Download the
Join VisionCritical and Forrester's webinar exploring the playbook of insight-driven business on April 4

How Should Market Research Best React To Social Influence?



By Edward Appleton

I recently read an excellent article in the January 2013 issue of Research Live ( on the role social influence has on our preferences.

The authors – Colin Strong, Paul Ormerod and Alex Bentley – quote an experiment conducted by Microsoft, showing how strongly influenced teenagers music preferences were by a social component, in this instance the ability to see how many times a song had already been downloaded by others. The difference between this group and one that hadn’t had access to any social influence was marked – amongst those that were “influenced” only a few songs were highly popular, the majority of songs received very low ratings. Amongst those who didn’t have access to the number of downloads available, the preference spread was much broader.

The message was clear: having access to others’ opinions had a heavy skew on preference. Mark Earls ( makes similar points in Herd and I’ll Have What She’s Having – we are social animals, indeed “perhaps the most powerful influencer on human behavior is other people” as he states on his website.

What does the above mean for Market Research? Should we never ask individuals opinions on their own, as essentially opinions are a socially formed construct? Here’s my take.

1. Methodologies should continually improve in Replicating Real-Life Situations

This may appear obvious – that we should aim methodologically to get as close as possible to mimicking a real-life situation or replicating a decision making process. However, take a concept test: ideally, we would need to understand to what extent an individual respondent is more influencer or influenced – and who precisely are those are that influence him or her?

Take any purchasing decision: before beginning on the diagnostic journey, we need to establish per respondent if for a given category this is a decision they (believe they) take on their own or jointly, with somebody else, however loosely? If the latter, then we would need to find out who that person is, and ideally bind them in into the evaluation process. Curious if anyone is aware of such processes going on already, and if so I’d be interested in details on the cost and timing ramifications.

2. Research needs to be an Ongoing Exercise

In a world of ongoing budgetary-pressures, it’s an odd thing to suggest, but essentially we need to re-examine the frequency with which we undertake all studies, including strategic ones like U&As and brand trackers, and ask if it is adequate.

More than ever, we are trying focus our lens on subjects that won’t sit still.

Online influence is happening all the time, more and more people are reading blogs, reviews, ratings – the likelihood is that opinions are increasingly swayed more frequently, and that traditional tenets about the robustness about brand imagery may becoming dated.

3. The pressure to merge Behavioral Data with Attitudinal Measures will increase

Most of us believe we have our own opinions, preferences, beliefs. Much of what the article quoted suggests, as does Behavioral Economics, is that we don’t really have access to our decision making processes. Asking us questions is fine – we have an image of ourselves, that’s an important piece of data – but we increasingly need to look at the behavioral evidence of the same person for a methodological reality check.

How long will it be before sampling methods allow us to easily match behavioral data with attitudinal data, I wonder? The two need merging sooner rather than later.

In conclusion: thought-provoking work coming from the broad area of Behavioral Economics has again made me appreciate the complexity of what we set out to do as professionals – understand and predict behavior – whilst sensitizing to the fragility of some of our tools.  It’s potentially overwhelming: our subjects’ viewpoints are fleeting, contextually derived and influenced highly by people who we probably haven’t included in our datasets. Hmm.

It’s important to remember that however complex the world is, however many cognitive biases we are aware of, Management is looking to MR to help solve problems, essentially to simplify, not highlight potential complexities that we don’t have the answers to.

To continue thriving, Market Research needs to outline the limitations and constraints of the methodology employed, and within that be bold and articulate in stating our views, recommendations and suggested actions.

Curious, as ever, as to others’ views.

Please share...

12 responses to “How Should Market Research Best React To Social Influence?

  1. It’s rare that I outright disagree with you, Edward, but this is going to be one of those times (and Lenny will tell you how shocked he is when he picks himself up off the floor). The influence of social influence is, imho, much over-stated. In reading Earls, the Microsoft study, and lots of Behavioral Economics, what you find is that IN CASES OF UNCERTAINTY social influence can be important. Yes, I’m shouting that because proponents often fail to mention that point. If we look back to social psychology, Social Comparison Theory clearly explicates the conditions under which we seek or are susceptible to information from others, and at the risk of simplification, it’s when we don’t know the stimulus well. For CPG/FMCG, I think your frequency of updating is less necessary systematically, but rather should respond to changes in the market. New products that appear to be gaining some traction could lead shoppers to re-evaluate, and that’s when we need to update, I think. And finally, the reason we can’t tap into System 1 thinking is because it’s System 1 thinking – we’ve already done our homework, handed it in, and had it graded. We can’t tap into System 2 thinking very well because we haven’t finished thinking about something yet.

    So I agree with your 3 needs whole-heartedly, but not why they are important.

  2. I totally comply with Edward. Consumer purchasing decision starts with a want or need and typically end with a product sale. The path a consumer takes to get from point A to B is often greatly influenced by their social circle. Often the path to purchase includes consideration of competitive products, feature and benefit evaluation, recommendations from others, reviews and of course price.According to the IBM/Ipsos study, more than half of the British, German, French and Italian web surfers check social networks before deciding to buy or not a product
    The three needs stated will aid to some extent in improvising present scenarios and ways of market research.

  3. Thanks for the interesting and thoughtful take on the article Edward, I think the three needs look interesting and relevant in this context. I would however argue the case with Steve Needel about social influence only being important in cases of uncertainty.

    Surely it is well documented that social effects work even in conditions of relative certainty, as demonstrated by the classic 1955 experiment by Solomon Asch? Here he recruited respondents to undertake the task of comparing lengths of lines which were displayed on a board. The differences between the lines were very clear and respondents invariably got the right answer. However, when ”confederates” were in the room giving intentionally incorrect answers then much of the time the respondent would conform to the majority. A clear demonstration of the way people conform to social effects even in situations of certainty.

    I do see the point that social influence can be more important in conditions of uncertainty but perhaps the mechanism works both ways – social influence can also create uncertainty.

    There seems to me to be many ways in which ‘copying’ can operate, from straightforward bowing to pressure to less explicit following of fashion. The work we have been doing demonstrates very clearly that in some categories and in some consumer segments social effects are in operation and measurable.

    But I think the broader point Steve raises is an interesting and valid one – what are the circumstances in which social effects are strongest and where are they less important? And why? The legacy of market research and indeed marketing generally has been to look at the individual and not fully consider social effects so the tools and frameworks we have are only now emerging. This has been helped by the explosion of digital data available which better facilitates the exploration of social influence. I think we are probably all agreed it’s early days but interesting times.

  4. @Colin – so I went back to re-read Asch’s studies (as well as the earlier work by Sherif that gave rise to some of Asch’s thinking). First, his experiments were hardly conclusive – he reports only 68% compliance. Second, to get this effect, he needed to have a unanimous majority giving the obvious wrong answer – the number of “errors” drops off radically with even one co-defendant. Third, and for me, most important, is the notion of uncertainty. In his write-up in Cartwright and Zander’s Group Dynamics series, he notes that those who yielded to influence believed their initial judgments were in error – that they now had uncertainty.

    So could you be subject to influence when you are not uncertain – sure. But the idea that we’d go searching for information in the absence of uncertainty goes against everything we know – information theory, Kahnemann’s work, etc. Two theories about social influence (and read advertising here as one source of influence) – the push and the pull. The push approach was what we all grew up with – bombarding you with messages. The pull approach is much newer (with the ubiquitous presence of the internet) that we seek out information and are thereby influenced. Reality is probably a bit of both, which is why we still see a lot of TV advertising research and effectiveness of TV ads (see Joel Rubinsonn’s recent work) and the rabid belief that social media matters to your CPG/FMCG purchases (of which we have little to no proof yet). Show me someone who is searching peer reviews on toilet paper and I’ll show you someone with way too much time on their hands.

  5. An excellent and thought provoking article, Edward. Thank you. I have a couple comments and the first one is a bit technical. Regarding need 1, you ask for methodological approaches that could account for external influences in the purchase decision. I have had good results using attitudinal data to explain conjoint utilities via structural equation models. I think that approach might be extended to include influencers. However, the most obvious approach would be to construct an agent-based model. These models are ideally suited to accommodate the large number of interactions that occur in a social setting. My second comment is a more general one about the role of social influence in purchase decisions. While I agree with Steve that social influence increases with uncertainty, I am convinced that most humans live with some uncertainty in all aspects of their lives, even regarding toilet paper. Humans are complicated and famous for self-doubt. While I would not search peer reviews, if an admired friend told me of their favorite brand, it could possibly affect my purchase decision at retail. There are many reasons for a human to buy toilet paper. Some of those reasons may not involve toilet paper at all. The web is stuffed with review sites, of course, but the local dinner party may hold more sway in many categories. Anecdotally, I have seen in numerous studies that show personal recommendations have tremendous influence on the purchase decision because of the credibility of the source as well as the lack of credibility of so many other sources. There is a tsunami-sized increase in available data. It’s sheer size demands that it be inconsistent. The avalanche of data available today may increase consumer uncertainty, rather than decrease it, making personal influences even more potent than previously.

  6. Edward (and commenters), what an interesting topic and discussion. Edward, you bring up some good perspectives on rethinking the market research process given the greater access to social influence.

    I believe that there is a continuum of social influence – there are categories where social influence is quite robust and others where it is virtually non-existent. As researchers we need to understand where a product/service is on that continuum and whether it can be changed.

    The social media bandwagon is in full force these days. With time, it will settle down to something more normal. Until then, researchers need to keep a finger on the pulse and move deliberately. Thank you Edward for presenting a new perspective.

    @Steve –While I am not well versed in the literature, the notion of uncertainty rings true. Uncertainty is an anxiety state – people want to have made good decisions and will seek out information or influencers to help reduce their anxiety and reinforce their decision. Again, it is dependent upon the type of product (toilet paper, no; a car, yes).

    I think about my own decision-making process (or lack thereof) for many products and services. As a natural born researcher (I’ve always asked questions), I have found the internet a boom to researching any of a number of uncertainties. Professional and layperson opinions, Pinterest type content, statistics, and blogs are all just a few mouse clicks away (sometimes it is overwhelming and tiring).

    As I’ve gone through this vetting process time and again (and experienced various frustrations and decision points), I wonder how much marketers really know about their customers (if they did, my experience would have been better). I’ve come to realize that marketers need to take a much more holistic and in-depth approach to understanding their customer’s approach to the category.

  7. @all – am humbled by the intelligence and erudition of all your comments. I have a question: does anyone have a relatively pragmatic way of determining when social influence is something to worry about? Is it simply a question of category – ie one can classify one’s field of activity, and that’s that? Or does every category have segments within it – user groups essentially – who are more or less likely to be influenced or influencers? I’m curious about the facility (and cost-effectivness) of the Agent based models you mention @paul. And @steve – I just wonder if your push-theory on change is right, waiting until a new entrant effects change, it’s like a politiican just sticking to their guns, losing sense of sentiment, and then getting outflanked.

  8. Edward, as always, good stuff. I think you have very good social influence skills!… I do agree that social influence (aka ‘social proof’ in Cialdini’s great advertising book, Influence) can impact in all choice environments (from certain to uncertain); ie. it’s sliding scale not an excluded middle situation. But more fundamentally, what is certainty anyway in a world of fragmented media. We are losing our societal reference points and common perspectives. The fact that when I do a vanity google search of myself in the news, I find that the top 5 or ten posts are about me. Is this what everyone sees or is that just what Google shows me since I am probably the most relevant thing to me? I don’t really know. What I do know is that search engines and all advanced content providers tweak their algorithms so that you only see stuff that they deem “most relevant” to you. This creates silo’d perspectives. In terms of certainty, it means that people today may be raised according to some fringe media channels that say that Elvis is still alive or that the holocaust never happened. For this reason, and to answer your question, I think segmenting these groups not just in terms of the product category or choice being made, but in terms of who they are and what specific media channels they are consuming is important…One other thought- until we knock out the mechnical turks/bots (another ripe area for Agent-based models) and get our social influence rating systems in order, it’s a brave new world in which marketing and research are just two sides of the same coin. Cheers, Tony

  9. @edward – my point regarding frequency of re-visiting or updating a particular type of research, that one should do it when the market has changed, is based on your assumption of a declining budget. It’s not my ideal approach, but if I can only do something periodically, doing it every January, or every third month or whatever seems less rational to me than doing it when something big is happening. For example, if I’m studying doing an A&U on file folder labels (just pulled this one out of thin air), I’d probably want to do one after Office Max and Office Depot merge – an industry changing event.A big new advertising campaign might be such an industry-changing event. If nothing’s changed in my category for three years – sales are stable, percent buying is stable, shares are stable, no interesting new products or innovations, don’t think I’m going to learn anything new with another A&U – better to put my limited resources elsewhere.

  10. In answer to the point you make Edward about whether there is a straightforward way to asess whether there are social effects the answer is a qualified yes. There are two broad ways this can be done:

    a) As outlined in the article and consistent with the work that others have been doing in the field it is possible to analyse sales data to look at movements over time and use mathematical modelling to identify the underlying patterns. If the data is skewed then ‘copying’ is typically taking place. Well documented by people such as Duncan Watts, Paul Ormerod and Mark Earls. However, this will only let you see what is happening at a category level.

    b) Survey data can be used when combined with modelling to identify the underlying network, or indeed whether there is an underlying network in place. This is more flexible in that it allows us to look at it by segment rather than just category level.

    The reason the answer is qualified is that there are no absolutes – more percentage probabilities – but that’s the complex world we now recognise we inhabit of course.

    And I think the earlier discussion touches on the way in which influence works. My reading of this is that it can often be very subtle – we do not need to be actively ‘pulling’ in order for there to be influence. As mentioned, Behavioural Economics is educating us that the way things are framed can have a massive effect on our preferences and behaviours, we often passively accept influence on our lives.

    1. Just saw this and thought it would be appropriate to add to this discussion:

      “Consumers seek product information based on social standing, research finds.”

      (—When it comes to buying decisions, it’s more about who you know than what you know for some people, according to a Colorado State University researcher, whose work may have a significant impact on how marketers tailor product messages to consumers.”

      Read more at:

  11. @Edward — you asked . . . “I have a question: does anyone have a relatively pragmatic way of determining when social influence is something to worry about? Is it simply a question of category – ie one can classify one’s field of activity, and that’s that? Or does every category have segments within it – user groups essentially – who are more or less likely to be influenced or influencers?”
    It’s a very good question and one which every marketer should be thinking about. However, I don’t know that I have a definitive answer as to how to discern when it is an issue. A marketer should have a good pulse on their product/service and to some extent should have some basic gut feel for the degree of social influence that is/is not present. However, if one wants to quantify, then that will require more regimented methods. Perhaps there is an opportunity for some company to compile social media influence measures by category (and brand).
    @Tony – you stated . . . “But more fundamentally, what is certainty anyway in a world of fragmented media. We are losing our societal reference points and common perspectives. The fact that when I do a vanity google search of myself in the news, I find that the top 5 or ten posts are about me. Is this what everyone sees or is that just what Google shows me since I am probably the most relevant thing to me? I don’t really know. What I do know is that search engines and all advanced content providers tweak their algorithms so that you only see stuff that they deem “most relevant” to you. This creates silo’d perspectives.” I can’t agree with you more. This process of “influenced searches” is something that really bothers me as a researcher. I find it biasing. I want to determine if something is worthy of my viewing, not some algorithm.

Join the conversation