Big Boulder: Data Science at LinkedIn

An interview with LinkedIn data scientist, Yael Garten, and LinkedIn business strategist, Jennifer Weedn, about data science at LinkedIn and the applications of LinkedIn data.

Yael Garten & Jennifer Weedn of LinkedIn

LinkedIn is the largest professional network out there where users can find compelling products, look for a new job, and connect with other professionals. Founded in 2003, LinkedIn gets 2 new sign ups per second, 4 billion people searches annually, and supports international professionals in more languages everyday. When it comes down to information, LinkedIn is  “Disney World for data nerds” as Chris Moody put it. And with data like “percent users who are decision makers” (42%!!!), LinkedIn has a jackpot at their fingertips. Amazing data sets of education level, company information, and number of companies worked for are just some of the pieces one can pull from LinkedIn. For example: What happens when someone senior leaves a company? Where do they go? This is all data that LinkedIn has a wealth of.

Yael Garten delved into the different areas of information that LinkedIn gathers and how they use it. As Yael put, normalized data is information. And what do we do with information? We gather knowledge. Within LinkedIn, “products” are defined as people, pages, groups, jobs, and more. At LinkedIn, they have a saying “If you can’t meaure it, you can’t fix it,” and this is how LinkedIn operates. To understand the data is to understand how users engage with LinkedIn’s services and products and how they can make products better for users. According to Yael, every decision is data driven, and with good reason.

To define a data scientist is to incorporate many different skill sets:

Data Scientist = curiosity + intuition+ data gathering + standardization + statistics + modeling + visualization + communication

Data Scientists use LinkedIn data in 3 ways:

  1. Products: LinkedIn’s newest product is “Skills”. Launched last year, it is a dictionary that allows for users to search by professional skills and qualifications to find new employees, see what’s current, and more. For example, the skill “Hadoop” has grown 43% year over year. Someone searching for “Hadoop” would also come across related skills. This new sections allows for really collaborative filtering, and huge insights for LinkedIn users and the company as a whole.
  2. Insights: When can look at growing and shrinking industries with LinkedIn. Even the government has used LinkedIn data for speeches on growing job industries. Data  around growing and shrinking in a geographic area can be found and used in practical ways. A great example of insights derived from this data: Where did people in the finance industry go after the 2008 financial crisis? A surprising answer, but they just redistributed into other financial institutions.
  3. Wisdom: With the wealth of data, it can be used to drive business. What is the value of an action that a user takes on the site? What early behavior on the site is predictive of future engagement? What is the value of a user? Does mobile usage impact site engagement? Mobile usage can cause actions that impact web engagement. LinkedIn uses this to understand implications for the product and the business, because wisdom is the ultimate goal of the data science team, and it is an art.

Jennifer Yang Weedn of LinkedIn’s Business Development launched into brand engagement and how brands are finding a presence on LinkedIn. Their sweet spot for brands is with B2B marketers. Jennifer says there are 5 steps to engagement on LinkedIn:

  1. Establish a presence on LinkedIn, on a community page. A brand’s company page is their record on LinkedIn.
  2. Attract followers organically or by paid sources
  3. Have a conversation with followers using targeting status updates.  Brands can send and target specific updates to certain types of followers who would most be interested in them.
  4. Drive amplification of engagement by followers sharing a brand’s updates and posts.
  5. Analyze follower base, and slice and dice the data to fit the brands’ needs.

There are 2 million plus companies on LinkedIn, and many have seen success on the site. Phillips, for example, increased engagement by 106% using targeted status updates. Jennifer says 70% of users follow or would follow companies on LinkedIn. Users follow companies for different reasons as well; some looking for job info, but more are following because they want insights from companies and content to help them make better professional decisions. From a data point of view: 60% of users want industry insights, 53% are looking for company news, and 43% are looking for products and services.

Brands can also have a presence in Groups. There are about one million groups on LinkedIn, and brands play a role in many. They can either be mentioned in the context of the conversations in groups, or they can sponsor groups on LinkedIn. GE was able to use groups to leverage their resources and become thought leaders in the space. By posting an infographic about how people should navigate their career path, they positioned themselves as thought leaders and could engage their customers better.

Data science at LinkedIn is taken very seriously, and with all of its uses, it should be. The valuable insights and wisdom gained from data are just beginning to show their uses, with many more possibilities on the rise.

Big Boulder is the world’s first social data conference. Follow along at #BigBoulder, on the blog under Big BoulderBig Boulder on Storify and on Gnip’s Facebook page.