Skip to content

Dashboard user interface design: layout, charts, filters and status

Better Design

Direct answer

Good dashboard user interface design answers one question at a glance. Lead with the number that answers it, choose bar, line or scatter charts, keep status colors quiet, and hold one spacing density across the page. This guide covers dashboard types, layout, chart choice, filters, mobile dashboards, accessibility and user testing, from Nielsen Norman Group, W3C WCAG and the platform guidelines.

A hand-drawn browser window with Dashboard user interface design written in it and a short list of what it does.

What a dashboard is for

Nielsen Norman Group defines a dashboard as a collection of data visualizations on a single page that gives at-a-glance information people can act on quickly. It differs from a portal, which gathers many tools and tasks. A dashboard has one job: communicate the critical information.

So good dashboard user interface design starts from the question a person brings, not from the data you happen to have. A dashboard should give an insight, not copy the structure of the database onto the screen.

Different users bring different questions to the same product. A support lead wants today's open tickets; a finance lead wants this quarter's revenue. Design one view per question, or let users pick the view they need, instead of one page that tries to serve everyone.

Operational vs analytical dashboards

Operational
Used for

Time-sensitive tasks, such as monitoring live systems or orders

What the design must do

Show what needs attention now, in seconds, on one screen

Analytical
Used for

Analysis and decisions that are not urgent

What the design must do

Support comparison over time and between groups, still at a glance

Start from user research, not from the data model

The data model tells you what you can show. Only users can tell you what they need to read first. Nielsen Norman Group separates research into what people say and what people do, and recommends you lean on what they do.

Talk to the people who will use the dashboard before you draw a single card. Ask which decision they make with it, how often they look, and on which device.

User interviews
What it tells you

The decisions people make and the questions they ask

When to use it

Before you design

Field studies
What it tells you

How people use data in their real work, on their real devices

When to use it

Before you design

Card sorting
What it tells you

How users group the metrics in their heads

When to use it

When you group metrics and plan navigation

Usability testing
What it tells you

Where people misread a chart or miss a status

When to use it

On every draft

Analytics
What it tells you

Which views, filters and drill-downs people actually use

When to use it

After launch

Lay the page out in reading order

A clear dashboard has four layers, top to bottom:

  • Title: what the page is, with at most one primary action in the header.
  • Metrics: the few numbers that answer the main question, with each value set larger than its label.
  • Records: the table or list of items behind those numbers.
  • Interpretation: what changed and why, stated only from real data, never an invented trend.

Make the most important number lead

Hierarchy is relative: a metric leads because it differs from the ones around it. Six dials set what leads, and strong dashboards lead with one or two:

  • Size: make the key metric clearly larger, and shrink the supporting ones before inflating the hero.
  • Color: keep accent color for what needs attention; status colors carry meaning, so do not reuse them for decoration.
  • Space: give the lead metric room, and keep its context close.
  • Placement: put the main status first in the scan path, and put an action next to the number that justifies it.
  • Visualization: turn a fact into a pattern people see at once, such as progress, comparison, trend or status.
  • Motion: animate to reveal a change or confirm progress, not for decoration.

Choose the chart from the question

Nielsen Norman Group recommends bar charts, line charts and scatter plots for most needs, because people judge length and 2D position quickly and accurately. Pie charts, gauges and radar charts rely on angle and area, which people compare poorly, and gauges also waste space.

How do these compare?
Chart

Bar chart, sorted

Why

Length is compared quickly and accurately

How has this changed over time?
Chart

Line chart

Why

2D position shows the trend at a glance

How do two measures relate?
Chart

Scatter plot

Why

Position shows the relationship

Which 3–7 items rank highest over time?
Chart

Bump chart

Why

Shows rank, but hides the gaps between items

How has the share of each part shifted?
Chart

100% stacked chart

Why

Shows share even when the total changes

What share is this of the whole?
Chart

Avoid pie charts and gauges

Why

Angle and area are hard to compare

Color status quietly, and keep it apart from actions

Status badges label rows; they are not the content. Keep them readable but quiet. Use dark text in the badge's hue on a light tint of that hue, not white text on a bright fill. A row of solid red and green pills reads as an alarm.

  • Keep action color and status color different, so a button never looks like a warning.
  • Save solid fills for the one element that must be seen first.
  • Measure contrast on the real surface: text at least 4.5:1 against its badge.

Filters that match the data

Derive each filter control from the column's data type, so people pick instead of type:

  • Show every active filter as a removable chip, plus one action that clears them all.
  • Combine filters with and by default.
  • When nothing matches, keep the filter bar and the active filter chips visible, say that no rows match all of them, and offer one action to clear them.
Free text
Filter control

A contains search box

Example

Customer name

Number
Filter control

Minimum and maximum, or a range slider

Example

Order value

Date
Filter control

A date range with presets such as last 7 days

Example

Created date

Status or category
Filter control

A multi-select list with a row count per value

Example

Order status

Yes or no
Filter control

A three-way choice: yes, no, any

Example

Paid

Show loading and data freshness honestly

  • Keep the previous numbers on screen while new data loads, but dim them or show an updating label with the last-updated time, so nobody reads old numbers as current.
  • Use a skeleton only on first load, when no data exists yet.
  • When a request fails, say what failed and offer a retry, never an endless spinner.
  • Keep the current rows visible while the next page of a long table loads.
  • Show when the data was last updated, which matters most on operational dashboards.

Show detail on demand with progressive disclosure

Nielsen Norman Group calls this progressive disclosure: show the few most important options first, and offer the rest on request. The fact that something sits on the first screen tells users it is important.

On a dashboard, the first screen shows the summary numbers, and a click on a card or a bar opens the records behind it.

  • Put on the first level everything users need often, and nothing they rarely need.
  • Label the link to the detail clearly, so users know what they will find.
  • Keep to two levels: Nielsen Norman Group reports that designs with more than two levels often have low usability, because users get lost.

Pick one density and hold it

A dashboard reads as designed when every section, column and card uses the same spacing. Use compact density for dense tools such as data tables and admin views.

Sidebar width
Compact

11.5rem

Comfortable

14rem

Page inset
Compact

1rem

Comfortable

1.5rem

Section gap
Compact

1.5rem

Comfortable

2rem

Column gap
Compact

0.75rem

Comfortable

1rem

Card padding
Compact

1rem

Comfortable

1.25rem

Fix a crowded dashboard before adding whitespace

When a dashboard feels dense, more whitespace makes it longer, not clearer. Three moves work better:

  • Group metrics by the question people ask, so they find the group first.
  • Add a visual anchor where a row refers to something, such as an avatar for a person or an icon for a file.
  • Vary the form: a chart, a short list and a table break one wall of numbers into parts the eye can skip.

How do you design a mobile dashboard?

A mobile dashboard is not the desktop page made small. People look at it for a few seconds, often with one hand, so the small screen must show the few numbers that matter and keep the detail one tap away.

Nielsen Norman Group describes responsive design as one site whose layout changes at breakpoints, based on the width of the screen. The layout must stay usable at every size, not only look correct.

  • Show one metric per card, with its label, its value and one small trend line.
  • Stack the cards in one column, in the same order as the desktop page, so users find the same number in the same place.
  • Move tables, long lists and secondary metrics to a detail screen.
  • Make touch targets large: Apple's Human Interface Guidelines give 44 by 44 points as the default control size on iOS, and Material Design recommends at least 48 by 48 dp.
  • Keep at least 8 dp between targets, as Material Design recommends, so a tap does not hit the wrong filter.
  • Swap wide charts for simple ones: a line with fewer labels, or a sorted bar list in place of a wide grouped bar chart.

Make the dashboard accessible: the WCAG checks that apply

Dashboards rely on color, small text and dense charts, so they fail accessibility checks easily. W3C notes that accessibility also helps people on small screens, older users and people in bright sunlight. These are the WCAG 2.2 success criteria that most often apply to a dashboard.

  • Give every chart a text equivalent that carries its main finding, such as a short summary of the trend or a data table behind it.
  • Make every filter, tab and drill-down work from the keyboard.
Text contrast
WCAG rule

1.4.3 Contrast (Minimum), level AA

What to do on a dashboard

Give text and numbers at least 4.5:1 contrast; large text needs at least 3:1

Chart and control contrast
WCAG rule

1.4.11 Non-text Contrast, level AA

What to do on a dashboard

Give chart lines, bars and input borders at least 3:1 against what sits next to them

Color is not the only signal
WCAG rule

1.4.1 Use of Color, level A

What to do on a dashboard

Add a label, an icon or a pattern to every red or green status

Target size
WCAG rule

2.5.8 Target Size (Minimum), level AA

What to do on a dashboard

Make every click or tap target at least 24 by 24 CSS pixels. A smaller target passes only when a 24-pixel circle centred on it touches no other target or circle

Test the dashboard with real users before you ship

In a usability test, a facilitator gives a participant realistic tasks and watches what they do. For a qualitative study of one user group, Nielsen Norman Group recommends five participants to find most of the common problems.

Write tasks that match the questions the dashboard must answer, and ask participants to think out loud as they work.

  • "Which region missed its target last month?" tests chart choice and hierarchy.
  • "Find the orders that are late and still unpaid" tests the filters.
  • "Is this number up to date?" tests the last-updated label.
  • Run one test on a phone if users open the dashboard on mobile.

Dashboard design best practices: a checklist

Seven checks before you ship:

  • The page answers one named question for one group of users, and its answer sits first.
  • Every chart is a bar, line or scatter chart unless the question needs rank or share.
  • Status colors are tinted, differ from action colors, and never carry meaning by color alone.
  • Text has at least 4.5:1 contrast, and chart lines and bars have at least 3:1.
  • Every section uses the same density, and loading never blanks the numbers.
  • On a phone, each card shows one metric and every target is large enough to tap.
  • Five users have tried the main tasks, and you fixed where they got stuck.

Dashboard design in four steps

The four steps, in the order you take them.

A hand-drawn four-step path: Name the question, Lead with the answer, Group the rest, Hold one density.
  1. Step 1: Name the question
  2. Step 2: Lead with the answer
  3. Step 3: Group the rest
  4. Step 4: Hold one density

Steal this workflow

Run this guide in your own assistant. The prompt carries the guide link and the workflow summary, so the assistant can adapt it to your project.

Read the guide "Dashboard user interface design: layout, charts, filters and status" at https://better-design.com/blog/dashboard-user-interface-design and help me apply it to my project.

Good dashboard user interface design answers one question at a glance. Lead with the number that answers it, choose bar, line or scatter charts, keep status colors quiet, and hold one spacing density across the page. This guide covers dashboard types, layout, chart choice, filters, mobile dashboards, accessibility and user testing, from Nielsen Norman Group, W3C WCAG and the platform guidelines.

Start by asking what I am building and which tools I use. Then walk me through the workflow step by step, adapted to my answers.

Claude and ChatGPT open with the prompt filled in. Gemini copies the prompt to your clipboard first, so paste it when the app opens.

Questions

What makes good dashboard user interface design?

One clear question, the answering number first, simple charts, quiet status colors, filters that match the data, and one consistent spacing density.

What is the difference between operational and analytical dashboards?

Operational dashboards support time-sensitive tasks and must show what needs attention now. Analytical dashboards support decisions that are not urgent and lean on comparison.

Which charts work best on a dashboard?

Bar, line and scatter charts, because people compare length and position quickly. Avoid pie charts and gauges, which rely on angle and area.

How many metrics should a dashboard show?

As many as the person's questions need, grouped by question. Crowding usually comes from metrics that all look the same, not from their number.

Should dashboards use red and green for status?

Use them as quiet tints with dark text, not solid fills, and never as the color of an action button.

How do you design a dashboard for mobile?

Show one metric per card in a single column, move tables and secondary metrics to a detail screen, and make touch targets at least 44 by 44 points on iOS or 48 by 48 dp on Android.

How do you make a dashboard accessible?

Follow WCAG 2.2: text contrast of at least 4.5:1, chart elements of at least 3:1, never color alone for status, and targets of at least 24 by 24 CSS pixels.

How do you test a dashboard design?

Run a usability test with about five users from one group. Give them tasks that match the questions the dashboard must answer, and watch where they misread or get stuck.

Where these rules come from

Each fact comes from Nielsen Norman Group, W3C WCAG 2.2, Apple and Material Design guidelines, and Better Design's UI and UX guidance, the same guidance its MCP server gives coding agents, read on November 15, 2026.

Build the interface in your own code instead

A coding agent such as Claude Code, Cursor or Codex can build these screens in your own code. Better Design gives that agent a design system, interface principles and review rules, so the result follows rules like these and looks designed.