Dashboards people actually open

Most business dashboards are dead. Somebody paid for the tool, spent a week building charts, demoed it in a meeting to real enthusiasm, and then nobody opened it again. If you have a dashboard right now, check its view history. If the answer is embarrassing, this post is about why that happened and what the dashboards people actually open have in common.

Why dashboards die

They die from three causes, and it's almost never the software's fault.

Too many charts. The classic failure is the fifteen-chart wall. It happens because building charts is fun and cutting them feels like losing work, so everything makes the page. But a person opening a dashboard has one question in their head, usually "are we okay?" Fifteen charts don't answer that question, they assign homework. People do the homework twice, then stop coming back. A dashboard that tries to show everything communicates nothing.

Stale data. The demo used fresh data. Then the refresh depended on someone exporting a CSV every Monday, that person got busy, and now the dashboard says it's three weeks ago. Here's the brutal part: it only takes one stale visit to kill trust. Once a viewer catches the dashboard being out of date, they go back to asking a human, permanently, because the human is never three weeks stale. A dashboard that isn't refreshed automatically is a countdown to abandonment.

No owner. A dashboard is not a project you finish, it's a small product someone runs. The business changes: a new revenue line appears, a metric gets redefined, a data source moves. With no named owner, the dashboard drifts away from reality one small change at a time until it's decoration. Every dead dashboard we get called about has the same origin story: the person who built it left, or was never really assigned to it in the first place.

What the living ones look like

The dashboards that survive are almost boring. That's the point.

One page. No scrolling, no tabs, no drill-down maze. If it doesn't fit on one screen, it isn't a dashboard yet, it's a report that hasn't made its decisions. The discipline of one page forces the question that matters: what do we actually need to see?

Five numbers, give or take. Pick the handful of numbers that tell you whether the business is healthy this week. For a lot of companies that's something like revenue this month against last, cash position, open orders or jobs, a pipeline number, and one operational canary like overdue invoices or open tickets. Yours will differ. The count shouldn't by much. Every number you add dilutes the attention paid to the others.

Updated live, or close to it. Connected to the source system and refreshing on a schedule, with zero humans in the loop. Not because you need minute-by-minute numbers, but because automatic is the only refresh schedule that survives vacations, turnover, and busy weeks. If a number can't be automated yet, better to leave it off than to include a manually-updated one that goes stale and poisons trust in the rest.

Comparison built in. A number alone means nothing. 47 what? Good 47 or bad 47? Every number needs context: versus last month, versus target, versus this time last year. The best dashboards make good news and bad news visible from across the room, without reading a single label.

An owner with a routine. One named person keeps it truthful, and the dashboard has a standing slot in a real meeting: the Monday huddle opens with it, or the monthly review starts from it. Dashboards that are woven into a ritual get opened. Dashboards that rely on curiosity don't.

The test before you build anything

Before choosing a tool or a chart type, answer two questions. Who opens this, and what decision do they make differently because of it? "The owner opens it Monday morning to decide whether to chase collections or chase sales this week" is a dashboard. "It would be nice to see everything in one place" is a screensaver with expenses. If a chart doesn't change a decision anyone makes, it doesn't earn a spot on the page. Park it in a monthly report instead, where depth belongs.

This is also why the tool matters less than people think. Power BI, Looker Studio, and half a dozen others can all render five numbers on one page. Dashboards don't fail in the charting layer. They fail in the choosing layer and the plumbing layer.

How to know yours is working

Three checks, ninety days after launch. People open it without being reminded, which most tools will show you in usage stats. Meetings reference it ("the dashboard says overdue invoices jumped") instead of relitigating whose spreadsheet is right. And when a number looks wrong, somebody says so within a day, because noticing means they're actually looking.

If you have a dead dashboard, the fix usually isn't a rebuild. It's cutting it to one page, wiring the refresh so no human touches it, and giving it an owner and a meeting slot. That's a shorter project than the one that built it, and it's the kind we do a lot. If you have no dashboard and a standing pile of "can you pull the numbers" emails, same offer, one step earlier.