In our builds, we have killed and revived enough dashboards to recognise the rot before it spreads. A dashboard rarely dies all at once. It dies the way a wooden boat does: a little water at a time, in places you stop checking.
Below are the four signals we watch for, the diagnosis behind each, and the move we make to save it. None of these are hypothetical. We have seen each of them more times than we can count.
Sign 1: the dashboard is opened on Monday morning, and never again
Open rates have a shape. A healthy dashboard is opened throughout the week, peaking on Tuesday and Wednesday. A dying dashboard is opened only on Monday, briefly, before the weekly meeting. The rest of the week the team has gone back to their old habits.
Diagnosis. The dashboard has become a reporting tool, not a working surface. The team uses it to remember what to say in the meeting, not to do their work.
The fix. Find the one tile that the team uses outside the meeting. Promote it. Add two more tiles that answer the questions they ask their colleagues at 11am on a Wednesday. If we cannot find that one tile, the dashboard was built around the wrong workflow, and the kindest move is to scope a second, smaller dashboard that actually fits the daily work.
Sign 2: the team double-checks the numbers
A user pulls up the dashboard and then opens a spreadsheet to verify the number. They do this once, then twice, and then they stop opening the dashboard at all.
Diagnosis. The dashboard has lost a trust argument. Usually one wrong number landed at a moment that mattered, and the team has been quietly suspicious ever since.
The fix. Two parts. First, add the data quality status pill to the top of the dashboard so the team can see that the checks ran. Second, find the bad-number incident in the past, write a one-page post-mortem, and send it to the team. Trust is rebuilt by acknowledging it was broken, not by hoping nobody remembers.
Sign 3: tiles get added but never removed
Three months after launch, the dashboard has eleven tiles. Six months in, it has fourteen. Every tile was added because someone asked for it, and every tile is technically used. And the dashboard reads as cluttered.
Diagnosis. The team is treating the dashboard as a feature wishlist. Each addition is a small win for whoever requested it, and a small loss for everyone who has to scan past it. The losses compound silently.
The fix.Run the cull from picking the right six widgets, this time with a quarter of usage data to back it up. For every tile, ask: in the last thirty days, did anybody make a decision because of this tile. If no, cut it. We move retired tiles to a second "archive" page, never delete them. Nobody complains when something they asked for moves there. Many people complain when it stays on the main board.
Sign 4: the dashboard outlives the workflow it was built for
The company reorganised. The product changed. Sales moved from inbound to outbound. The dashboard, still loyally showing the same six tiles, no longer matches the work.
Diagnosis. This is not a dashboard problem. It is a calendar problem. Nobody scheduled a moment to ask whether the dashboard still fits.
The fix. A standing review, once a quarter, with the operational owner and one stakeholder. Forty-five minutes. Three questions: what has changed in our workflow, what tiles are now lying to us, what tiles do we wish we had. The dashboard either survives the review intact, or gets a small rebuild. Both outcomes are healthy.
The pattern across all four
Dashboards die because nobody is paying attention. Every sign above is a quiet symptom of the same root cause: the person who owned the build moved on, and the dashboard kept running without an owner.
The single best defence is naming a steward. Not a builder, not a sponsor. A steward. One person who has thirty minutes a week of permission to watch the dashboard the way a gardener watches a garden. Without one, even a perfect build decays. With one, an imperfect build keeps adapting.
What revival looks like
A dashboard we resurrect usually loses about a third of its tiles and gains two new ones. The remaining ones get relabelled into the team's current vocabulary, because the words have drifted in the months since the build. The whole revival takes about three days. It is cheaper than starting over, and the team appreciates the continuity.
The only dashboards we cannot revive are the ones where the workflow they were built for no longer exists. Those we retire with honours, and we build the next one.
