Nobody counts the dashboards

Every BI team has heard a version of the claim that most dashboards are opened once and never again. I wanted a number for this column, so I went looking for a study from 2025 or 2026 with a stated sample and method.

I did not find one. The figures that circulate either come from surveys several years old or arrive with no method at all. For an industry that sells measurement, we are oddly bad at measuring our own output.

What does exist points the same way. A small study by Datalogz, based on interviews with 55 data and analytics leaders and practitioners, found more than two-thirds dealing with the problems of BI sprawl.¹ In Sigma’s 2025 survey of more than 500 BI users and leaders, 76% cited slow performance, with delays in dashboards and reports, and 37% of executives wanted BI to build custom data apps as well as dashboards.² Salesforce found that fewer than half of business leaders, 49%, could reliably generate timely insights.³ None of these counts unused dashboards. Together they describe estates that have grown faster than anyone has pruned them.

Budgets have not been the constraint. Dresner Advisory’s 2025 study found more than 90% of respondents increasing or maintaining their BI budgets.⁴ Buyers do seem to be tiring of assembling their own stacks: in Dresner’s 2025 analytical data products report, as quoted by Domo, 52% preferred a complete platform from a single vendor and 37% a composable approach.⁵ Microsoft said in July 2026 that it had more than 40,000 paid Fabric customers, up more than 60% in a year.⁶ Consolidation onto fewer platforms is a good moment to count what you are carrying across.

Then there is the AI layer. Gartner found more than half of analytics and AI leaders saying their organisations already use AI tools for automated insights and natural language queries, and it forecasts that 75% of new analytics content will be contextualised for intelligent applications through generative AI by 2027.⁷ ThoughtSpot reported that more than 52% of its customers were actively using its natural-language agent, from its own platform data.⁸ If people increasingly ask questions in plain language, the pile of rarely opened dashboards gets less useful each quarter, while still costing money to maintain and to keep correct.

Count your own

You do not need an industry statistic to act on any of this. Most BI platforms keep usage logs. Pull ninety days of view data and sort every dashboard and report by distinct viewers. The bottom of that list is worth bringing to the next planning meeting, because it turns an argument about taste into a list of names.

• Anything with no viewers in ninety days goes to its owner with a retirement date.
• Anything with no owner gets one, or goes.
• Anything that duplicates a certified dataset gets rebuilt on that dataset or merged into it.

Do it once and publish the result. Do it every quarter and the estate stays small enough to keep correct, which matters more now that an agent may be reading the same numbers your executives are.