Unfortunately, getting value from data isn’t just about using it – it’s about how it’s used. If applied incorrectly, data can inhibit productive learning that promotes improvement. So to encourage innovation, agencies should use data to learn, not evaluate.
Why evaluation can hurt innovation
Agencies often use data to evaluate performance. Employees are promoted for hitting quarterly targets, or departments lose funding for not maintaining a certain metric. This might not seem bad – on an individual basis, there are benefits to rewarding people for results and setting an objective accountability standard. Unfortunately, applying this mindset at the agency level can create incentives discouraging staff from engaging productively with data.
Why? When using data primarily to evaluate impact and performance, it pressures staff to set conservative targets, avoid risks, and share wins. It discourages people from discussing data that looks bad, and it can incentivize people to misinterpret or spin data. An evaluation model limits the range of activities employees will try and discourages active learning that leads to better decision-making.
Of course, an innovation culture doesn’t make sense in all contexts. In some lines of work - for example, accountants or surgeons - experimentation might even be a liability. However, most organizations work in dynamic environments. People, norms, and technology change, and organizations must adapt. Cultures that encourage continuous learning will be better suited to work through uncertainty and respond to change. Here are some recommendations.
How to start using data to encourage a learning culture
At the program or portfolio level, set broad north stars. Whether qualitative or quantitative, they should motivate your team, rally people around shared priorities, and guide the bets people choose to work on.
At the project or initiative level, set hypotheses, not targets. At a project’s beginning, teams should document expectations and assumptions that lead them to that prediction. This has two key benefits. First, it helps teams align expectations and remove projects they presume will have limited effects on outcomes. Second, it allows teams to compare results against their hypotheses, which helps them refine assumptions and make smarter bets.
At the leadership level, model good data practices. Leaders’ actions often influence culture more than their words. They should set a good example by documenting their hypotheses for strategic initiatives, acknowledging uncertainty, and engaging with “bad” data to learn.
A hybrid approach: learn to hit targets
For federal agencies, reaching measurable goals is essential, especially considering the amount of data they are required to gather. Agencies may need to hit certain targets to meet regulatory, compliance, or reporting requirements. But while targets are valuable for accountability, they may discourage the risk-taking necessary for innovation and encourage agencies to prescribe targets for their teams.
However, targets don’t need to conflict with learning cultures. After all, teams who learn from data after completing a project are more likely to determine how to deliver results. If your agency is subject to external targets, consider a hybrid approach: treat targets as your north star, but encourage your teams to set hypotheses for their projects. If you’re a leader, acknowledge uncertainty and encourage discussion for both good and bad results.
Data doesn’t have to be a performance judgment or a compliance task. When agencies operationalize data well, they create a culture of innovation that encourages people to learn, try new things, and get creative within their sphere of responsibility. Data can be used to incentivize teams to continually improve. For agencies that aspire to be data-informed, don’t just aim to build processes around data – consider how your approach can contribute to the culture you want to create.