Under the influence

We are social animals. We live in a world where our values, attitudes and activities are influenced by others. And there are many of these others, as trends of urbanisation, globalisation and technology development mean that we now have “far greater knowledge of the choices, decisions and behaviours of other people”¹.

Social influence and more specifically network effects have a huge influence on each of us in many spheres – from the partners we choose and the careers we develop, through to the goods and services we purchase². While this has long been recognised in the social sciences³ it has more recently been conveyed to a wider audience by Paul Ormerod (Positive Linking) and to market research audiences by Mark Earls (Herd). The resulting discussion is generating real focus on the way in which social effects are influencing our consumer behaviours.

The number of choices available to modern consumers is orders of magnitude larger than at any other time in history⁴. With thousands of different choices in almost every category or sub-category – laptops, phones and especially digital information – we are prone to decision fatigue.

With such information overload, how do we cut through to arrive at the right decision? One way to simplify the process was identified by Keynes, who wrote that a good rule of thumb is to copy other people⁵. Copying makes sense in many circumstances. Imagine you are trying to decide where to eat on holiday. The various restaurant menus appear similar, prices are close and each has an attractive appearance. So which do you choose? The busy restaurant, of course. The information of others informs your decision.

Copying is so much a part of normal human behaviour that we are scarcely aware of much of it, but practically all the cultural rules by which we live – how we stand in a lift, speak, dress for work – are learned from other people.

How we make decisions

Much of consumer research focuses on identifying consumer preferences, but research of the past decade has led to increased awareness of social influence on our decisions.

Duncan Watts, a professor of mathematical sociology, now working at Microsoft, used the music downloads market to explore the tension between individual decision making and the influence of social effects⁶. He and colleagues recruited 14,000 consumers from a teen-related website, asking one group to rate a list of previously unheard songs from unknown bands. The participants were then given an option to download a track – an ‘individual’ decision made without reference to others. There was a normal (bell-shaped) distribution of song preferences, with the most popular being around three times as popular as the least.

A second group did the same task, except that they could see the number of times the songs had been downloaded. This social visibility resulted in a complete shift in consumers’ preferences, with a few songs being highly popular and the majority of songs getting much lower ratings. In this scenario the ratio of most to least popular was at least thirty to one. Also, the tracks that were popular when selected individually bore little relationship to the tracks that were selected when consumers could see what had been downloaded by others in their network. So the opinions of others matter.

But do some opinions matter more than others? Ed Keller and Jon Berry propose that ‘The Influentials’ are “everyday consumers who are substantially more likely than average to seek out information and to share ideas, information and recommendations with other people”⁷. Their firm studied tens of thousands of influentials by identifying people highly active in their communities, and found them consistently ahead of the curve; over the last 20 years they began using computers, mobile phones and the internet years before the mainstream. Real-life examples of marketing success using the Influentials approach include a word of mouth campaign by GSK to support sales of Odol-Med3, a gel foam toothpaste. Engagement of influentials was considered key to increasing sales of the product by 15%⁸.

More recent work has called into question the degree to which highly connected individuals are directly responsible for the spread of ideas, attitudes and behaviours. Ormerod and Ellie Evans showed in a theoretical study of the mathematical properties of networks that the main determinant of whether a product spreads is the degree to which those connected to early adopters are open to try new concepts, rather than whether the early adopters are themselves influentials⁹. Watts used email, Twitter and computer modelling to discover that influentials are indeed more effective than average in starting fashions and spreading trends¹⁰, a result that was confirmed in a recent Facebook study as well¹¹. However, these studies find that influentials need to affect not only their own contacts, but also the connections of these contacts in turn in order to explain shifts in behaviours and attitudes within a population. In other words, it is not just the social effect from individual to individual that causes the tipping point, it is the effect of the network as a whole. So while influentials have a role, we need to understand better how social effects can support or detract from the call to action that brands are trying to generate.

Tools for social influence

There is a growing consensus that new, more practical tools for measuring social influence are becoming available. A good example of this approach¹² aims to measure the effect of social influence and identify the type of network in operation. By network, we mean the pattern of social relationships we have with those around us. Amid the endless possible specific networks, most social and economic activity fits into a small number of network categories, as in the chart below (adapted from Bentley, O’Brien and Ormerod, Mind & Society, 2011, and Earls MRS, 2012 ). In this chart, the horizontal axis shows propensity to copy others and the vertical axis shows ability to discriminate between products in a category. For example, a sports car may be at the top end of this axis while a laptop or pension plan may be at the bottom.

Sales data are used in order to place a product or service category along these axes. We can identify when copying is taking place in a category by the distribution of sales data into a skewed distribution. This was illustrated very nicely by Watts’s experiment outlined earlier. Using this framework, sales data can also be used to identify when the choices within categories are easy or hard to discriminate. In hard-to-discriminate categories, such as household cleaning products, we characteristically see movement between products or services within the category.

Much current marketing strategy implicitly assumes that people are all in the north-west corner; they are atomistic – making individual, well-informed decisions without reference to any network. However, data shows that most product categories can be found in the east of the chart, reflecting the prevalence of copying.

In the north-west corner information is widely available and the costs and benefits can be clearly expressed. House buyers fit in this quadrant; literature is thorough and individual participants are knowledgeable and have the time and motivation to consider their choices. The strategy for brands is to make sufficient data available to support decision making. The south-west is where we find multiple poorly differentiated options such as many financial products and laptops. The challenge for brands here is to move the product out of this quadrant by emphasising its differentiated features.

Social effects

The north-east corner is the domain of the influential network; listening to experts or copying from the most successful people is a strategy with a long and profitable tradition, going back to hunter-gatherer society. The challenge is in identifying influentials who, although connected, are not necessarily prestigious or even persuasive.

The types of networks that operate in the south-west quadrant are more diffused, and people copy more indiscriminately – many categories can be found here, especially when propelled by influences such as recommendations, top 10 lists, and ‘most popular’ search results¹³.

The strategy for brands is to generate critical mass for the product and then move it north, positioning it as a more considered purchase.

Movement within the model

Of course categories can change and this approach helps us to understand the nature of this change. We may see that early on a new technology product sits in the north-west corner. It is discovered and adopted on an independent basis. As more people discover and use the product, news spreads and it starts to move into the north-east where it is talked about and discussed. Influencers start to play their role in disseminating news of the products benefits. As the product grows in popularity it is vulnerable to other market entrants and it can then start moving to the south of the model where discrimination becomes harder. By tracking these movement it is possible for brands to start anticipating the way in which the category may be developing and take early steps to avoid pitfalls for their product.

The role for survey methods

This approach to understanding and measuring social effects serves as a highly effective focussing ‘lens’ for marketing efforts. This saves wasted effort. For example, too many brands are chasing influentials in categories that are simply not driven by this dynamic. Once a brand understands the social dynamics of the category, survey data is also needed to help brands navigate the marketing challenges that are thrown up. If sales data map in the north-east corner, survey methods can be used to identify the characteristics of influentials as well as the content and style of their communication.

If sales data for the category are not available, then survey data can be used in combination with computer modelling to assess which of the model’s quadrants represents the best fit. And similarly, this approach can be used where there simply is no sales data – for example, when examining social trends. This was demonstrated to good effect by Ormerod and Greg Wiltshire in a study¹⁴ they did on binge drinking.

Looking forwards

Overall, we believe that this approach leads us to explore entirely different hypotheses. Does the north side of the map typically lend itself to higher-value items? Do time pressures influence the way in which different categories, types of networks and individual needs interact? Similarly do we see different patterns with different segments of consumers? How is the increased use of mobile phones changing the way in which networks potentially operate in traditionally closed retail situations at the ‘point of sway’? These critical questions examine the rich intersection between network analysis and consumer insight, between big data analytics and survey data.

Measurement tools are now starting to develop which provide us with an opportunity to make real progress in this area. Of course our tools need to be appropriate for the task – survey methods, which have the individual as the unit of measurement, need to be supplemented by modelling. Analytics of sales data are proving to be highly valuable but survey data provides the nuanced understanding – a classic application of smart data.

Colin Strong is managing director of the technology division at GfK UK, Paul Ormerod is a partner at Volterra Partners and visiting professor at the University of Durham and Alex Bentley is professor of anthropology and archaeology at the University of Bristol

Reference:

1. Paul Ormerod ( 2012 ), Positive Linking: How Networks Can Revolutionise the World, Faber and Faber
2. For further discussion see Alex Bentley, Mark Earls, Michael O’Brien ( 2011 ), I’ll Have What She’s Having: Mapping Human Behavior, MIT Press
3. Borgatti, S. P. Mehra, A. Brass, D. J. & Labianca, G. ( 2009 ), Network Analysis in the Social Sciences, Science 323:892-95
4. Eric Beinhocker ( 2007 ), The Origin of Wealth: Evolution, Complexity, and the Radical Remaking of Economics, Random House Business
5. John Maynard Keynes ( 1937 ), The General Theory of Employment, Quarterly Review of Economics, 51, 209-223
6. Salganik, Matthew J., Peter Sheridan Dodds and Duncan J. Watts ( 2006 ), Experimental Study of Inequality and Unpredictability in an Artificial Cultural Environment, Science 311 ( 5762 ): 854-56
7. Keller, E. and Berry, J. ( 2003 ), The Influentials: One American in Ten Tells the Other Nine How to Vote, Where to Eat, and What to Buy, Free Press
8. Ben Corah and Katherine Kinnear ( 2012 ), Odol-med3: How a Word-of-Mouth Campaign Helped Toothpaste Win a Place in Everyday Conversations, IPA Effectiveness Awards entry
9. Paul Ormerod and Ellie Evans ( 2011 ), Ex-ante Prediction of Cascade Sizes on Networks of Agents Facing Binary Outcomes arxiv.org/abs/1103.3457
10. D. J. Watts and P. S. Dodds ( 2007 ), Networks, Influence, and Public Opinion Formation, Journal of Consumer Research, 34( 4 ), 441-458
11. Sinan Aral and Dylan Walker ( 2012 ), Identifying Influential and Susceptible Members of Social Networks, Science, 337 ( 6092 ), 337-341
12. R. Alexander Bentley, Michael J. O’Brien & Paul Ormerod ( 2011 ), Quality Versus Mere Popularity, Mind & Society, 10:181-191
13. Alexander Bentley, Mark Earls and Michael J. O’Brien ( 2012 ), Mapping Human Behaviour for Business europeanbusinessreview.com.
14. Paul Ormerod & Greg Wiltshire ( 2011 ), ‘Binge’ Drinking in the UK: A Social Network Phenomenon, Mind & Society