Build a dashboard that fits on one screen

A dashboard's purpose: decide at a glance

A dashboard isn't a collection of pretty charts. It's a decision instrument: in a few seconds, it must answer "is everything fine, and if not, where do I act?". A good dashboard reads like a car's dashboard — speed, fuel, temperature — not like the engine's technical manual. If you have to think for thirty seconds to understand a screen, it's failed.

The golden rule: a dashboard fits on a single screen and reads in under a minute. Anything beyond that is an analysis report, not a steering dashboard.

Visualization tools, by profile

• Google Looker Studio (formerly Data Studio): free, connects natively to GA4, Google Ads, Google Sheets, BigQuery. It's the natural starting point if your data is in the Google ecosystem. You build dashboards shareable by a simple link.
• Metabase: open source, plugs into a database (PostgreSQL, MySQL, BigQuery). Lets you ask questions in near-natural language and build dashboards. Ideal as soon as you have a real application database. Paid cloud version or free self-hosting.
• Databox or Geckoboard: specialists in the "TV" dashboard that aggregates dozens of marketing sources via ready-made connectors, no technical work. Handy but paid.
• The spreadsheet itself: Google Sheets and Excel make perfectly honorable charts. For simple tracking, no need to add a tool — a "dashboard" tab in your Sheet is often enough.

Practical rule: Google data → Looker Studio. Application database → Metabase. Simple tracking → a chart tab in your spreadsheet.

Choosing the right chart for the right question

A common mistake is using an unsuitable chart type that muddles the message. A few guideposts:

Two classic traps: the twelve-slice pie (unreadable — beyond 4-5 categories, prefer bars) and the truncated axis that doesn't start at zero and visually exaggerates a variation. An honest dashboard doesn't lie through the chart.

The structure of an effective dashboard

A good entrepreneur's dashboard is organized into reading layers:

graph TD
    A[Top: 3-5 main KPIs<br/>as big numbers] --> B[Middle: trends<br/>lines over time]
    B --> C[Bottom: detail by segment<br/>source, channel, product]

1. At the top, the vital numbers: three to five KPIs displayed large, each with its variation versus the previous period (a number without comparison says nothing — "1,200 visitors" only makes sense next to "+15% vs last week").
2. In the middle, the trends: the lines showing the trajectory over time.
3. At the bottom, the detail: the breakdown by source, channel, or product, to know where to act.

You read top to bottom: first whether things are fine, then why, then where.

The ritual matters more than the tool

The finest dashboard is useless without a reading ritual. Set a recurring appointment — for example every Monday morning, fifteen minutes — to open the dashboard, note what moved, and decide on an action. Many tools (Looker Studio, Databox) can even email the dashboard on a regular schedule, which removes the excuse of forgetting.

A dashboard landing in your inbox every Monday that you read over your coffee is infinitely better than a sophisticated one you never open.

Start small, without blushing

To start this week: create a single screen with your three most important KPIs (for example visitors, conversion rate, revenue), each compared to the previous week, plus a revenue line over the last twelve weeks. Whether it's in Looker Studio or a Google Sheets tab matters little. What counts is finally having one place that answers, at a glance, "how is my business doing?". You'll sophisticate later, at the pace of real questions.