A Dashboard Redesign That Cut Stakeholder Questions in Half

A client's marketing dashboard had forty-one metrics across six tabs. Every week, the marketing ops team fielded six to ten Slack messages asking what a specific number meant, why it had moved, or which of two similarly-named metrics was the 'real' one. The instinct when a dashboard generates confused questions is usually to add more context — a tooltip, a definitions tab, a Loom walkthrough. We did the opposite. We cut the dashboard to fourteen metrics.

Why more context wasn't the fix

We audited every one of those six to ten weekly questions over a month and found a pattern: almost none of them were 'what does this metric mean' questions that a tooltip would solve. They were 'why are there two versions of this number and which one matters' questions, caused by the dashboard surfacing both a raw count and a normalized rate for nearly every metric, plus a legacy metric definition that hadn't been retired when the team switched attribution models eight months earlier.

You can't tooltip your way out of a dashboard that's answering a question nobody asked, twice, with two different numbers.

What actually got cut

We interviewed the five people who used the dashboard regularly and asked each of them to name the three numbers they actually checked weekly to make a decision. The overlap across all five was fourteen metrics. Everything else was either a duplicate representation of something already on the list, a legacy metric nobody had formally deprecated, or a 'nice to have' someone requested once two years ago and never looked at again.

• Removed every raw-count metric that had a normalized-rate equivalent already on the dashboard — kept whichever one the interviews showed people actually acted on.
• Retired the pre-attribution-model-switch metric entirely and archived its history in a separate reference tab instead of the main view.
• Renamed the remaining metrics to match exactly what stakeholders called them in conversation, even when that didn't match the underlying table's column name.
• Added a single 'last updated' and 'definition source' line per metric instead of a separate definitions tab nobody opened.

Weekly clarification questions dropped from an average of eight to under four within a month, and most of what remained were genuinely new questions about new campaigns, not confusion about the dashboard itself. The lesson that stuck with me: a dashboard's job isn't to represent all the data you have. It's to represent the fourteen numbers five specific people need to make a decision, and every metric past that is a liability, not a bonus.