To be blunt: how can you be customer-led if customers don’t know or can’t tell you what they want themselves? From the company angle the Holy Grail is a market game-changer: the automobile to replace the horse, say, or an iPhone, or an iTunes or Spotify. But if as a customer you’ve been thinking in terms of more rapid equine transport, a telephone that’s better at talking on, or a larger collection of CDs, what do you make of a machine with four wheels and a clattering engine, a supercomputer with a touch-screen, or a virtual jukebox hosted in something called ‘the cloud’? Round the other way, how come despite decades of experience, pollsters still find surprises and contradictions in people’s actual behaviour, even when polled the day before an election about how they are going to vote?
The answer to the questions that came back from an intriguing Foundation Forum on 10th June was that human reactions remain more or less as infuriatingly difficult to call as ever. We have a lot more data about why we react as we do. But while that helps us understand the degree of our unpredictability, knowing more accurately the scale of the problem doesn’t make it easier to get business decisions right. To paraphrase Lord Leverhulme, we know that half our knowledge about the way people will jump in any situation is wrong: we just don’t know which half.
‘If you’d told me in 1987 that I’d employ 100 filmmakers in 2015, I’d have wondered why I’d ever do that, or that I’d employ 20 economists’
The challenge for the research industry therefore remains as great as ever. As Ben Page, chief executive of pollster Ipsos MORI, noted, while some of the techniques are remarkably unchanged – companies still knock on doors, do telephone interviews, and send out postal surveys – they have deepened. ‘If you’d told me in 1987 that I’d employ 100 filmmakers in 2015, I’d have wondered why I’d ever do that – or that I’d employ 20 economists’. That reflects more sophisticated attempts to tease meaning out of the information gathered, with a discernible shift of emphasis from asking questions to watching and observing what people do. Today’s key trends, says Page, are speed (clients want reports in hours, not weeks), mobile (location recording, instant selfies at the breakfast table) – and the dawning realisation that since subjects aren’t completely rational, it’s not enough just to record what people think they do.
That means that no one research tool is adequate on its own. ‘Instead what we’re seeing is layering of these different techniques, so clients will be looking at a whole range of different data sources. And as we better understand these things using all the techniques at our disposal we’re getting a much better and richer understanding of human behaviour’, he said. So it’s not a question of either intuition or research, but both/and, and a lot of other things besides. Indeed, intuition remains a powerful force – ‘an Ipsos MORI chief executive wouldn’t be advising anyone to scrap research and just listen to instinct, but it’s amazing how many clients pay millions of pounds to evaluate an ad campaign and then cheerfully ignore the data and go with their gut feel.’
To understand how people work ‘you need research, you need data and you need psychology’
Marc Michaels, the second panellist, agreed that to understand how people work ‘you need research, you need data and you need psychology’. And as someone initially recruited to set up a government direct-marketing unit and then more generally to work on changing behaviour – persuading people to eat more sensibly, give blood, join the army: a tough brief – he is clear that in making research and data actionable, the new findings of psychology and behavioural economics are critical. It’s not that we lie, he says; but as Daniel Kahneman demonstrated in Thinking, Fast and Slow we each have two brains, a System 1, ‘Homer Simpson’ organ for instinctive, holistic, and instant decision-making, and a System 2, ‘Spock’ brain for more analytical, deliberative, demanding thought. There are cognitive biases that undermine strict ‘rationality’ but which also give opportunities to ‘nudge’ people towards certain choices or behaviours by fitting the way they are engaged to the innate predisposition to respond in one direction or another.
Thus there’s a general human tendency to fear losses more than to value gains (‘a bird in the hand is actually worth two point five, sometimes even three, in the bush’). The slacker, Homer Simpson brain will try to get away with answering an easier question than the one asked. Big, complicated issues are often avoided, so ‘chunking’ them down is likely to win a more positive response. As Stanley Milgram’s famous 1960s experiments showed, people respect and obey authority, sometimes to a frightening degree. So, in one celebrated example, the Department of Health and COI recruited Anne Diamond, a respected and well-loved newsreader who had suffered a cot death, to counter the grandmother-sanctioned traditional wisdom of putting infants to sleep on their fronts and persuade young mothers to sleep them instead on their back or side. The result: a reduction in cot deaths of 70 per cent. But doing it needed authority to fight authority.
In a sense, research has gone full circle, observed Clive Humby, co-founder of data-led research business dunnhumby and now Starcount, which uses social media and ‘fan science’ to craft brand influence strategies. ‘Really, what we’re really talking about with data is understanding customers through the things that we observe. We’ve heard about watching them through videos, asking them questions, and obviously looking at what people physically do close-up and the transactions they’ve made using information. It’s gone through a complete revolution’.
Humby is credited with the line that data is the new oil, and he drew two important comparisons. First, the gusher in its raw state has little value, only becoming usable when it is processed into something else. Data is the same: ‘Data is everywhere. I’ve got 147 devices in my house that have their own IP address – smart TVs, a lighting system, computers, phones, all those items are generating data about you all the time. The data on ourselves generated in the last day, exceeds all the data generated in a year 12 months ago. So the real challenge isn’t in collecting data any more, there’s far too much of it – it’s making it useful’. So it’s the algorithm guys, the pattern-identifiers that are the stars, the car designers building on the potential of oil.
‘The real challenge isn’t collecting data any more, there’s far too much of it – it’s making it useful’
Yet solving one problem often just reveals another. Made usable, into petrol for instance, oil becomes volatile. So too with data, which turns not just volatile but nuclear when it collides with privacy. Benign nudging and behaviour-recording with consent are one thing; but what about using your shopping list as a basis for insurance premiums? Or a real example from Tesco: ‘One of the most important correlations we found in terms of data we could have commercially exploited, was the one between £4.99 Chardonnay and condoms. But we never acted on it. The reality is that just because you know, doesn’t mean you should. And that is the dilemma we’re all facing’: Is it cool or is it creepy?
The dilemma can only intensify with the rapidly emerging internet of things, in ways that have barely yet been registered. Suddenly the issue is no longer the quantity of personal information being given away to a faceless corporate. ‘We think about privacy in terms of our big corporate systems and frontline operators who talk to customers’, Humby pointed out. But actually the data is available to developers, app people and potentially everyone in the organisations that has access to it. The people who repair your car know everything about where it has been, how fast it was driven, how long it was parked. How easy would it be for someone to get this and use it for something unforeseen and with bad intent? ‘Once that happens, everyone becomes a possible liability. And we have to really worry about that as leaders in organisations’.
The paradox of research, as with most things human, is that the more we know, the more complicated it gets
The paradox of research, as with most things human, is that the more we know, the more complicated it gets – and a simple scientific synthesis seems as far off as ever. At least for the time being, it’s humans that rule, not algorithms. Noted Page: ‘It’s taken some time, but I think the industry is getting there. One of the things that’s holding us back is that sometimes we’re conservative, and our clients are just as conservative because they’ve been tracking data the same way for 30 years, and it’s consistent and tells them things’. His advice is: relax a bit, be creative, and remember the lesson of Alex Salmond – who on the basis of extremely expensive US analysis through social media knew for certain that the Scots would vote for independence, never mind the opinion polls.
‘People aren’t rational,’ summed up Michaels. ‘When they tell you they want to do something, you may see from the data side that they’re doing something, but you’ve got to think about what is going on there. You can think data, but you need to talk human’.
The Foundation’s thoughts
Four of the most significant points which emerged for us were as follows:
• Collaborative businesses are succeeding because they bring at least three useful characteristics together in a way that reinforce each other. Any new and better business model tends to do this, creating a virtuous circle that is different enough to the incumbents’ for it to be impossible to copy with a simple adjustment:
• To get to a good understanding of what people think, feel, and crucially, do, can take the application of all of the approaches described above. On the evening we talked about triangulation, using market research to understand the landscape, then in the areas of interest conducting deeper exploration. This might use more extensive real-world data, and the vastness is made useful by developing and testing hypotheses based on human understanding that respects the instinctive ways we often act. Another way to describe the process is one of detective work – an overall conundrum to be solved, and lines of enquiry established around the possibilities. Each then explored creatively (what could be going on here?) then challenged to eliminate as much as possible from enquiries using data, further specific research or conducting experiments.
• As Clive reminded us, there is a rear view mirror issue with data. It can tell us what’s happened, and used well it can give us insight into why. But it can’t predict the future. Which might make some of the big data investment going on right now look a bit optimistic
• The vastness of the data we each generate creates real ethical issues that aren’t currently being addressed. It is much easier than we realise to share information on everywhere we’ve been, everyone we’ve spoken to and a fair bit of what’s been exchanged with all sorts of organisations and individuals that we might be wary of if we sat down and thought about it. As we heard, modern cars contain information on where they have been and how they were driven, all easily accessed by your local car dealer, the police or your insurance company ... or anyone who knew how to hack and steal it. We often allow apps to get this kind of information from our mobile phones, because we click ‘allow’ and because the Apple Ts&Cs are, in Clive’s words, longer than Shakespeare’s The Tempest.
• Our human intuition isn’t just at the end of the telescope trained on customers. The users of insight have just the same biases, from the more entertaining ‘I don’t care what the research says, we’re running the ad’, to the more important problems we find with inconvenient market research findings getting short shrift from a leadership team trapped in a world they see from the inside of their business looking out. It can be useful to see the conclusions from insight work as the start of another challenge, giving it the impact it needs to get the organisational response it requires. For example, getting leaders speaking to customers themselves so they create their own stories and beliefs in line with the bigger picture.