Thursday 2 September 2010

Mark Earls - "Herd": the Da Vinci Code of marketing?

Mark Earls is clearly in the wrong profession. He's an ad man by trade, but Herd: How to change mass behaviour by harnessing our true nature lurches between a vast array of subjects covering all types of popular science yet is so tangential that it rarely mentions any direct applications to marketing and market research. Instead, Earls has created a tome to make you go and think for yourself, and explore topics in greater depth.

It's a beautifully printed volume - nice paper, nice typeface and a bright pink cover of the sort that makes people on the tube squint to see what you're reading (and when they see the "how to change mass behaviour" title combined with a suitably megalomaniac look in your eyes, you'll get a bit of extra standing room, I promise you). The writing style is very much in the catchy mould of the advertising professional, albeit the author is a planner by trade, not a creative. Short, sharp sentences. With lots of sentences starting with 'with'. And many more starting with 'and'.

I'm going to start at the end, rather than the beginning: the bibliography is one of the highlights of the whole book, and the sceptic might suspect that Earls may have gone to great lengths to deliberately cram in as many offbeat and varied references as possible. From Freud to Popbitch, from Goebbels to Thatcher, at times it feels like Earls is desperate to show his versatility and open-mindedness. It should also be pointed out that Earls pays humble tributes to many writers before him, and is quick to pay credit. Indeed his praise for several books put them straight on my wishlist. One of them is Thomas Kuhn's The Structure of Scientific Revolutions, which, by sheer coincidence, I found my dad (who is neither a marketeer nor a scientific revolutionary, but a psychoanalyst) is currently reading. As an aside, I asked him about repressed memory and the controversies surrounding it; his explanation was almost word-for-word identical to the description in Herd - which certainly added a lot of credibility to the book in my eyes.

Another I did pick up: Micromotives to Macrobehaviour, Nobel Laureate Thomas C Schelling's study of the way group behaviour is affected by tiny fluctuations in individual perceptions or opinions, and the way the decisions of individuals relate to overall crowd behaviour. The examples at times seem so obvious, yet seeing them written down makes you think twice about crowd behaviour. It's absolutely compelling - one of the best I've read in ages. Herd builds on the solid foundations placed by Schelling and sexes it up; he also begins to muse on how these crowd-behaviour phenomena might affect marketing campaigns, although, perhaps wisely, leaves many questions unanswered: what would be the fun in creating loads of mysteries only to clear them all up?

Schelling touches on traffic modelling - something of which I know little, other than it's a fascinating subject and well worth exploring. One fundamental principle is that traffic jams flow backwards with a wave-like motion, as first set out by Lighthill & Whitham in their 1955 paper On kinematic waves II - a theory of traffic flow on long crowded roads (Proc. R. Soc. Lond. Vol 229 pp 312-345) - something which I'm intending to dig out of the British Library at some point, although to get started I have a copy of Robert Banks' Towing icebergs, falling dominoes and other adventures in applied mathematics (I'll report back if I ever become an amateur traffic expert). Hat tip to a Durham maths student called SG Hockey for the links - his dissertation is well worth a read, although it comes with a health warning - if, like me, mere mention of partial differential equations is enough to trigger a heart flutter, then it might be best avoided. He also links to this experiment by Yuki Sugiyama et al which is an experimental verification of, in their words, "a dynamical phenomenon of a many-particle system" of which "in general, such a system drastically changes its macroscopic aspect owing to the collective motion of many interacting particles". A direct analogy of the sort of behaviour Schelling and Earls are talking about! Keep those two chevrons' distance on the motorway, and you'll do your bit to keep traffic flowing as sudden braking slows everything down. It's easier to see on a crowded escalator: if you're in the "fast lane" (left hand side if you're from London!) bunched up close, if the person in front of you stops suddenly, you'll make a sudden stop, as will the people behind seconds, the whole escalator will come to a standstill. Next time you're heading down to the Northern line at Euston in rush hour, try looking behind you to watch the jam flow backwards. It works. Keep a sedate distance, and you might just help speed people's journeys up.

So much for a book review, I hear you say. This deviated from the mainline ages ago. OK, I may be going off at a tangent, but that's probably Herd's greatest strength: for all I might be sceptical of some of its conclusions, and for all I might scoff at the writing style at times, it doesn't half inspire you to think outside the box. I found myself drifting off into daydreams of herd-like behaviour as I was reading. Earls' enthusiasm is infectious. Let's put the traffic modelling to one side for now, however - if anyone knows an expert who can explain it to me properly, please let me know.

[Update, 7 September: I'm now reading Banks' Towing Icebergs which is very interesting if you like maths, and is bringing back rather more vivid memories of differential equations than I'd like to remember; and how could I forget to link to this post comparing traffic modelling to online communities?]

Meanwhile, Mark Earls is able to cover huge amounts of ground in subjects close to his heart. He launches into a discourse on one of hs favourite subjects with gusto in the early pages. a self-confessed amateur primatologist, he explores the human/chimp boundary and concludes that socially, as well as physiologically, we are infinitesimally close: humans are an example of a super-social ape. Chapter 1 doesn't say anything particularly radical. Rather, it sets the scene for Earls' later dramas, a scene with all humans as a naturally social species, with interactions with other people playing a central role in the way we approach all problems and decisions.

The second chapter carries on where the first left off, with helter-skelter, high octane voyages of discovery covering illusions and memory - and where the two meet in the middle. Then, out of the blue on page 72, comes the first killer blow: the claim that attitudes change after behaviour, not before.

This is based on work by psychologist Daniel Kahneman, whose writing sounds fascinating and whose book Judgement and uncertainty: heuristics and biases has shot straight to the upper echelons of my wishlist.

I take issue slightly with the way Earls treated Kahneman's work, although I must stress that I haven't read the original material. Earls seems to sensationalise everything - using Kahneman's "lazy minds" theory to suggest that nobody ever makes decisions for themselves and we're kidding ourselves if we think we do. It's easy enough to understand the point he's making - and it's an important one - but does he need to exaggerate so much? It's a rather tabloid style that turned me off the book somewhat; indeed, my enjoyment of of the book fluctuated as I went through - looking something like this:
How good is Herd? It varies as you go through. (Pictorial representation only!)
Kahneman's model talks of heuristics as mentioned by Nigel Hollis - subconscious, instinctive, even irrational judgements or actions that we make. Perhaps I'm wrong, but the impression I get is that Earls takes a rather defeatist attitude: he seems to suggest that it's not possible to break down and analyse the motivations for these heuristics, let alone take advantage of the heuristics themselves and influence people's decisions in this way. Surely that is one of the great challenges for researchers and marketers: to discover the heuristics at the point of decision-making and then throw a spanner in the works by affecting those unconscious thoughts. Earls, it seems to me, seems to take a rather fatalistic attitude (repeated later in the book) - you can't do anything about it so don't bother trying.

The attitudes change... statement might be controversial, but, as Earls points out, it challenges the awareness-interest-desire-action model of marketing. Decisions might not be made in the frames of reference we assume.

The question is: how can we work out in what ways, and at what times, that decisions are made? The most obvious example of mass behaviour not working out as expected that I can think of in recent months was Cleggmania in the run-up to the general election in May this year. Before the election, I attended a social media summit where respected political commentator (and influential blogger) Paul Waugh proffered the opinion that the result of the election would depend less on social media than on television, as the televised leaders' debates would change more attitudes than anything else.

The first debate took place and Lib Dem leader Nick Clegg was almost universally acknowledged by those who watched the show live to have "won" it. Sure enough, the various opinion polls, with their varied fieldwork dates (some with rolling fieldwork dates; there was much excitement on the UK Polling Report discussing methodologies!) displayed dramatic increases in support levels for the Lib Dems.

But let's consider. Say 10 million people watched the debate on TV and another 5 million saw clips on YouTube or on the news, a total of 15 million who had seen the debate in some form, and not all of those would be eligible to vote. How many genuine floating voters among that lot? Surely not that many. Yet the changes in the opinion polls were dramatic. The Liberals gained up to 10% of the vote in the space of a couple of days in some polls. What was even more interesting was the fact that the bounce wasn't just an immediate thing: the Lib Dem share of support increased as time went on - and stayed high throughout the remaining weeks.

Surely this was a classic example of herd behaviour going on? Had the Twitter campaigns (for example #iagreewithnick) succeeded? Critically, were there people changing their preference who had not actually seen the debate at all, but were reacting to the hype and opinions of others around them? It's often said that voters like to choose a winner so that they can feel like they have contributed something personally to that success; this is why such positive language is used in electioneering. The Lib Dems, masters of the campaign trail thanks to the genius of Chris Rennard, have monopolised the phrase "Winning Here" as a result.

And yet come election day, the herd phenomenon seemed to vanish completely. Pre-debate polls were actually more acurate than post-debate. This seems to have been a dramatic example of the herd effect/word of mouth affecting opinions, yet when it came to the decision-making that really counted, the voters lost their nerve, or else there are other heuristics involved inside the ballot box. My feeling is that Earls' Herd theory needs re-evaluating after this, but also the research industry as a whole: the post-debate polls were largely worthless in predicting the election result, given that they gave a Lib Dem share of up to 30% right up until polling day, which dissolved completely. That's not to say there was necessarily any flaws with the methodologies used at the time... just that people's intentions can be different from their actions (as Earls would agree - it's a fundamental point of his book).

What can researchers learn from this? That intentions and behaviour are very different - that surveys can be a very inaccurate way of predicting future behaviour (what's the disclaimer that you get on investment ads?) - that perhaps word of mouth has its limits. Were the polls useless? Not entirely; they may be able to give a clue to the heuristics involved in making a decision on who to vote for. Besides things like fear, optimism, wanting to contribute to a success story...what could they be? I have no idea, but anyone who works them out accurately could be a person in great demand. I'd love to know what Mark Earls' thoughts are on the mechanisms at work in the weeks preceding the election.

Earls' argument, taken more generally, is that people are not really in control of their own lives and opinions. The key point for research is that what we think we believe in, or what we aspire to believe, or what we believe we do, and the actions we actually take, may not tally up at all. Now researchers have known for ages about the dangers of respondents giving "socially acceptable" answers to questionnaires. earls takes things further by arguing that traditional focus groups can never provide a natural environment in which we operate and interact, and that we need more organic ways of monitoring behaviour - ideally at the point at which those heuristics kick in. Does this simply point to social media research? I'm not so sure. Monitoring naturally occurring conversations - in forums and on Facebook, for example - can give a wealth of opinion data; but as with a focus group, one is reliant on opinions being put forward, however naturally that may be. Perhaps some of the newer, more sophisticated techniques might be the way forward in analysing decision-making processes - are they just hot air though?

Rich use of case studies is one of the best things about Herd. The Milgram Experiment is one terrifying example of just how irrational human behaviour can become when we feel that there is an "accepted" way of thinking. I'd never heard of it before and barely dared to breathe as I read Earls' two page account. I distinctly remember staring at the wall and saying "shit" repeatedly after reading about it. There are other, similarly dramatic (if less horrifying) examples which most "marketing" books won't come within a hundred miles of. For example, how much idle fun can you have with this Mexican Wave generator? (There are other similar modelling simulations here).

Having set the scene, Mark Earls proceeds to lay down his Seven Principles of Herd Marketing. Don't get too excited: this isn't a step-by-step bible on how to double your turnover in a year. Rather, they're a set of rather nebulous ideas, some of which are frustratingly obvious. Instead, you should continue to focus your attention on the game-changing case studies and analogies with which his arguments are made.

It starts unpromisingly. Urinals? C'mon Mark, it doesn't need a laborious analysis of the rules - that's known to anyone who's ever had a few drunken conversations (if you haven't, there's plenty of stuff on the internet). My attention wavered. Following this, however, we get into the real meat of the first chapter (simply entitled Interaction): it's lengthy, but rich in case data and ideas, and convincingly presented. We learn about markets; whether it's betting markets or financial markets, much of their behaviour and volatility is a result not of external factors, but purely the interactions between people concerned. Betting markets are similar - particularly when you start to think about either starting-price betting, or the new betting exchanges such as Betfair, where you take on the market directly. There are comparisons to be made with game theory - another subject which Earls touches on, and Thomas Schelling concedes that his entire book is really about game theory. I was chatting to a hedge fund trading mate of mine the other week, who is keen to learn more about game theory in order to improve his work; if I remember to dig it out of a locker in Holborn, I'll lend it to him - I might point him in the direction of Herd at the same time.

A fascinating discussion on metastability ensues, and how phase transitions can be compared to other social situations (for example crime levels). The pedigree of the theories is impressive: Earls rehashing Phillip Ball rehashing Campbell & Ormerod influenced by Schelling. It's no less entertaining for all that. Thrillingly, there appear to be quite a few comparisons to be made between physical systems and human ones.

I wonder if Earls has ever come across percolation theory, an area of statistical mechanics which has some striking similarities to some of Earls' material. It deals with lattices connected by nodes and the probability of some form of path finding its way through the lattice to an infinite degree. In other words, if each individual bit of mesh in the coffee percolator has a certain probability that coffee will manage to drip through that one bit of mesh, what is the probability that some coffee will make it all the way through? Admittedly that's an oversimplification, but percolation theory, which has applications in geology and materials, as well as, for example, the propagation of forest fires. Networks vary, depending on the number of connections at each node, and the number of dimensions. But the key property is that, for an infinitely large lattice - the simplest model - across a range if individual node probabilities (how porous one "junction" is) the probability of the percolation taking place jumps from zero to one around a critical probability. So there's a critical point above which the information will always find a way through. The similarities with human networks are clear: each node or person or organisation, connected to a certain number of other nodes, has a certain chance of "getting their message across" to the next person. The spread of information greatly increases as the influence or effectiveness of transmission of information at a particular junction reaches a critical level. I wonder if the Mexican Wave model could be predicted using percolation theory? It's a fascinating subject - one I came across at university - and one that merits further reading.

Earls only makes direct reference to market research a few times in the text, but when he does, he tends to be pithy. He cites the example of when a different methodology gave a completely different answer to a descriptive research problem he was involved with on shoe-buying habits. It's easy to see where his scepticism towards traditional research methods comes from: a methodology could conform to all the usual rules on sapling, validity and so on, and still give wildly different from another methodology which was similarly robust. (Could this provide a clue to where the opinion polls went wrong? Could we try a radically different methodology next time?)

Following on from Milgram's experiment, Earls then examines the models of influence: how we are all influenced, by whom, and in what way. He thoughtfully considers various descriptions of the types of people who are "influential" (though the definition of influential remains somewhat shrouded). There is the Opinion Leader approach - where one in fifteen are "social influencers". There is the "early adopters" model. Malcolm Gladwell has his ideas. Earls doesn't conclude in favour of any one approach - or against any for that matter; personally, I think that you can't just define what an influencer looks or sounds or behaves like - it simply depends on the situation. Opinion Leader talk about MPs, CEOs and community leaders. That all seems a bit predictable to me. I'd say it's much more of a social personality thing: in my experience, those people who are natural leaders and natural influencers within social spheres are the most exuberant, most outgoing, funniest, most interesting people. They're the ones who are good at everything - from rugby to pub quizzes - and don't waste a minute of their time lying around on the sofa but are involved in loads of activities (although they always seem to be good at computer games too, ironically). The ones with god jobs and attractive partners (they're probably attractive themselves, too; good genes, I suppose). In other words, the ones you're jealous of. The ones who suggest a trip to the pub and the rest follow. The ones everyone else is jealous, and wants to be like, and wants to copy. Is it possible to pinpoint what sets these people apart? I don't know. This does, however, extend to internet forums, where again there are influencers and leaders; this does not correlate with post counts. It's something I might write a post on at a later date.

The next couple of chapters of Herd fade away very slightly, with a rather forgettable discussions on word-of-mouth marketing - although I wholeheartedly agree that buzz isn't something you can conjure up, and that half of these word of mouth, buzz and social media "conversation" agencies are charlatans. A slightly sanctimonious chapter saying "just be yourself" follows; presumably this was the chapter where Earls convinced his publisher that this was really a marketing book rather than a popular science one, as there are quite a few case studies. Again, however, while a little predictable in its ideas (make good products, be nice and smiley to everyone and everything will be OK) it's written in a gripping style. He defines a "Belief Business" as one which applies its ideals across all forms of its operations. He gives several examples; the most obvious one that  can think of in my experience is The Lexi cinema. Run by volunteers, with all its profits going to charity, it's not just for the bleeding hearts - it's a lovely comfortable interior with great decor, friendly staff, a bar (yes, you can take in your pint), no ads, a personal introduction to the film by one of the friendly staff...all in all, it's easily the best cinema in London (and 10 minutes walk from me to boot). The damage is a tenner a time, but the overall experience is so far removed from the nearest alternative (the multiplex on the North Circular) that it's worth every penny. The Lexi has been cited before as a social media case study (not sure where the article is or who wrote it but suffice to say that their Facebook and Twitter pages are active, conversational, multilateral and give you plenty of reasons to follow them).

Earls is back on compelling form when talking about co-creativity - with countless examples, across platforms and industries (and outside industry!), of group collaboration proving to be more effective than one "genius". His nineteenth century engineering example was particularly strong; more up to date was his discussion of co-creativity in the software industry. Perhaps he missed the two more obvious examples of collaboration in the technological world: APIs, where programs and applications such as Twitter and Google Chrome are opened up, allowing developers to create extensions and tools to really enhance the user experience; and of course the ultimate co-creative project, Wikipedia. There are plenty of academic studies on the wiki phenomenon out there so I won't embarrass myself - but I will point you in the direction of one of the most interesting articles on Wikipedia: about its own accuracy.

Herd finishes on a slightly damp note of don't bother defeatism again (hence the slight tailing off of my enjoyment graph!) but I found myself coming away with more and more examples of group Herd behaviour flooding into my head, and am in serious danger of considering myself a disciple.

One example that always gets to me, although not strictly an exercise in herd behaviour (it's my blog and I'll cry if I want to), is choosing who to sit next to on the bus. If you go upstairs and each two-person seat has exactly one person sitting on do you choose where to sit? Now don't say "random". There will be some sort of reason. Perhaps there's a simple rule you take: the front seat, or the one closest to the stairs. But let's say you don't. OK, if you've got the whole bus to choose from, then certain people rule themselves out straight away: those sporting loud headphones, gross obesity, and cans of special brew are unlikely to have their adjacent space taken away from them early on. But then who? Trying to deconstruct my own subconscious, I think I aim for someone as neutral and bland as possible, someone who is unlikely to make a fuss, someone who keeps their stuff on their side of the seat, someone who won't make a big deal of getting out. I don't think I'm fussed by male or female...although if I decide to sit next to a woman, it can't be the most attractive one there (too obvious), although I'll err on the attractive side of average if possible. It'll probably be someone of nondescript age (ie middle-aged). And although consciously I would never choose on racial grounds, having read Schelling I'd love to know statistically/historically whether my seating habits are biased towards white people. They probably are. Incidentally, as a schoolkid, I always used to get paranoid if I was the last person to be sat next to. In truth, I probably still do.

I could go on all day providing examples. Just the other day I was in Edinburgh at the festival (I've put some reviews up here). I went to see stand-up comic Stephen K Amos - one of the better and more dependable middle-of-the-road comedians in this country. At one point, a bloke got up noisily to go to the toilet; when he was gone, Amos decided to try a little experiment in herd behaviour (or peer pressure as he put it). He briefed us on what to do, and when the chap came back with a newly empty bladder, Amos casually said "now, c'mon folks, let's be honest here. How many of you used to pick your nose and eat it when you were younger?" As one, we put our hands up...and, sure enough, chappie's hand went up with the rest, whereupon he was stitched up royally by Amos, to his embarrassment (and our mirth).

Mark is a constant critic of the traditional focus group - his arguments perhaps point to weaknesses in even the latest trendy research-by-crowdsourcing and online communities (MROCs) - that they are artificially created and therefore can never be a place to watch true, natural interactions between people. But there is a difference between online communities and focus groups and that is timescales. Scientists can create artificial environments very easily - whether it's artificial reefs or contrived forests - and, although nature will take a while to adapt, in time the relationships between creatures and plants will adjust as normal. Hell, Big Brother was a great concept at first, and no matter how much the producers contrive to force increasingly incompatible and decreasingly interesting people who they think will be good for a fight/shag/ratings (delete as appropriate), even in a short space of time true human emotions, relationships and frailties poke out from under the veneer - and thanks to the camera work of Tony Gregory and his team, every glimpse, every pained expression, every faltering relationship will be captured on film (although, of course, the editing reduces it all to fighting, shagging and ratings). If left alone, a garden will start to sprout weeds and brambles and similarly, now matter how artificial an online community is at first, if left alone, the true insight can appear. Weeds and brambles encourage the real money shot back garden wildlife like bees, mice and foxes; perhaps if left to go to weed, online communities can also provide that rich interaction of the type that Mark Earls thinks is so elusive.

My mum (like my dad, also a shrink; breakfast-table conversations could be quite stressful) was talking to me the other day about a conference organised by the Tavistock Institute she attended in Leicester many years ago. Ironically, it's a conference on group behaviour; she told me that what was fascinating was the way that over the course of the week-long conference, groups, factions and schisms grew naturally - which all began with a rather heated discussion between the smokers and non-smokers at the conference. In the space of a few days, delegates had clustered together around natural leader-types, to the extent that cross-group interaction was almost non-existent.

"An indispensible manual for the Web 2.0 era" extols Matthew D'Ancona in the list of endorsements on the back cover. Would I be right in thinking that D'Ancona has missed the point of the book entirely? The "Web 2.0 era" isn't something new, and nothing has changed about the way people interact and behave. The behaviour just manifests itself more clearly. Opinions, trends, and the propagation of content and views can just be traced quantitatively much more easily.

Herd could be described as the Da Vinci Code of marketing. It relies a little too heavily on shock tactics, the writing style is an acquired taste, and it draws some conclusions that might be distinctly dodgy, but damn, it's good fun getting there: you'll gobble it up page after page, and come out on the other side feeling quite liberated, with some pretty major questions in your mind about how human beings work, and whether our communications efforts could all be in vain. One of those rare things that might actually justify the description "essential".

No comments:

Post a Comment