Mon – Sat: 9:00 AM – 7:00 PM IST
Analytics

What is a KPI dashboard?

A small set of measures someone acts on — and the reason most dashboards are opened twice and abandoned.

In short

A KPI dashboard presents a small number of key performance indicators for a specific role, in one view, with the ability to drill from a figure to the transactions behind it. The defining constraints are that it is role-specific and small. A dashboard showing forty metrics to everyone is a report nobody reads.

Why It Matters

Why this matters in practice

Dashboards fail for a consistent reason: they are built to display what is available rather than to answer what a particular person decides each week. If a measure does not change anybody's action, it is decoration. And a figure that cannot be drilled into is not actionable — the first question anyone asks about a bad number is which transactions caused it.

Built for one role's decisions, not for everyone
Few measures, each tied to an action someone can take
Drill-down from figure to source transactions
Comparison against target or prior period, not a bare number
How It Works

How it works in practice

Start from decisions

What does this person decide weekly? The measures follow from that, not from what the database contains.

Choose few measures

Five to eight per role. More than that and attention disperses across all of them.

Give every figure context

Against target, against last period, or against a trend. A number alone cannot be judged.

Enable drill-down

One click from the figure to the underlying documents. Without it the dashboard raises questions it cannot answer.

Define measures once

Centrally, so every view of revenue or margin means the same thing.

Worked example

A worked example: a dashboard with 34 tiles

An executive dashboard is built with 34 metrics across six categories. It is comprehensive, visually impressive, and within two months nobody opens it.

The failure is that 34 numbers with no hierarchy demand that the reader decide what matters, every time. Without thresholds, nothing stands out. Without ownership, no number implies an action. Without drill-down, a concerning figure cannot be investigated without asking someone. The dashboard reports; it does not prompt.

A rebuild with six metrics — each tied to a stated objective, each with a target, a threshold and a named owner, each clickable through to the underlying records — gets used daily. The other 28 numbers are not lost; they live one level down, available when a top-level measure prompts a question. The principle is that a dashboard's job is to direct attention, and a screen where everything is visible directs attention nowhere.

Key terms

The vocabulary, briefly

Term What it means
KPI A measure tied to an objective, with a target and an owner. Without those it is just a number.
Leading indicator A measure that predicts an outcome, so there is still time to act.
Lagging indicator A measure that reports an outcome after it has happened.
Threshold The level at which a measure demands attention, defined in advance rather than judged.
Drill-down Moving from a summary number to the records behind it, which is what makes a dashboard actionable.
What Goes Wrong

Common mistakes worth avoiding

Everything on one screen

A dashboard serving every role serves none. Build per role.

Measures nobody acts on

If a figure never changes a decision, remove it. Its presence dilutes the ones that matter.

No drill-down

A figure that cannot be investigated gets ignored once it is first disbelieved.

Vanity metrics

Totals that only ever go up look reassuring and tell you nothing about performance.

What to measure

Numbers worth tracking

Metric What it tells you What distorts it
Dashboard usage Who opens it and how often. The honest verdict on whether it is useful.
Metrics with a named owner Measures nobody owns produce no action regardless of how they move.
Drill-down usage Whether people investigate or merely glance. Low drill-down suggests the numbers are not trusted.
Time from data event to dashboard Latency against decision need, which is often over-specified.
FAQ

Questions we are asked about this

Typically five to eight for a given role. The constraint is attention, not screen space — people act on a handful of measures and skim the rest.
Operational ones often should — today's despatches, open exceptions. Performance dashboards rarely need to be; a daily refresh is usually sufficient for trend-based decisions.
Five to seven on a top-level view, with depth available underneath. The constraint is attention rather than screen space: beyond about seven items, readers stop scanning and start skipping. If more measures genuinely matter, that usually indicates several audiences rather than one dashboard, and separate role-based views will serve all of them better than one comprehensive screen.
It is tied to an objective someone is accountable for, it can be influenced by the people who see it, it has a target and a threshold defined in advance, and it is hard to improve without actually improving the underlying thing. That last criterion rules out a surprising number of popular measures — any metric that can be moved by changing how work is recorded rather than how it is done will eventually be moved that way.
Only where decisions are made in real time. Warehouse throughput during a shift and order flow on a peak day genuinely benefit. A monthly margin review does not, and building it real-time adds cost and fragility for no decision benefit. Match the refresh to the decision cadence, and be sceptical of real-time as a default requirement — it is usually easier to request than to justify.
Govern the definitions rather than the dashboards. If revenue, margin and on-time delivery have single agreed definitions in a semantic layer, then multiple dashboards built on them still agree, and proliferation is harmless. What causes damage is not many dashboards but many definitions, which produces meetings where the first half hour is spent reconciling whose number is right.
Related

Where to read next

Operational vs analytical reporting

Read more

OTBI vs BI Publisher

Read more

Our ERP software solutions

Read more

Need this applied to your own systems?

We build and support the systems this page describes. Start with a free working proof of concept — no cost, no commitment.