← Back to Blog
Design Analytics Dashboards 6 min read · 2025

Smart Solutions to Build Beautiful Dashboards

How I broke out of the grey-grid trap and started using Canva and Claude Code to make dashboards that people actually want to look at.

Before — default
24,310
Sessions
3.2%
Conv. Rate
Traffic by channel
Organic
Direct
Paid
After — designed
24,310
Sessions
3.2%
Conv. Rate
Traffic by channel
Organic
Direct
Paid

Same data. Same KPIs. One is a grid to be closed. The other is a story to be read.

The grey-grid problem

If you've built enough dashboards in Looker Studio or Power BI, you know the feeling: you spend hours getting the data right, the metrics are accurate, the filters work — and then the client opens it, skims it for 30 seconds, and never comes back. Not because the data is wrong. Because it looks like every other dashboard they've ever seen.

The default design language of BI tools is functional, not engaging. White backgrounds, thin grey borders, the same 5 chart types, the same blue. It gets the job done, but it doesn't create the feeling that someone put real thought into communicating something to you.

I got tired of it. And since I come from a design background — not professional, but genuinely passionate — I started looking for ways to make dashboards that actually feel like they were designed.


Method 1: Canva as a dashboard design layer

The first tool I tried was Canva, and it changed everything. Not as a replacement for Looker or Power BI — you still build your charts there — but as a design layer for the look and feel.

Here's the workflow:

  1. Define the layout and visual identity in Canva — background color, card shapes, icon set, typography, color palette for the charts
  2. Export the background as an image (PNG or PDF)
  3. Import it as the dashboard background in Looker Studio or Power BI
  4. Drop your charts on top, matching the positions to the layout you designed

The key insight: you're not fighting the tool's limitations — you're bypassing them entirely. Canva has hundreds of templates, icon libraries, gradient options, and layout systems that BI tools will never have. Use them for the shell. Use the BI tool for what it's actually good at: data.

What this unlocks in practice:

"A dashboard is a communication tool. If it doesn't communicate clearly and beautifully, it doesn't matter how correct the numbers are."


Method 2: Claude Code for HTML dashboard backgrounds

The second approach is more technical — and produces results that Canva can't match. Using Claude Code, you can generate full HTML/CSS dashboard layouts and then use them in two ways: as a static background image (screenshot the HTML and import it), or as a standalone interactive HTML dashboard.

What's powerful here is the level of customisation you get from a plain-language prompt. You describe what you want, and Claude Code generates the HTML. Then you ask it specifically for the background and icon components, and you drop your real charts on top inside the BI tool.

What you can generate with Claude Code

E-commerce HTML dashboard
€84k
Revenue
1.8k
Orders
4.1%
CVR
Weekly revenue
Paid Media HTML dashboard
€2.4
CPA
3.8x
ROAS
12k
Clicks
Spend by platform
Google
Meta
TikTok
Operations HTML dashboard
99.4%
Uptime
1.2s
Avg. Load
3
Incidents
Daily status — last 7 days
Web Analytics HTML dashboard
41k
Users
2:34
Avg. Time
58%
Bounce
Device split
Desktop 60% Mobile 30% Tablet 10%

All four of the above were generated purely with HTML/CSS — no design tool, no images.

These are the kinds of outputs you can generate with a single Claude Code prompt. The workflow is:

  1. Describe the dashboard to Claude Code: industry, KPIs, color palette, layout
  2. Ask it to generate the background + icon layer only — placeholder boxes where the charts will go
  3. Screenshot the result at the exact dashboard resolution (e.g. 1280×720 for a Looker Studio page)
  4. Import the screenshot as the background image in your BI tool
  5. Overlay your real charts on the placeholder positions

Tip: ask Claude Code to output the layout with a grid overlay — labelled boxes like "KPI 1", "Chart A", "Table B". This makes it trivial to align your real charts on top when you switch to the BI tool.


When to use which approach

Both tools solve the same problem — breaking out of the default aesthetic — but they suit different situations:

In practice, I often combine both: rough concept in Canva, refined in Claude Code, then back to Looker Studio or Power BI for the live data layer.


The real shift

The most important thing I learned from this is that dashboard design is a communication skill, not a technical one. The data can be identical. The structure can be identical. But the difference between a dashboard that gets used every Monday and one that gets ignored after the first demo is almost entirely visual.

Good tools — whether Canva or Claude Code — just lower the barrier to applying that skill. You don't need to be a UI designer. You just need to care enough to try.

Like this post

Comments

Leave a comment

Guillermo García

Guillermo García

Digital Analytics Engineer · TNK Design & Analytics

Get in touch →

Related posts

Design

Design Principles for Data Products

Dev · AI

I Built This Entire Site with Claude Code