Social Commerce

5% Sales Boost: Real Research Finds Friends Do Influence Purchases on Social Networks [Report]

Amid all the opinion survey nonsense reporting whether people do or don’t think they are influenced by their social networks when it comes to what they buy, here’s a smart study that looks at behaviour, not opinions.

Research at the Harvard Business School has found that in Korea, where social networking and commerce is more established, that social networks do in fact influence purchases, with an interesting wrinkle – the strength and direction of the influence depends on the usage level and connectedness of a user:

  • For light users of social networks with few connections (48% of users), purchases are unaffected by social network activity
  • For moderate users of social networks with average connections (40%), purchases are influenced by social network interaction, boosting vendor sales for this group by 5%
  • For heavy users of social networks with a high number of connections (12%), purchases are also influenced by social network interaction, but negatively; these users avoid buying what their friends have bought and are talking about, leading to a 14% drop in sales from this group for vendors

Whilst this research looked at Cyworld users in Korea,the results could have interesting implications for f-commerce and Facebook marketing.

  1. Firstly, if your target market is made up mostly of light users of social networks, then f-commerce or Facebook marketing should not be a priority
  2. Secondly, if your target market is made up mostly of moderate users of social networks, then deploy shares, likes, and comment wherever you can – as this group is particularly susceptible to social network influence
  3. Thirdly, if your target market has a significant proportion of heavy users of social networks, then shares, likes and comments are likely to hinder sales, so use alternative ‘direct’ social marketing strategies to sell to this group in/with social networks

Download the Harvard research report, Do Friends Influence Purchases in a Social Network? by Raghuram Iyengar Sangman Han Sunil Gupta here.

The broader point to be made is that the current debate on the utility or futility of f-commerce will be resolved by facts, not opinions; buyer behaviour, not survey results. Why? Because opinion surveys are notoriously lousy at getting to any reliable truth, particularly related to how people think and decide – for at least two reasons.

  • Firstly, most of us have no idea what really influences us or the cognitive process underlying that influence; asking us whether were influenced by our social network is tantamount to asking whether we were influenced by the neural network in the orbital frontal cortex of our brains (the bit responsible for decision-making).  In other words, pointless.
  • Secondly, people lie in surveys, or as researchers more politely call it, they display a social desirability bias (responding to enhance one’s image, rather than report the truth – as they see it). Hence if you ask a shopper if they were influenced by an ad or a social network, then unless they want to present themselves as a gullible sheep, there will be pressure to say no.

Which is why most opinion surveys suck when it comes to studying what really influences us, and why opinion surveys are usually about as reliable as reading tea-leaves – unless you’re studying opinions rather than realities, that is.  So whatever you think about Forrester’s contrarian PR stunt – that social media has negligible influence on sales – don’t base your opinions on opinion survey data.  Look for the behaviour not the opinion.


Chartered psychologist specialising in consumer behaviour, wellbeing and technology. Certified CX professional experienced in Design Thinking. A researcher, writer and speaker, Paul is head of Digital Insight at SYZYGY.

8 Comments Add New Comment

  1. theothermatt says:

    It’s facile to dismiss surveys as a means of gauging the likely impact of a given trend. Their weaknesses are well understood by anyone involved regularly in research of this kind and the Forrester report you refer to does take account of this to some extent. For one, the “opinions” solicited from the participants are based on empirical fact, such as clickthroughs and sales. Opinions, when cast against hard numbers, remain a broadly robust means of assessment.

    In a similar vein, it bears noting that the Harvard study is based on small-sample research from a single context from 2004. That by no means invalidates it, but it does render it rather less emphatic a rebuttal to the Forrester work than you paint it to be here.

    The Harvard research is interesting, of that there’s no doubt. It’s not massively surprising, as it itself obliquely acknowledges in its multiple references to well-known previous research about the dynamics of peer influence. But it does give us an additional perspective that adds to our capacity to estimate where and when social commerce may, or may not, be of any use. That’s an advancement.

    I’d be interested in seeing some research conducted on the likely tipping points between the three ‘classes’ of socially influenced consumer. When and where, for instance, does the middle group reach a saturation point where diminishing returns become a factor? That would seem to be pertinent particularly to your bold assertion that marketeers should “deploy shares, likes, and comment wherever you can ” for this group. Likewise, how does one accurately segment ones own social audience in accordance with these three groups and tailor the extent to which social influencer mechanics are deployed accordingly? And finally, yes, as someone said above, how realistic is it to apply any of these findings and conclusions to “Social Commerce” generally. The strongest influencer will always be what it is your selling and the price at which you’re doing so.

  2. Kevin C says:

    there’s also a point that it remains difficult to assess how good a lot of products are (or even which one to consider in the first place). there are plenty of legitimate review sites that do a pretty good job on more standardized goods (e.g., various technology sites like cnet for computers, arguably yelp for restaurants), but for certain types of products / verticals, there isn’t a killer solution… yet.

    ultimately, seeing a friend buy something (particularly one who’s credible or who you know does their diligence) is validation. obviously, if everyone did that then theoretically, everyone’s subsequent purchase is based on the buying rationale of the first person who bought it. but the world’s a bit more nuanced and complex than that in reality.

    what’s a somewhat interesting dynamic is what happens when you build that kind of “validation” on a site that’s oriented around discovery / curation. is the “lift” from social purchase transparency bigger in that case? in thinking about what my company’s doing, an interesting question that we’ll eventually be testing

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.