The Future’s Dark

Research we are told is changing: from a business that asks questions to a business that listens to what people are saying. Why lure people into focus groups and surveys when they are talking about brands and services spontaneously, in public, without asking for any incentive? Why bother to construct artificial market research conversations when they are talking naturally in the public domain?

Dark social is one short answer.

This idea, which we encountered in an engrossing conversation with Paul Armstrong, was first used by Atlantic journalist Alexis Madrigal.[1] It refers to all the digital conversations that happen in private: our IM’s and emails, the 18 billion WhatsApp messages and 19 billion texts sent every day. They take place out of sight, influencing us minute by minute, but they remain unobservable to anyone except the data owner, or the NASA. They are certainly not captured or modeled by any of the fashionable social media metrics.

Does this matter? Well, Madrigal’s Atlantic article suggests it does. He and the analytics company Chartbeat conducted an experiment that found that 69% of social referrals to websites came not through Facebook, Twitter or other social networks, but through ‘dark sources’. They reached this startling conclusion by looking at the number of people who landed on individual story pages without being referred by social networks. As Madrigal points out, it seems extremely unlikely these people had memorised or guessed the exact URLs they would have needed to type[2]. They had almost certainly clicked a link they had been sent, and yet the sender, the conversation, the process of recommendation remained invisible.

This invites us to pause for a long beat and consider what we actually learn when we ‘monitor social media’? This monitoring, it would seem, may be showing less than half the picture. Furthermore, which fragment of the picture is it showing? Think for a moment about what you say and show on Twitter, Facebook, Linked In, Tumblr; then think about the contents of your texts, IM’s, WhatsApp messages….Which is the real you?

The social psychologist Erving Goffman explores this in his model of front-stage versus back-stage behaviour – the behaviour people present and perform consciously to an audience and the behaviour people exhibit when they feel they are in a more restricted, trusted, social sphere. He captures this difference rather brilliantly with a quote from George Orwell, observing a waiter slipping from kitchen to public dining room: “As he passes the door a sudden change comes over him. The set of his shoulders alters; all the dirt and hurry and irritation have dropped off in an instant. He glides over the carpet, with a solemn air”.

It would be an oversimplification to say that people’s private conversations are necessarily more authentic. But they are different. And if social media analysis is only monitoring people’s ‘front stage’ behaviour, it is only revealing one side of who they are, only one glimpse of how they discuss and discover products, services and content.

How, then, do researchers respond to this problem? There is of course no simple solution, but there are five thoughts that strike us…

Firstly, we should recognise that the boundaries of ‘dark’ and ‘light’ social are liable to change, rapidly at times. Recently the WhatsApp founders have been insisting that they intend to maintain the privacy of comments made on the platform. However, there is already speculation that this will change under Facebook’s supervision. Conversely new innovations, ranging from Secret to Blackphone are encouraging people to take more of their conversations out of the light, bring public social networking deeper underground.  Meanwhile the growing range of ‘anti-social media’ aggregators, such as Flipboard and Zite are allowing people to find content more privately, outside the bright lights of public recommendations[3].

Secondly, we need to think carefully about how people present themselves on the different networks we are able to monitor – and about how this can vary by audience. For instance, Jean Paul Edwards notes that older users tend to use Twitter as a more open medium – good for interacting with strangers – and use Facebook for closer ties. Twitter is more ‘front-stage’, Facebook more ‘back-stage’ for older users. This pattern, though, is often reversed among teenagers: where Twitter is used more for restricted conversations with closer friends, whilst Facebook is a more public medium.

Third, we can start to monitor and model which kinds of products, services and brands are more likely to be spread through public social forums and which through ‘dark’ social media. Using the same methods Madrigal employed, we can analyse our web traffic to discern how many people are being referred by social media and how many come through other sources. The temptation is often to put resources into the public buzz of Twitter or Tumblr, but in certain categories private recommendations are more likely to get results. It will be increasingly valuable for clients to be able to map these differences.

A further consideration is the ethics of all this. As the importance of dark social becomes more widely recognised, we need to be very careful about how we analyse it. There is a huge difference between the approach taken by Madrigal – which analyses dark social indirectly, through what public media doesn’t tell you – and a direct attempt to analyse messages that people have sent in private. Market research relies on having open, honest conversations with consumers. We will quickly lose trust if we ignore this.

Our last reflection is a broader one and concerns the way we respond to the sudden abundance of data that researchers have been granted. It is tempting to look inward into the pool of buzz metrics, return path data, self-tracking and so forth. But the patterns we find there can be misleading[4]. In our experience, the more data we get the more we value stepping back from it: the more we value the human point of view.  We may partly be able to tell a lot about what people think of a service or brand by studying buzz metrics, but the full picture means meeting people, meeting them front-stage, back-stage, in real life.

[1] here

[2] Who would ever type something like this into a search engine?

[3] See Josh Davis’ critique of Madrigal’s original article,  which argues that these aggregators may account for an increasing amount of the so-called ‘dark’ recommendations.

[4] For two contrasting examples of an over-reliance on data versus human judgement see here and here

Tags: , , , ,