Real-time dashboards: what to measure and how to build them

"We want a real-time dashboard" is one of the most common requests we hear. But a screen full of moving numbers isn't useful on its own. The value comes from seeing a problem early enough to act on it.

So what should a real-time dashboard measure? Only the metrics tied to decisions people make every day and that change fast enough to need a response — orders, queues and throughput, machine and system status, stock levels for fast-moving items, and error rates. Slower strategic metrics like monthly revenue belong in daily or weekly views.

Start with decisions, not data

Before choosing charts, list the decisions people make every day:

A warehouse manager deciding what to reorder
A clinic administrator balancing doctor schedules
An operations head spotting a delayed shipment

For each decision, ask: what would I need to see, and how quickly, to make it better? Those answers become your metrics.

Not everything needs to be live

Real-time is valuable for things that change fast and need a response:

Orders, queues and throughput
Machine and system status
Stock levels for fast-moving items
Error rates and anomalies

Strategic metrics — monthly revenue, customer lifetime value — are better as daily or weekly views. Mixing the two on one screen creates noise.

Design for action

A few principles we follow:

One screen, one audience. Leadership and floor teams need different views.
Show thresholds, not just values. Colour and alerts should tell you when something is wrong.
Make every number clickable. Drilling into the records behind a metric builds trust.
Alert where people already are — email, WhatsApp or Slack — instead of expecting someone to watch the screen.

The architecture behind it

A typical real-time dashboard has three layers:

Sources — your ERP, CRM, databases, devices or third-party APIs.
A streaming pipeline that captures changes as they happen and computes aggregates.
The dashboard, which receives updates over WebSockets and redraws only what changed.

Keeping the pipeline separate from the dashboard means you can add new sources or views without rebuilding everything.

Getting started

Pick one team and one decision. Build a focused dashboard for it, measure whether it changes outcomes, then expand. See how this works in our real-time visualization case study, or explore our dashboard and real-time analytics service.