Staring at a wall of sensor charts and not knowing what matters? Good iot dashboard design shows only the right data, labels every unit, and uses colour the same way everywhere.
Dashboards are one piece of iot user experience design, and this page covers the screen where live sensor readings meet a human decision.
| Design rule | Value |
|---|---|
| Traces in a single chart before you rethink it | More than five individual traces |
| Default visualizations to avoid | 20 default visualizations |
| Questions to answer before building | Five questions |
| Meaning of a status colour such as red | The same everywhere on the platform |
| Main goal of the dashboard | Deliver the right data, not the most data |
These figures come from Method and SumatoSoft. Published guidance can change, so check the source before you rely on them.
Start with the decision, not the data
Every dashboard exists so that someone can decide something quickly. Method puts the priority plainly: value isn’t delivering the most data, it’s delivering the right data. That sentence is a good filter for every widget you are tempted to add.
Before you sketch anything, answer five questions. SumatoSoft advises settling them before building any dashboard, and they cover these points:
- What problem the dashboard should solve.
- What goal you want to reach.
- What a successful result looks like.
- Which charts fit that goal.
- Whether real-time updates matter.
Then prune hard. Method’s advice is not to hand users 20 default visualizations but to let them select and configure the ones they need with dropdown menus and expandable graphs. The same page warns that too much information at once hinders a user’s ability to absorb any of it.
TagoIO’s walkthrough starts from a weak example that presents all data simultaneously, which can overwhelm users. Its fixes, covered below, are mostly about removing and clarifying rather than adding.
Label every reading so it makes sense on its own
A number without context is a guess. In the TagoIO example, the temperature data lacked units, leaving users unsure if the values were Celsius or Fahrenheit. Adding the unit is the cheapest fix on any dashboard.
The same example improved its text in several ways:
- Widget titles and labels became specific about what they represent.
- Temperature values gained a visible unit.
- Table columns got real names instead of one repeated variable.
- Unreadable option IDs were replaced with human-readable labels.
- Chart subtitles now explain the severity levels.
Alert counters need the same care. The widgets that show the number of alerts did not specify the severity of the alerts, so the team made it explicit and defined severity 1 as the highest level.
TagoIO also argues that clearer text pays off before any styling does. In its words, the dashboard may look bland and cluttered, but at least reading the text allows the user to make sense of everything.
If your users span regions, let them choose. TagoIO notes that you can provide a great experience to your users by allowing them to select units according to their preferences, which is very important in a global context.
Use colour and icons the same way everywhere
Method calls colour one of the most instinctive ways users group, prioritize and make decisions about data. That instinct only works if your palette is predictable. The same page says consistency is key: do not make red a positive state in one area of the platform and a negative in the next.
TagoIO applies the rule like this: if a color means something that needs the user’s attention, every other thing that needs the same amount of attention will use the same color. It also added icons, because severity and seen-status columns with icons act as a quick visual representation that does not make the user parse a bunch of text.
IoT For All shows the pattern on a fleet map. Siren Marine’s dashboard has two general types of icons, a normal state and an alert state, and boats shown in red direct the user to click to learn more. When one icon looks different, the eye goes straight to it.

Choose chart types that fit the reading
The chart should match the question the user is asking. Bestiario suggests a few pairings that work well for sensor data:
- Line charts and time-series graphs for trends over time, like temperature changes or energy use.
- Heat maps for areas of high activity or performance problems, which suits location-based data.
- Scatter plots for relationships between variables, such as humidity and machine performance.
- Pie charts and bar graphs for comparing categories, used sparingly to avoid clutter.
Whatever you pick, label it. The same source says you should clearly label all axes, legends and data points to make information easier to understand.
Avoid the rainbow chart, too. Method’s rule of thumb is that if you are displaying more than five individual traces in a single visualization, it is probably time to ask how that information might be better displayed.
One alternative is to make a graph adjustable. IoT For All suggests considering one graph that can be altered to depict different variables, instead of several separate graphs for the same feature.
How do you keep a dashboard from becoming cluttered?
Start by deciding what the headline is. Method says the data hierarchy on a page is typically achieved with font sizing, page location or color highlighting. The reading people check most should be the largest thing on the screen.
Then cut what is not essential. In the TagoIO example, the team kept indoor temperature and humidity and removed outdoor information and the less severe alert counters, reducing noise. Moved details can live on another tab, another dashboard, or behind a button.
Hide detail until it is asked for:
- Keep graph legends inside a tooltip or hover state, as Method recommends for whitespace.
- Show a hover card with the essentials. On the Siren Marine map, hovering over a boat reveals its name, speed, location and last-updated timestamp.
- Make navigation menus collapsible, since Method notes that real estate is always a hot commodity in dashboards.
Finally, test with real users. TagoIO says that if none of your users care about a piece of information in the dashboard, it takes space from more important things.
The same readings often end up on a small screen as well, and the iot companion app design guide covers how those layouts differ.
Show alerts, live readings and predictions honestly
Users need to know what is happening now and what might happen later. Method says predictively derived data needs to be differentiated from live information so nobody confuses a current problem with a possible future one. A different style, label or panel does the job.
Let people choose their time window. Method observes that real-time data visualization almost always comes with a time window selection, so do not lock users into one slice of history.
Plan notifications early, not at the end. In-platform notifications require real estate to execute and need colour indications consistent with the rest of the platform. If alerts may also be spoken aloud, the voice user interface design guide explains how the wording changes.
Frequently asked questions
How many widgets should an IoT dashboard have?
Show answer
The sources behind this guide set no ideal count. They point to a test instead: a widget nobody cares about takes space from more important things, so ask your users which ones they actually open.
Do I need to show units on every value?
Show answer
Yes, wherever a reader could misread the scale. The temperature example above shows how a missing unit leaves people guessing, and letting users switch units makes the screen friendlier across regions.
Should a dashboard be interactive?
Show answer
Usually, a little. Bestiario notes that drill-down lets users click a data point for more detail, and dynamic filters let them tailor the view to specific parameters.
Who should I design the dashboard for?
Show answer
Design for the person who will act on the data. For non-technical users, simple, high-level visualisations that convey key insights are more appropriate than dense analytical views.
