Removing charts that do not support a decision: a method

For each dashboard visual, test its decision value, confirm who relies on it and where any needed detail will remain, then remove it and review the next decision meeting for lost information.

Audit actual use

Identify who reads the visual, the question it answers and what decision or action it supports. Keep it if it helps that audience make a decision, monitor the current state or understand necessary context.

Flag a chart for removal if it has no identifiable audience or decision or monitoring role, duplicates another visual, or shows detail readers do not need to monitor on the dashboard. A popular chart may still be decorative if it reports a number nobody can influence.

Use click frequency as one input, not a pass-or-fail test. A rarely viewed risk indicator may still be essential, so check its purpose with the people who rely on it before removing it.

Check for duplicate views: a headline number, gauge and trend for the same measure may use three spaces where one trend with the current value and target would do. If a visual is needed but unclear, replace an unclear pie or dense table with the simplest form that communicates the comparison.

Microsoft Learn's “Tips for designing a great Power BI dashboard” describes a dashboard as an overview for monitoring the current state, with detail in underlying reports and semantic models. Readers can drill into those reports, so leave detail off the dashboard unless they need to monitor it there.

Remove safely

Before removing a chart, confirm with people who rely on it and tell affected readers about the change. If the information remains necessary, point them to a supporting report or underlying report instead.

Remove the visual from the dashboard while keeping any still-needed underlying report or semantic model available. Keep the metric definition and history with the metric.

At the next decision meeting, check whether questions are clearer and action is faster. If people rebuild the chart elsewhere or lose information needed for a decision, reconsider its role and design.

The test is whether the dashboard directs attention to changes that need a response, with evidence close enough to support that response.