How should companies build behavioral science functions into their organizations? This is a hot topic nowadays, warranting a recent report by McKinsey on how companies are currently setting up behavioral science units.
The Uber Labs team just posted an article on how they did it. Uber and behavioral science? We were skeptical, too. The NYTimes article on how they used psychological principles to get drivers to drive longer is still in our memory.
But from this transparent post from their team, we are encouraged to see that it seems like they have figured out and operationalized a new role at the company. One that uses academic literature to inform product concepts, one that considers ethics, and one that adds the much-needed rigorous quantitative methods to typical heavily qualitative user research.
These are the key components of Uber’s approach:
The behavioral science team takes a consultative approach
We’ve seen this model in many organizations; the key here is for the team to be deeply knowledgeable about the product details (business objectives, constraints, etc.), to be able to act hand-in-hand with the product team vs. be perceived as outsiders. As the article points out, when applying behavioral science, context matters. It is important for the behavioral science team to understand all the details about the specific product context.
They focus on quantitative research
Behavioral science research indicates that we humans are great storytellers. Ask us in a user feedback session to explain our behavior or decision at any given moment, and we’ll give you a full rationalization. Uber appears to recognize this and prioritizes rigorous quantitative analyses to inform feature development.
Everyone speaks the same (behavioral science) language — not just the PMs
It’s important to work across a product team: product management, marketing, UX, engineering, data science. Sure, this seems obvious. But we’ve seen product solution meetings where the designer isn’t in the room, for example. When you’re solving for behavior, you understand that seemingly small details like design or copy really matter, and you keep those people directly involved in the solutioning process.
They take a proactive vs. reactive approach
This is a sign of a truly advanced organization. Typically a behavioral science unit within a company starts off as being reactive, but the successful ones manage to get ahead of the bus and start to drive product strategy and feature development. Part of this transition involves sharing wins and building confidence and trust across the organization.
They have built for scale by democratizing experiments
To be a company that prioritizes experiments and questioning intuition, people working at the company need tools. This means A/B testing tools. For Uber, it meant empowering all teams with the top-notch data science tools built for civilians. e.g., a sample size tool. By democratizing experiments, they allow more people at the company to run tests and they reduce the dependency on the behavioral science team.
Overall, we commend the Uber Labs team not only for doing the hard work to build this rigorous organization, but for sharing this peek under the hood by writing this article. This is the kind of transparency we need more of in the field, and kudos to Uber for taking a step in the right direction.
Taking this further, we’d love to see Uber’s behavioral science team (and other companies) share their ethical positions or detailed mission statements.
Public awareness of the use of behavioral science tactics is already high; the opportunity here is to increase trust and buy-in by sharing the ethical principles the behavioral science teams will stick to. This could take inspiration from The Behavioral Scientist’s Ethics Checklist or Cass Sunstein’s concept of a Bill of Rights for Nudging. Regardless, increased transparency is needed and there is an opportunity for companies to take a leading role.