You don’t need that many dashboards. You need a few quality dashboards, with robust, trusted data sources and simple visuals. Maybe one dashboard (one tab!) for every 10 analysts in your company.

One big draw to using dashboards is that they are new and shiny, and they feel like analytics. Dashboarding gives teams and departments the ability to say they’re doing advanced analytics, without actually doing any. Executives seem to eat this story up, because they keep seeing slick KPI-ridden dashboards in their HBR / Gartner / Sloan reports. And if the other guys have NASA control center-style dashboards, we’ve got to have them too, darn it!

This all becomes an even bigger issue when your company develops a culture of “dashboards are data science”. When this happens, attempts to build a team that actually does data science get squashed because “we already have a data science team”. Nevermind that the team only does dashboards - in the mind of the executives, analytics and data science are being done because dashboards are being done.

Dashboards are not analytics in and of themselves. You still need someone to look at the dashboard and find something interesting, and then someone needs to execute on that interesting thing.

Instead of making a dashboard for everything, what is an analyst supposed to do? Fundamentally, visualization tools are meant to provide a way for humans to see trends. However, with high-dimensionality data (most business data), humans are terrible at spotting trends, regardless of how pretty the graphs are or how seasoned the humans are. This is where you need machines to pick out details for you. However, your humans can’t program the machines if they’re busy making dashboards.

Whenever considering making a new dashboard, ask yourself “what will I do with these insights”? Don’t make dashboards just to make them.

What can be done to cure this? Cull the unused dashboards. I guarantee you if you look at your usage reports for your company’s dashboards, you will find a usage ratio that would make Pareto weep.

Analysts should not be making their own dashboards in the same manner they would make Excel workbooks or Jupyter notebooks. This leads to bloat and confusion as to which dashboards are trustworthy and which are not. It also makes for a dashboard avalanche if those analysts leave the company. Take that opportunity to pitch those dashboards after a month of no one using them.

Free your analysts and data scientists from dashboarding, and let them actually analyze data instead.