11 April 2016 10:43:04 IST

What does your social graph say?

Marketers are gauging your personalities through social analytics. But are they accurate?

At this year’s Adobe Summit, the larger than life annual digital marketing conference in Las Vegas, a big draw was IBM Watson’s booth at the experience zone. It was dishing out personality traits of people based on their social graphs.

Watson is the thirsty supercomputer that guzzles data from all sources and makes sense of it within seconds. It does a lot of things but at this booth, IBM was showcasing Watson’s cognitive analytics ability by figuring out ‘what-kind-of-a-person-are-you’ based on a person’s tweets, the language used and so on.

All I had to do was feed in my Twitter handle and within seconds, Watson threw up my behavioural profile. Much to my chagrin, it painted me as a disagreeable introvert! Watching my chastened face, the IBMer at the stall tried to pacify me, explaining that most people tend to post a lot of rants and negative stuff about brands on twitter, which is why they end up with poor scores on the “agreeable” quotient.

While this was just one microblogging site, Watson has the capability of charting an in-depth social graph using a number of behavioural data sources — Twitter, Facebook updates, transaction records, blogs, and profiles. Basically, it is sniffing at the heels of your entire digital footprint.

It can point out who the customers are, who would more likely propagate brand information, which shoppers are given to using discount coupons and responding favourably to advertising, as well as singling out people who might say negative things about a brand.

Social graphs

Ever since Facebook came up with social graphs a few years ago, there’s been lots of work in this area. But what are social graphs?

To put it simply, when you go on a food review site like Zomato, and if you happen to be signed in through Facebook, the reviews of restaurants posted by your friends will pop up first because Zomato uses Facebook’s open graph.

Social graphs open up the interlinked relationships in the digital world for companies and digital marketers to exploit because chances are, you will trust the review of someone you know more than that of a stranger.

As IBM’s big data evangelist, James Kobielus posts in a blog, “Social graph analysis is most popular, thriving on the gusher of customer intelligence flowing from online communities of all shapes and sizes.”

He points out that social graphs can incorporate big data from not just social media but from B2C communities, B2B supply chains and so on. The graph models that are drawn up, help companies identify the likely behaviours of individuals in the context of groups, influence and other parameters.

How it can be used

Ecommerce companies are the ones who’ve used social graphs well to sell products. These websites integrate dozens of social graphs together to give you recommendations.

For instance, sites like Amazon or Flipkart offer you recommendations of books and products based on your preferences displayed in other sites. The more information you share across the web, the more you digitally handshake people, the richer your social graph will be. There are endless possibilities on what marketers can do today using social graphs.

At the summit, Melissa Lemberg, global partner of IBM Interactive Experience, pointed out that hoteliers could gauge what type of movies or books would appeal to a guest (through social viewing sites like Netflix for the former and Goodreads for the latter) checking into the property. The room can then be prepared accordingly, with preferred selection fed into the TV in the room, or on the bookshelf.

But that’s just the tip of the iceberg. American pharmacy chain CVS has apparently used Watson’s predictive analytics to help estimate which customers are at risk of chronic conditions like diabetes!

Boundaries of intrusion

Sounds a bit creepy, right? Every time you leave a bit of information about yourself on the internet, you are adding to your social graph.

But the question really is about how accurate these conclusions about you are. After all, machine analytics and artificial intelligences can goof up big time, as was seen with Microsoft’s chatbot Tay, that is on Twitter as @TayandYou . Holding real conversations, it put back as replies the racist insults that were fired at it. Microsoft had to bid a hasty retreat and took it offline.

Another case in point is American chain Target. Around three years ago, using its own store’s analytics and based on things a teen bought, Target had eerily figured out that she was pregnant — even before her father did.

Well, the social graph may tell you lots of things about yourself that you never even knew — or wanted others to know!