Creating a Data-Driven Culture

Being data driven can help your company ship better products faster. You can cultivate a data driven culture in your company in four stages: stimulate curiosity, quantify past wins, estimate impact of projects in flight, and finally use data to prioritize future efforts.

Whether we acknowledge it or not, people make most decisions based on their “gut” - a mixture of past experiences, anecdote, social norms, and their personal worldview. When companies and product teams are able to thoughtfully introduce data and other forms of quantitative evidence into their decision making process, they can gain new insights, make better decisions, and ultimately outperform teams that primarily make decisions on intuition alone. A “data driven” team is one that consistently incorporate quantitative evidence into their decision making practice.

But creating a data driven culture can feel like an uphill battle - people’s natural tendencies are to make decisions based on gut, and becoming data driven involves learning a new set of skills, putting trust in an a new system, and acknowledging that your gut may be wrong. To counteract this and organically grow a data driven culture, you can couple being data driven with two things that people strive for: success and personal recognition.

The first step is to simply start surfacing data. Spark conversations by presenting interesting statistics about the health of your business or the use of your product. Do some research yourself, and start including data wherever you can, from the water-cooler to 1:1s to presentations. One technique for generating curiosity around data this is publishing an “Interesting Fact of the Day” in a very public place. This can be as high level as your current Alexa rank to as nuanced as the conversion rates of specific funnels. Print these out and tape them on the door to the bathroom, and send them to the team Slack room. Another technique is to hold regular “Data Office Hours,” where people can come by to get questions answered or just learn interesting things about the part of the company they work on. The goal is to start the conversation and get people asking questions about the data. These two techniques will also help you identify gaps in your data collection and querying capabilities, if there are questions that you cannot answer.

Next, start habitually running analysis on changes that the company has recently made, and surface that information. It feels good to ship a feature, it feels really good to ship a feature that increased signs up by 5%, and it feels awesome when the whole team is congratulating you for how awesome it is that you increased conversion rates by 5%! Celebrate these victories. After a few such celebrations, people will come to Data Office Hours wanting to know what the impact of their changes were, and how they can be sure that the other things they’re working on will have similar impact.

Third, for changes that are currently being worked on, start asking people to estimate what they think the effect will be. The goal is not to hold people to specific estimates - first, allow people to get comfortable predicting impact and improving their tactics for doing so. Their first attempts will be wildly off, but over time the team will get more accurate. More importantly, people will build the muscle memory that when a project is being proposed, they should anticipate questions about what metrics it might move, by how much, by what date.

Finally, take this to its logical conclusion and ask that when new ideas are put forward for prioritization, they should have estimated goals. Facilitate the transition by making yourself available to help people do the research and make their estimations as accurate as possible, and continue to celebrate the successes that the team has had with data-driven decision making so far. Usually this step isn’t explicitly necessary, as success with steps 1-3 usually starts getting people in the mode of doing this anyway, but it’s helpful to clearly articulate that data should now be part of the prioritization process.

Cultural change is a gradual process, but if you follow the above steps, you should start seeing people opt-in to incorporating data into their decision making in around 2-3 months. These people are the seed of your data driven culture, and as that decision making process yields success - and that success is celebrated by referencing the quantitative results - the rest of the team will get on board as well.