Every picture tells a story, don’t it? Well, turns out that’s true in the enterprise as much as on our Facebook pages. In this case, the picture is the enterprise graph of the workforce – who interacts with whom, when, in what context. And the story is what the patterns of interactions revealed by the graph may say about employee engagement, influence, and how to better leverage all that to the business’ – and the employees’ — benefit.
When Marie Wallace, IBM analytics strategist, looks at social and collaborative networks and other sources of enterprise communications and channels for business processes, such as CRM systems, “I am interested in the narrative,” she told an audience at the Sentiment Analytics Symposium earlier this month. “There is a lot of information in CRM systems – who met with whom, what industry the client is in, what products were presented. All this is valuable and contributes to the enterprise graph.”
Wallace relies on Hadoop and graph database technology, with network data represented as a property graph. “Property graphs are utterly, totally extensible and flexible,” she said, and “the system gets smarter as you add more data into it.” The enterprise social network data generates triple sets (that John Smith created X Document that was downloaded by Jane Doe) that get pocketed into the graph, for example, as is metadata extracted from relational databases. A general set of algorithms can find a user within the graph and calculate his or her engagement level – activities, reactions, eminence and so on. “We now have a Big Data service with a set of APIs so people can query the enterprise graph,” she set, and then run analytics on those results that can drive applications.
Queries can be infinite, she said, but categories of particular interest to the enterprise may be those that are human resource and line-of-business centric. For instance, querying about employees’ engagement levels can provide insight into their happiness, which can help businesses assess attrition risks, and be proactive about them. It’s been shown, she said, that employees demonstrate less social activity in the months prior to their leaving a company. “If someone in your organization is at risk and leaves, there’s a knockdown effect that can be significant to an organization,” she said.
Or sales leaders can use the graph to discover who might be the individuals — including those you might never even have thought of — that actually can be valuable to help close on an opportunity. “If you don’t know how you closed a similar deal [in the past], how can you know [how to make another one] happen in the future,” she said. “The how question is what the network [graph] tries to answer.” Through the graph users, can take the discovery about John Smith’s document being downloaded by Jane Doe and follow it through to Jane’s sharing it with Susan Jones who shared it with XYZ Corp. to close that million dollar deal, for instance.
Marrying social and CRM interactions and information at IBM, including who was at what meetings, who created what content, and who tagged it, for instance, now can surface in an app recommendations for the sales team about who to bring into a project based on employees’ past experience with similar engagements, or even with the potential customer in a completely different context. “It’s a weighted graph and evidence-based, which is key,” she said. “So all evidence can come into play when we make recommendations.”
Users have to be willing to contribute data to make this all effective, which means, she said, that they have to trust the organization with that data. So you need to focus on the employees first and what they themselves can get out of the enterprise graph, such as gaining an understanding of how useful the content they have contributed has been to the organization and how they compare to their peers, without waiting for the end-of-the-year performance review, or how they can activate their network to achieve a goal like getting another position in the company.
Using the enterprise graph to feed easily understandable engagement dashboards accessible only to the employee himself helps build trust and incentivize contributions. “When you do complex analytics you have to give the user a sense of how you came to that analysis but [all the data] can be very overwhelming,” she said, which is why in the work at IBM she’s broken things down into four KPIs. They include a measure of employee activity, how people respond to that, how people perceive the user, and the quality of the user’s network and his role in it.