This article is a brief summary of the contents of the course “People Analytics: Transforming Management with Behavioral Data”, led by Ben Waber and Sandy Pentland at the MIT Media Lab in July 24 and 25, 2017. The content of the article is based on their explanation and bibliographic references.
People Analytics is a discipline of Social Physics that aims at understanding human behavior through data frequently used for organizational development. People Analytics is a human centric discipline different from Human Resources Analytics, whose ultimate objective is more focused on the employee’s performance.
Leveraging a variety of data sources, People Analytics is used as a multidisciplinary foundational management method based on the scientific method to generate and test hypotheses, and solve business problems. Contrary to traditional organizational development, People Analytics relies on data to make observations, think of interesting questions, formulate hypotheses, develop testable predictions, gather data to test predictions, refine, alter expand or reject hypotheses, and develop general theories about the future of the organization.
Analytical Framework and Data Sources
People Analytics leverages a great variety of data sources, such as facilities data (HVAC, Elevator Usage, Power consumption, seat sensors), people data (sensor badges, RFID badges, consumer wearables, cell phones, wi-fi tracking); behavioral data (demographics, team, tenure, performance, contextual qualitative knowledge); and digital data (email, calendar, chat, project management software, phone, computer usage).
Relying on social network analysis, People Analytics aims at discovering how people are connected to each other and how that interaction influences their productivity as a company. Local metrics measure cohesion (how tight knit the network is), visibility (number of unique participants in each interaction) and exploration (how much members of one team interact with other teams). Global metrics include clustering (overall network cohesion); centrality (how nodes in the network connect to each other) and diameter (distance between nodes).
To solve privacy issues, people analytics is best applicable under an opt-in frameworks that are clear, concise and easy to understand, and with complex aggregation methods, such as hashing and salting, that ensure that no individualized data is shared.
A study on Call Centers demonstrated that cohesion between employees predicted productivity and that facilitating the employee interaction during breaks and lunchtimes could reduce stress and improve the level of engagement.
A study on retail banks that analyzed how employees communicated between each other, demonstrated that active employee engagement policies improved interaction between employees could increase productivity by an 11%.
The case of Yahoo remote work policy, demonstrated that co-located employees could increase communication between employees and increase by $150M per year the total revenue of the company.
The case of EToro highlights the value of social learning and how closed and isolated communities generate losses and represent a risk.
- Pentland, Sandy. Social Physics
- Waber, Ben. People Analytics
- Watts, Duncan. “Small Worlds: The Dynamics of Networks between Order and Randomness.” Princeton University Press, 2003.