SCIENTIFIC DECISION-MAKING

We help apply the scientific method to the process of decision-making, which is less challenging with applicable data. Through comprehensive data inquiry, we surface assumptions and underlying context. This helps us determine hypotheses, which inform experimentation. We then execute those real-world experiments (conditioning a business on x, y variable), to statistically evaluate what is true and directly test assumptions and competing explanations or hypotheses. We help you use statistics and modeling to determine, as a function of chance, what decision is “best.”

This process is perhaps best described with an example:

In 2015 the U.S. operating unit of a large global pharmaceutical company wanted to enact a behavioral and cultural change in response to environmental and market changes. Typically, Human Resources would select "change agents" (individuals with a perceived ability to promote change amongst peers) and task them with promoting the cultural and behavioral change. 
We worked with this organization to understand the underlying problem, and after performing initial experimentation, we suggested leveraging social contagion, a process in which behaviors and attitudes get transmitted from person to person like a virus, to try and accomplish the much-needed organizational change.
We ended up testing two approaches to selecting change agents:
  1. The organization's traditional methodology, using HR-selected change agents
  2. Scientific methodology, in which we selected change agents using a network analysis to identify influential employees based on their structural location in the organizational email network.
The outcome?
Our approach was 3x more effective and saved the organization an estimated 5 million in lost opportunity cost. (This work also received an award at the 2016 Wharton People Analytics Conference. Please contact us if you're interested in reading the abstract.)

This example shows what happens when you apply the practice of scientific decision-making (mathematics and statistics) to affect large-scale behavioral and cultural change within an organization. Additionally, our methodology was causally unambiguous: we understand exactly why it worked, and can do it again, in other contexts, for other clients, with the same expectation of success.