Saikat Mukherjee is Principal Data Scientist at ShareThis working on analyzing the online social behavioral data of hundreds of millions of users. He is leading the data science efforts on reliably identifying users across devices and building user models for online advertising. Prior to ShareThis, he was at Intuit where he used data mining and machine learning techniques to build innovative offers and recommendation products from small business and consumer transactional data, and analytics for increasing conversion and customer retention in various Intuit products.

He obtained his Ph.D. from Stony Brook University and bachelors from the Indian Institute of Technology and is a frequent speaker in data science conferences.

Here is my interview with him:

Anmol Rajpurohit: Q1. What does ShareThis do? What role does Analytics play in the firm's strategy?

Saikat Mukherjee: ShareThis is a social media company. We provide social tools to publishers so that users can share their favorite content to social networks of their choice. Currently we integrate with more than 100 social channels and 2.4 million publishers. At the same time, we help advertisers buy media and find the right audience for their message. Analytics is extremely critical to what we do with data at ShareThis. Our platform is based on finding the right insights for our publishers and advertisers. That insights can range from, for example, informing publishers about what content is resonating with their users to finding the right users for advertisers at the right time.

AR: Q2. How do you define "Social TV"? What are the popular channels for Social TV and how do they compare against each other?

SM: Social TV is typically defined as online social activity about TV shows. For example, you might be tweeting about the show while watching an episode of ‘Breaking Bad’. In this case, it is real time but it need not be real time activity and could be social activity before or after the show has aired.

Since ShareThis has an extensive coverage of all the major social channels, we are in a great position to measure social TV activity across channels as opposed to analysis on just any particular channel. We found that Facebook and Twitter indeed dominate the social TV conversation in the ShareThis network data but there are also interesting niche players in Pinterest, Reddit, and Tumblr. For example,

while Twitter and Facebook have a lot of activity on sports, drama, and family genres, Pinterest and Tumblr stand out in comedy. Pinterest also has significant social TV activity on music genre shows — maybe an effect of that network being used more by women.

AR: Q3. What kind of user behavior insights can be obtained by tracking Social TV?

SM: We have looked at device usage (i.e. is the activity from an iPhone or a PC, etc.), distribution of TV show genres, demographics, and geographic location of users across social channels in our current analysis. While that itself gave us a lot of interesting insights about user behavior, we can go much further particularly on analyzing the impact of the social activity on other users.

AR: Q4. What are the major challenges in measuring and monitoring the user activities on Social TV?

SM: One of the challenges in doing these kinds of large scale analysis is on using data from different sources - both first party as well as third party. There is the usual challenge on integration as well as trying to figure out how good the data really is and what kind of conclusions can be reliably drawn from it. For social TV in particular, being able to associate a piece of social content to a TV show can be challenging too. For example, the content can refer to the show name or it might refer to the main cast and a particular episode of the show so there are lots of variations.

AR: Q5. What are some of the most unexpected insights that you have obtained through Analytics based on ShareThis data?

SM: Mobile has been a surprise. While it’s commonly known how content consumption is increasingly becoming mobile and how networks such as Facebook and Twitter are successfully riding that wave, it has been surprising to see how other social channels are becoming so mobile-centric. For example, we found that almost 75% of the social TV activity in Pinterest is mobile compared to 55% for Facebook.

Also, we found that while there are more android smartphones than iPhones even the US, but in terms of social activity iPhones are more popular than androids. Goes to say something about the differences between iPhone and android mobile users.

AR: Q6. What advances do you expect in the field of Social Analytics over the next few years?

SM: I think, from a device standpoint, it will be interesting to see how social plays out in mobile apps. From a marketing perspective, I think we are beginning to understand how social affects a user’s path to purchase and I think there will be interesting advances over the next few years in how marketers use this data to reach their audience of choice.

AR: Q7. Based on your extensive research of users' social sharing habits, what recommendations would you give to marketers? What metrics should they focus on?

SM: I think it’s important not to fixate on any one particular social channel but have a mix in the bag. Different social networks tap into different demographics and the usage behavior, in terms of content, devices, location and time, also vary across them. Hence it’s important to target audiences across networks. I think some of the metrics around social are still developing and we might very likely go beyond counting likes to metrics more connected to brand awareness and advertising campaign performance.

AR: Q8. Data Scientist has been termed as the sexiest job of 21st century. Do you agree? What advice would you give to people aspiring a long career in Data Science?

SM: I agree that it’s a great job and very fulfilling!

To be good at it though, I think it’s not enough to be a good scientist. There is certainly the science part but arguably it could be less critical than being able to work with data (data engineering skills), being able to understand an abstract problem and solve it at scale, and communicate with peers and the community. So my advice to aspiring data scientists would be to also work on your soft skills and your engineering skills.

AR: Q9. What are your favorite books or blogs on Data Science?

SM: Again to be a good data scientist, you need to be very aware of your particular industry. I try to read as much as I can of adexchanger, adotas, adage which are blogs/publications in the advertising industry. As far as data science goes, I am a hadoop stack + python + linear models kind of guy so try to catch up as much as I can on books and blogs in these areas.

Related:

• Viewpoint: Social Media Analysis: What is missing
• Aparna Pujar, eBay on Evolution of Behavior Analytics for User Engagement
• Vita Markman, LinkedIn on Discovering Customer Insights through Sentiment Mining