This is the third post in our series on using data to improve your IT operations. The second post on making your metrics meaningful is posted here.

In tech, there’s no shortage of data. It can help you manage your systems and teams better, but getting the most out of the data available to you is about more than just getting the numbers. You need to have a culture that pushes to make and measure the success of decisions with data. At least in theory, relying on data lets managers not only make good decisions with lower risk but also have the confidence to make them quickly. It also provides a way to know whether a particular decision paid off.

Actually creating such a culture though, is more complicated than simply declaring that from now on your operations will be data driven. What data do you measure? How do you respond to it? And what steps do you take to get your team to buy into the whole idea in the first place?

Following are suggestions for how to implement a sustainable data-driven company culture–one that the team will buy into and is self-reinforcing–along with some pitfalls to watch out for:

1. Determine what to measure. The purpose here is to use data to make your business more nimble and confident. Managers need to understand the priorities of company executives and select the metrics that support those objectives. If you measure everything and treat it all as equally important, you’ll become bogged down in irrelevant minutiae. To get you started, here are four key operational metrics you should be tracking.

2. Relate the metrics to both the specific business goals and the team’s role in achieving them. Mean Time to Repair (MTTR) is a great high-level performance indicator, but it’s not always easily actionable by the team. Mean Time to Acknowledge an incident (MTTA) is a component of MTTR and is usually more actionable. Track both key performance indicators, as well as the metrics that lead or contribute so that you understand how the team’s work contributes to the overall goal.

3. Democratize the information. In a data-driven culture, everyone’s a data analyst. But to do that, the data has to be made more transparent than most companies are in the habit of, and the team needs the tools to access it. Make sure everyone has some kind of dashboard or other window into the data and that they understand (through training, if necessary) how to extract insights from it. Our newest feature, Advanced Analytics, is a great way to share data with everyone on your team.

4. Empower the team to speak up and take action. Everyone should feel free both to propose their own insights and suggest actions and to question the proposals of others, including upper management. “Do you have data to back that up?” should be a question that no one is afraid to ask (and everyone is prepared to answer).

5. Never stop testing. Before you start measuring, you don’t know everything you don’t know. New questions will arise that you may not already have a means to answer. Be prepared to test to get new data—which means being prepared to be surprised.

6. Act on the data. Nothing is more discouraging than to announce with great fanfare that “we are a data-driven organization,” only to have the team watch data-supported ideas languish or, worse, decisions still be made on the basis of what the boss thinks. A data-driven culture actually has to be driven by the data.

There are also several obstacles to becoming a true data-driven culture. Don’t make these mistakes:

1. Don’t get stuck in the past. Metrics, by their nature, reflect things that have already happened, and it’s easy to get sucked into spending a lot of time dissecting reports, discussing reasons, or assigning blame. What happened, happened—now what does it tell you about what to do next?

2. Don’t just focus on the numbers. It’s easier to manage to numbers than goals, and metrics will incentivize people to try to “work to the test.” Remember, the metrics are a means to an end—keep their relation to your business goals front and center.

3. Don’t get paralyzed. What you want out of metrics are insights, and simply measuring more things won’t necessarily help. Avoid the “analysis paralysis” that can come with having more information than you need.

Finally, a data-driven culture is a feedback loop. Top-performing Operations teams typically discuss what their data shows on a weekly basis. You have transparent access to data and a crew trained to interpret it. Close the loop by reporting the results of the actions you took, empowering everyone to start the process again.