Panel debate: Big data and the future of social #TheNewNow

01 April 2016 / Jaywing

Social media is changing rapidly. We are going from a world of simply tweeting about our cat and periscoping about our lives, to analysing that information to create more impactful and tailored messages.


Social data provides a new, unsolicited, uncensored and real-time view into the minds of existing and potential customers. But does it hold the key to future success for brands looking to harness big data and understand their customers better?


To answer this question and more, we hosted a panel discussion with some of the most authoritative experts on the topic of big data and the future of social. Our panel included Twitter’s VP for Europe, and three leading academics actively involved in the analysis and visualisation of the big data associated with social media.


Our panellists included:


Bruce Daisley, Vice President, Twitter Europe

Bruce runs Twitter’s business in Europe, was dubbed “one of the most talented people in advertising” by Campaign magazine and voted individual of the year at the Drum Social Buzz Awards.


Professor Ken Benoit, Quantitative Social Research Methods, London School of Economics

Professor Benoit’s current interests focus on big data analysis and text mining, including in social media.


Professor Sophia Yaliraki, Theoretical Chemistry and Professor Maurico Barahona, Biomathematics – both from Imperial Collage Data Science Institute’s Social and Cultural Analytics Lab

Professors Yalikari and Barahona have developed a series of methods that derive interest communities and roles in Twitter based on flows to understand who is talking about what and the various roles they play.


Q. What is the role of social data in predicting upcoming events?


Analysis of social data can reveal hidden patterns and provide powerful insights. The notion of changing your brand and operations based on what people are thinking and saying is a powerful one.


We can see at a macro-level when people are doing things in their lives. For example, people are talking about shopping on Sunday evenings more than any other time of the week.”  Bruce Daisley


Does this present the case for longer opening hours on Sundays? Does it suggest that’s when people are going online shopping, or are they turning their attention to retail therapy when life is at its quietest? Imagine what you could do with this insight to improve marketing, would you send targeted offers at the time people are most receptive? Or share reasons to shop with you?


“We also discovered the likely hotspots of a flu outbreak. The tweets proved a pretty accurate signal to where outbreaks were. In fact, tweets were ahead of searches.Bruce Daisley


What’s interesting here, is that you can predict an outbreak in a particular area as people start to say they feel a bit under the weather. So what can you do with this information? Improve NHS staffing, for example in hospitals (for the vulnerable), or in GP surgeries? Or promote ideas or products to make people feel better?


These kinds of interesting findings have the potential to flip old ideas on their head, and it shows how valuable this kind of data can be as a new way of looking at things.


Q. Due to the increase in emoji, Gif and video updates, are we getting to point where identifying sentiment trends will be more challenging?


The way businesses and customers talk has changed fundamentally. It’s not just about text-based language anymore. But our panel believes insight isn’t just gleaned from what you post – but who you follow and retweet.


People don’t realise how much they declare about themselves through who they follow and who they retweet”. Professor Yaliraki


What’s more, text updates can be difficult to mine due to our good old British sarcasm. But Bruce Daisley explains how emoji’s are far less ambiguous.


Sentiment analysis is hard, but one thing that has helped is emojis. The crying emoji was retweeted over 5 million times. It is far less ambiguous compared to sarcasm.” Bruce Daisley


Q. Can we work out the different personas people wish to portray and who they really are?


The panel debated whether it is possible to identify the persona someone is portraying on social media… and if it really matters? Professor Barahona talked about how you can get an idea of someone’s persona based on who they follow.


Studies show that when people consistently follow similar people it is a good indicator of their persona.” Professor Barahona


Professor Benoit, went on to explain how it might not even matter what persona you portray on social media because we do portray different personas throughout our lives depending on whether we’re at work, with close friends and family or acquaintances.


What you say isn’t necessarily what you mean. Politicians will always show their public persona. But that’s the persona we care about. So it might not really matter if we can identify personas or not.” Professor Benoit


As marketers, do we agree it doesn’t matter?


Q. Do we have the ability to identify genuine influencers?


Professor Yaliraki talked about discoveries with influencers and how many influencers sit in different communities and connect them.


People with the most influence connect communities that wouldn’t normally talk. If you want to innovate you want to know who that is!Professor Yaliraki


Bruce Daisley and Professor Barahona, talked about how people are looking for that human-to-human connection.


More than half of all tweets consumed are created by 2-3% of users – these users are predominantly journalists and celebrities. But the challenge is looking at where influence resides. All the individual members of One Direction have more followers than @onedirection. In business, the most followed accounts are the journalists rather than the actual newspaper. People are of course more inclined to follow a person than a brand!” Bruce Daisley


The idea of people being more inclined to follow people rather than brands may give more credence to the increased penchant for ‘characters’ in brand advertising, for example Vinnie from Fox’s biscuits and Aleksandr from Comparethemarket. These characters are more three-dimensional with personalities in their own right, which has helped develop closer relationships with customers.


People are not on social media to listen to an organisation. What’s interesting is how people interact with people!Professor Barahona


So what does this mean for brands? Brands know they have to respond to customer service queries and usually do this quite ‘humanly’, insofar as they usually identify as a human voice. But how do we speak more humanly in our brand messaging?  


Final thoughts

This intimate, real-time look into consumers’ lives, ideas and behaviours presents an unprecedented opportunity for smart marketers to get ahead of the game and understand consumers as individuals with unique needs. Even now we can see social data completely changing the way businesses approach certain things, as our panellists illustrated.


We are hugely excited about the potential of social data to serve marketers, but to do big data and sentiment analysis right requires talented mathematicians.


What’s more, while social media may have started with the marketing team, now it’s spreading and becoming something everyone can use. It doesn’t matter what department you’re in, social data is valuable to everyone.


Social data is revealing patterns of routine behaviour and things that can’t be predicted. Our beliefs are challenged and proven by data.” Bruce Daisley