The Care Stack

Your data team is too busy building dashboards nobody opens

Victor Wilson · · 4 min read

Ask your data team how many dashboards they maintain. Then ask how many of those were opened last month.

The second number is usually much smaller than the first. In most health systems I’ve worked with, a large share of reports get little or no regular use. Each one still has to be kept running. Data feeds change, definitions drift, and someone files a ticket when a number looks off. That upkeep adds up.

Meanwhile, the team says it’s at capacity. New requests wait weeks. The work leaders actually care about sits in a backlog.

These two facts are connected.

How this happens

Nobody sets out to build dashboards that go unused. It happens one reasonable request at a time.

A director needs numbers for a meeting, so the team builds a dashboard. The meeting passes. The director moves on to other things, or leaves. The dashboard stays.

A department wants its own version of a report another department already has, with a slightly different filter. Now there are two to maintain.

A project launches with a set of tracking reports. The project ends. The reports don’t.

No one turns anything off, because turning something off feels risky and building something new feels helpful. So the pile grows, and the team spends more of every week keeping old things alive.

Find what’s used and retire the rest

Most reporting tools keep usage logs. Ask your team to pull them. For every dashboard and scheduled report, you want to know how many people opened it in the last 90 days, and how often.

Then sort the list. You’ll usually see a short set at the top that people open every week, and a long tail that almost nobody touches. The long tail is your opportunity.

Here’s the process I recommend:

  1. Flag every report with no meaningful use in the last 90 days.
  2. Find the owner, if there is one. Send a short note saying the report will be archived on a set date unless they tell you they need it.
  3. On that date, turn off the ones nobody claimed. Archive them so you can bring one back if needed.
  4. Repeat every quarter.

Expect fewer objections than you’d think. Reports that nobody opens rarely have anyone willing to defend them. When someone does speak up, have a short conversation. Sometimes they need the report. Often they need one number from it, and that number can go into something they already use.

Look at duplicates too. If five departments have five versions of the same volume report, agree on one with the right filters built in. That conversation is harder, because it means agreeing on definitions. It’s worth having anyway, since it ends a lot of arguments about whose number is right.

Spend the time on fewer, better things

Once you’ve cleared space, be careful not to fill it the same way.

Put a little friction on new dashboard requests. No committee, just a few questions. What decision will this help you make? How often will you look at it? Who else needs it? If the answers are vague, a one-time pull of the numbers may be all they need.

This matters more than usual this fall. Kaufman Hall’s October report found that about 43% of hospitals are operating at a loss, up from about 33% before COVID. Few health systems can afford a data team that spends its week on reports nobody reads.

Then point the freed-up time at work tied to outcomes. Fix the data behind your most important measures so people trust them. Help one department work through a costly problem from start to finish, like a repeat denial cause or off-contract supply spend. That kind of work takes longer per project than a dashboard. It also pays back far more.

You may hear that AI will make this whole problem go away. Ask a question in plain words, get an answer, no dashboard needed. EHR vendors are building toward that. In March, Epic previewed Agent Factory, a no-code tool for health systems to build their own AI agents. Tools like this can help. But an agent that reads from a number nobody trusts just gives a wrong answer faster. Agreeing on definitions and fixing the data comes first, whether people see the answer on a dashboard or in a chat window.

Some dashboards earn their keep. A small set that people open every week and trust is worth a lot. It’s the long tail that eats your team’s week.

One more thing worth saying plainly: your analysts usually know which reports nobody uses. Many of them would be glad to stop maintaining those reports. They just need a leader to give them permission.

So give it. This week, ask your head of analytics for one list: every dashboard and scheduled report, with the number of people who opened it in the last 90 days. Set a date to retire everything near zero. Then ask what the team will do with the time it gets back, and expect an answer tied to money or operations.

Discussion

    Join the discussion