How Many KPIs Should a Dashboard Have? The Case for Radical Focus

## The short answer

An operational dashboard meant for steering should usually carry **between five and nine KPIs**, with a single, unambiguous primary metric at the top. Beyond roughly nine, the dashboard stops being a decision tool and becomes a data dump that nobody reads carefully. The discipline is not in choosing what to add, it is in choosing what to leave off. If you find yourself unable to cut below a dozen, you are almost certainly mixing several distinct dashboards into one.

## Why more metrics make worse decisions

It feels prudent to track more. Surely the more we measure, the more we know? In practice, comprehensiveness and clarity pull in opposite directions.

- **Attention is finite.** A human can hold only a small number of things in working memory. A wall of thirty numbers is processed as texture, not information, the eye glazes and nothing is truly read.

- **Signal drowns in noise.** When everything is on the screen, the one number that actually changed this week is no more prominent than the twenty-nine that did not.

- **No metric feels owned.** Thirty KPIs cannot each have a clear owner who acts on them. Diffuse responsibility means nobody acts on any of them.

- **Review meetings sprawl.** A focused dashboard makes a stand-up sharp. A sprawling one turns every review into a tour of trivia.

## The hierarchy that replaces the dump

The answer to "but we need all this data" is not one giant dashboard. It is a hierarchy of focused ones.

1. **The north-star view.** One primary metric and a handful of supporting KPIs, perhaps five to seven. This is what leadership glances at daily.

2. **Team operational views.** Each team gets its own focused dashboard of the metrics it directly controls, again capped at around nine.

3. **Diagnostic views.** The deep detail, dozens of metrics if you like, lives here, consulted only when a higher-level KPI signals a problem.

The detail is not deleted; it is *demoted*. It sits one click away, available on demand, rather than competing for attention by default.

## How to choose the survivors

When pruning to five to nine, these criteria help decide what stays.

- **One primary, non-negotiable.** Every dashboard needs a single metric that matters most, the one you would keep if you could keep only one. Place it largest and first.

- **Cover the balance, not just the obvious.** A purely financial set can hide quality or customer problems. A small, balanced spread, outcome, customer, operational, financial, beats a deep dive into one dimension.

- **Prefer rates over totals.** Ratios and rates carry more decision value per pixel than cumulative absolutes.

- **Drop redundant cousins.** If two metrics almost always move together, keep one.

- **Cut anything without an owner or a target.** No owner, no target, no place.

## The objections, answered

**"But different stakeholders need different things."** Exactly, which is why they need different dashboards, not one shared dashboard trying to serve everyone. Tailor the view to the audience.

**"What if we miss something by not showing it?"** You will not miss it, because it lives in the diagnostic layer and surfaces when a headline metric breaches a threshold. Focus is not blindness; it is deferred detail.

**"Our business is complex."** Complexity is the argument *for* focus, not against it. The more moving parts there are, the more essential it is to elevate the few that matter most. Complexity handled by a thirty-metric dashboard is complexity hidden, not managed.

## A simple capacity rule

If you want a single heuristic: a person should be able to read your dashboard and state, within about thirty seconds, whether things are on track and where attention is needed. If they cannot, you have too many metrics, too little hierarchy, or no clear primary. Trim until the thirty-second test passes.

## Watch for dashboard sprawl over time

Dashboards rarely start bloated; they accrete. Each new initiative adds "just one more" metric, and within a year the discipline has eroded. Guard against this with a standing rule: **adding a metric requires removing one.** A fixed budget of slots forces the team to argue for relative importance every time, which is exactly the conversation you want.

## Where tooling helps

Enforcing a metric budget, surfacing a clear primary, and keeping diagnostic detail one click away rather than on the main screen are all easier with a platform designed around hierarchy instead of infinite tiles. At neart.ai we build enterprise-grade products in this area, and the pattern that holds across every deployment is that the most-used dashboards are the most ruthlessly edited ones.

## Takeaway

Cap each operational dashboard at five to nine KPIs with one clear primary, push the rest into a diagnostic layer a click away, and require a removal for every addition. If a viewer cannot judge the situation in thirty seconds, you have too many numbers, not too few.