A KPI (key performance indicator) dashboard is meant to be a comprehensive, digestible view of all the metrics that are important to a business, department, project, or initiative. Some KPI dashboards are simple summaries of a few metrics; some have pages and pages of charts and figures.
KPI dashboards differ from other other reports in that they’re not meant to answer one-off questions or drive specific decisions. Instead, they’re meant to be monitoring tools—not unlike a car dashboard—that alert people when something needs their attention. In this way, KPI dashboards often generate more questions (e.g., why are flights less full, or why are our margins improving) than they directly answer.
A KPI dashboard for an airline, for example, might show total revenue, the daily number of flights and passengers served, the percent of seats that have been sold, fuel cost trends, and a measure of weekly profit margins. Dashboards like these are meant to provide company leaders a survey of how the business is performing. If problems are emerging—say, too many flights are underbooked, or that rising fuel costs hurting margins—KPI dashboards should bring those to light.
Though most people agree with what KPI dashboards are, actual dashboards tend to deviate from this definition in two ways.
First, teams often create too many of them. Data teams are asked a lot of questions by their colleagues; to keep from having to answer those questions over and over again, data teams tend to respond by building people dashboards. Moreover, a lot of BI tools also make it easy to build new dashboards, so they’ve become the default form factor for answering questions. This creates a vicious cycle: People ask questions, data teams respond with a dashboard; the dashboard leads to more questions; people ask for more dashboards to answer them.
Second, a lot of KPI dashboards include too much information. Because KPI dashboards generate a lot of follow-up questions, it’s easy to think that they’re incomplete and should include more information. This leads to a second vicious cycle in which more and more metrics get added to KPI dashboards.
Counterintuitively, more, bigger dashboards create more confusion, not less. People won’t know which dashboards and metrics are truly “key,” and which are meant to address smaller issues. The more metrics that are on a KPI dashboard, the less people pay attention to each one—and with lots of dashboards to choose from, people create personalized pictures of their own corners of the company. As the business changes, dashboards go stale and drift apart from one another. And people come to expect analysts to just crank out dashboards, rather than work collaboratively to figure out how to help people make better decisions.
Nevertheless, KPI dashboards are still important. The solution to the problems above isn’t to get rid of dashboards, but to be disciplined about how you build them:
- KPI dashboards should show what’s everything that’s important to the business (or project, or whatever scope the dashboard has). KPI dashboards are meant to be monitoring tools. If something critical isn’t captured—say, there’s no measure of fuel costs on the airline dashboard—then company leaders might miss important changes that they need to respond to.
- But they shouldn’t show what’s not important. People can only digest so much information, and the point of a KPI dashboard isn’t to answer every question, but to highlight the crucial concerns of the business. Minor metrics, like the percent of passengers who purchase in-flight wifi, distract from more important ones.
- KPI dashboards shouldn’t frequently change. KPI dashboards are valuable because people know how to interpret the numbers they present, and know when things look out of place. If metrics are constantly being added and removed, people will struggle to learn what’s normal and what isn’t.
A KPI (key performance indicator) dashboard is a tool that gives a comprehensive, easy-to-understand view of all the metrics important to a business, department, project, or initiative.
Related terms:Analytics tools , ETL tools , Data warehouse