client was an outsourcer that ran a call centre for one of the UK’s largest savings banks (having over 20 million customers). OEE Consulting were developing a number of new processes and systems, based on lean principles, to deliver better processes in the call centre. These had both an efficiency (i.e. money-saving) objective and an effectiveness one (i.e. delivering better service for customers).

I was brought in to advise on how we could deliver better customer service through addressing what customer service representatives (CSRs) were saying on the phone. That is, using behavioural nudges to improve the quality of outcomes for both customers (more successfully answering their reason for calling, such as making a balance transfer) and the bank (reducing the duration of calls so they could handle more, as well as encouraging customers to take up online and paperless offerings).34

One example: our analysis found a surprisingly high number of people were failing the mandatory security checks. After listening to calls, we discovered this was because the framing of these checks was very formal, and slightly confrontational. CSRs were in effect saying that if customers could not prove their identity, the bank could (and would) not help. With older customers in particular, this interrogatory approach was causing them undue stress – which has been proven to affect mental availability35 and the ability to recall information. As a result they would frequently panic and get their answers to the mandatory security questions wrong. This lengthened the call, as well as making it unsuccessful and frustrating for the customer.

With a few small tweaks to the wording, we changed the scripts to frame them more positively (e.g. from “if you prove your identity” to “when you prove your identity”)36 and even said to customers that they could “take their time”, to put them more at ease – a counter-intuitive solution. By slowing down the conversation, this would actually reduce the overall length of the call.

It is an example of how behavioural science tells us that how you say something is as important as what you are saying – if not more so.

This was one of multiple interventions (nudges) employed. For practical reasons it was not possible to run a full RCT to isolate each nudge. Instead we ran a controlled pilot where a representative sample of CSRs in the call centre were trained and coached in using these nudges over a 12-week period, and we monitored the outcome of those calls versus the rest of the call centre.

Referring back to our three criteria for a behavioural business from chapter 1: we were using data to build an accurate picture of what worked; we verified it through experimentation; and we had hard data on what was actually happening through data based on behavioural outcomes (the outcome of the phone call). There was a clear, direct link between what our decisions were as a business, and a behavioural outcome.

Over the course of the pilot, there was an 11% reduction in the duration of calls versus the control,37 worth potentially millions of pounds due to the thousands of calls handled every day. Customer satisfaction levels increased, and we could prove overall success in terms of efficiency and effectiveness based on behavioural outcomes. Subsequently the training and process was rolled out to the other 300 CSRs in the call centre.

The value of testing

But, you may be thinking, for my business to make best practice use of insights from behavioural science, does this mean I need to be conducting RCTs every time I want to nudge a behaviour? Do I need a team of behavioural science PhDs conducting longitudinal, statistical analysis on the most effective subject line before I send an email?

Well, as in the example above, experience says no. RCTs are not the only way to experiment, and in the world of business they are often not practical for reasons of time or money. Besides, in the real world, human behaviour is complex. With over 200 different behavioural biases identified in research, the sheer number of influences on our behaviour often make it impossible to isolate the impact of individual nudges.

Richard Shotton, author of The Choice Factory and expert in applied behavioural science in marketing, says the most important thing for experimentation is creating a realistic context, not just sample size. “Context is hugely important, and hugely under-estimated,” he says. “The two reasons for testing are for persuasion and proof, using observed and not claimed data.”

For example, I have seen the exact same phrasing used in scripting for two different call centres achieve two entirely different outcomes. But this simply emphasises the importance of testing in the relevant context. In the example above, had we simply applied the academic principles blindly without piloting first it would have been an unacceptable business risk – and completely unscientific.

Leigh Caldwell, co-founder of The Irrational Agency and author of The Psychology of Price, agrees. “You never know for sure what’s going to work until you test in the field,” he says. “Everyone is influenced by context, because everyone has their own view of the world.”

Even Richard Thaler himself says: “Make your research about the world, not the literature.”38

The difference is that in business, what we do is not subject to peer review, nor do we publish our experiments in academic journals. That would wholly undermine any competitive advantage. As Rory Sutherland,39 author of Alchemy: The Surprising Power of Ideas That Don’t Make Sense, vice-chairman of Ogilvy UK, and the foremost advocate of behavioural science in marketing, puts it: “Let me briefly explain what business and behavioural science have in common. They both do experiments. Apart from that, everything is slightly different.”40

Businesses seek competitive advantage above all else, and, as we have seen, a scientific approach to changing behaviour can drive progress and innovation to deliver that. Whether that is achieved by academically robust experimentation, or simply adopting a mindset of continuous testing and learning, hypothesising and deducing, is largely irrelevant. This also means businesses should not be unduly concerned about the current academic debate about whether certain experimental psychological studies can be reproduced – the so-called replication crisis. Especially when one considers the replication rate of social psychology experiments actually compares favourably with medical disciplines, like oncology.

As Sutherland states: “In science, the dream is to uncover a universal, timeless truth or law. In business, we don’t need to be right in general – we just need to make the best decision for the situation at hand … In business, you don’t need to be ‘right’. You just need to be right enough … Sometimes all you need is to be less wrong than your competitors.”

Or, to put it another way, a business can science the shit out of a problem without needing a PhD. But a business also has to become comfortable with the fact that some of the results may not always be what you expect.

In fact, because behavioural science is grounded in understanding the nonconscious, hidden drivers of behaviour, they almost certainly will not be as expected. Sutherland says this is an inherent (competitive) advantage.

“Every time you test these things you find significant events – not necessarily predictably. But they’re significant enough at the very least to be worth testing. And the gains are monumental … I think there’s a massive sweet spot, because people only test what’s rational. The burden of proof we apply to a rational suggestion is very low. And the burden of proof we apply to an irrational suggestion is very, very high.

“But actually irrational suggestions, if they succeed, are much more valuable, because it’s knowledge you have which might give you one up over your competitors. That is a really valuable insight, whereas merely confirming what you already know is almost worthless.”

The conclusion is clear. “Test counter-intuitive things,” he advocates. “Because your competitors won’t.”

The value of this has been, in my experience, that the budgetary commitments are also much less. Should you hire a traditional management consultancy like McKinsey, for example, to address a problem they may spend hundreds of thousands of pounds developing a report of several hundred pages, that advocates one solution.

Hire a behavioural scientist and you will likely get a ten-page report with ten solutions that you can practically test to prove what works. Behavioural science solutions are as open to small businesses and start-ups as they are to large corporates – and can level the playing field.

McKinsey themselves even acknowledged this critical role in their 2019 review of behavioural science in the corporate world: “Creating an effective nudge unit requires much more than hiring a few experts who understand heuristics and statistics. It’s up to senior management to create the conditions for success by helping to focus the unit, situate it in the organization, celebrate its impact, and hold it to high ethical standards.”41

My interviews with leading behavioural practitioners confirmed they have found great value in an experimental approach – what I like to call ‘test-tube behaviours’.

David Perrott, an established South African practitioner, says an experimental approach actually fosters creativity, rather than the opposite: “Experimentation allows for more creativity, more counter-intuitiveness and innovative techniques, because it is viewed as a test. Because no one’s neck is on the line if it fails.”

In the following parts of the book, we will explore how an understanding of behavioural science, and applying test-tube behaviours, can lead to benefits in key areas of business, and how some leading businesses have effectively applied this knowledge.

Next we will see how removing this stigma of failure through a growth mindset, and an experimental approach to understanding the drivers of human behaviour, has driven the growth of the most successful global businesses in the 21st century. In addition, we will see how it is baked into the culture of these organisations.

28 www.theguardian.com/public-leaders-network/small-business-blog/2014/feb/03/nudge-unit-quiet-revolution-evidence

29 Halpern hypothesised that this is due to large taxpaying businesses viewing themselves as unique, and so what other people do (the essence of social proof) was viewed as irrelevant.

30 Explained overleaf.

31 This is agreed by international treaty.

32 Data from the Bureau of Aircraft Accidents Archives (B3A), a non-government organisation based in Geneva. Worth noting also that the volume of air traffic (i.e. the number of people carried by air) has increased over that period from 331m annually, to nearly 4bn (International Civil Aviation Organization, Civil Aviation Statistics of the World and ICAO staff estimates).

33 Now GoBeyond Partners.

34 In line with Thaler and Sunstein’s objective of going with the grain of behaviour, there is a win-win situation here in that no one likes spending more time on the phone to the bank than necessary, so successfully achieving the objective of the call more quickly benefits both parties. Call duration inversely correlated to customer satisfaction.

35 See the description of availability bias on page 14.

36 This uses the concept of self-efficacy, the behavioural bias that our own belief in our ability to achieve an outcome affects the likelihood of that outcome, to make the customer more likely to successfully complete security.

37 The effect was actually greater in relative terms as the rest of the call centre experienced an increase in call duration over the course of the pilot, for various operational reasons.

38 On Twitter, when asked to give his most important advice to PhD students.

39 Also, in the interests of full disclosure, my former boss.

40 behavioralscientist.org/it-isnt-a-replication-crisis-its-a-replication-opportunity

41 ‘Lessons from the front line of corporate nudging’, McKinsey Quarterly, January 2019.