// logs · data

Why data visualization matters

A table tells you what happened. A good chart tells you what to do next. Notes from years of designing dashboards people actually open.

// date
2026.10.31

// category
data, design

// reading time
9 min

// author
zoltán

// fig. 01 · a dashboard on a wall screen in a meeting room

// 01

Your eyes read pictures first

Show someone a row of twelve monthly numbers and watch what happens: they start reading, left to right, holding each value in their head and comparing it with the one before. By the eighth number most people have lost the thread. Show them the same twelve numbers as a line and the story arrives at once: steady growth, a dip in summer, a spike in November that needs explaining.

That's the whole case for visualization in one sentence. Our visual system finds patterns, outliers and trends long before the conscious mind has finished parsing digits. A chart isn't decoration laid over the data. It's the shortest route from a number to a decision.

It also works the other way. A bad chart sends people down the wrong route just as quickly. A truncated axis, a 3D pie or a rainbow palette can make a flat quarter look like a crisis. Visualization is powerful exactly because people trust what they see, so it has to be designed with care.

// in this log

01 your eyes read pictures first
02 a dashboard is a language
03 five rules we design by
04 designed dashboards get used
05 tools are not the strategy
06 how we work

// 02

A dashboard is a language

Every good dashboard has a grammar, even if nobody wrote it down. Position and length carry the values, because our eyes compare them precisely. Colour points attention at what matters. White space groups things that belong together and separates things that don't. Type tells you what to read first.

When those rules are kept, people stop decoding the screen and start reading it, the way you read a sentence without spelling out the letters. When they're broken, with five colours that mean nothing, labels at odd angles and a legend you need to memorise, even correct data turns into noise.

We treat that grammar as a design system. Chart types, colours, number formats and spacing are defined once and reused everywhere, so a sales report and an operations report feel like they come from the same company. They do.

// author

zoltán
data and dashboard design.
tableau, power bi, figma.

// fig. 02 · sketches of chart layouts pinned next to a laptop

// one chart, before and after the rules

// 03

Five rules we design by

01 — Start with the question. Each chart answers one business question. "How did revenue develop by channel this quarter?" is a chart. "Revenue" is not. If we can't write the question down, the chart doesn't ship.

02 — Pick the chart for the comparison. Bars compare amounts, lines show change over time, scatter plots show relationships and maps show where. A pie is almost never the answer, and a donut is a pie with a hole.

03 — One accent, one message. Everything stays grey except the number that needs attention. If everything is highlighted, nothing is.

04 — Label directly. Put the name on the line, not in a legend at the bottom. Every glance between the chart and a legend is a small tax on the reader.

05 — Cut what carries no data. Heavy gridlines, borders, shadows, 3D effects and decorative icons go first. What remains should be the data and just enough structure to read it.

// checklist

[ ] one question
[ ] right chart
[ ] one accent
[ ] direct labels
[ ] no chart junk

// a rule we keep on the studio wall

“If a chart needs a legend to be understood, it needs another draft.”

// 04

Designed dashboards get used

Most BI projects don't fail on data quality or tooling. They fail on Monday morning, when someone opens the report, needs two minutes to find their number, gives up and goes back to the spreadsheet they trust.

That's a design problem, not a data problem. So we treat dashboards as interfaces, with a clear hierarchy, consistent components, sensible navigation and an obvious place to start. The most important number sits top left, big and alone. Detail lives one click deeper, for the people who need it.

We also design for the real screen: the laptop in a meeting, the phone in a corridor, the TV on the office wall. A dashboard that only works on a 27-inch monitor is a dashboard for one person.

// key takeaway

treat a dashboard like an interface:
one obvious place to start,
detail one click deeper.

// 05

Tools are not the strategy

Tableau, Power BI and Looker are all excellent. None of them decides what your dashboard should say. The tool choice usually follows what the company already runs. The design work, deciding what to show, in which order and for whom, is the same in each of them.

What changes everything is a clean data model underneath. A beautiful chart on top of three conflicting definitions of "customer" is still wrong, just more convincingly. That's why we work closely with the data team from day one, so design and data meet in the middle.

// tools

tableau
power bi
looker
figma

// fig. 03 · the same report on a phone, a laptop and a wall screen

// 06

How we work

01 — Workshop the questions with the people who actually decide, not only with the people who build the reports.
02 — Sketch with real data, not lorem numbers. Real data breaks layouts in ways placeholder data never will.
03 — Build in Tableau or Power BI on a reusable component system: colours, chart types, number formats.
04 — Test it with the people who'll use it, watch where they hesitate and fix that.
05 — Cut a third of it. The last step is always taking things out.

A good dashboard feels obvious afterwards. That's the point, and it's a lot of work.

// work with us

need a dashboard people actually open?
hello@khahn.design

// keep.reading.

More logs

Two more notes from the studio, on brands and on AI.

// logs · branding · 10 min

A brand is a promise you keep

A logo is a signature. A brand is everything you do the same way, every time. How we build identities and why the brand book matters.

// logs · ai · 11 min

AI in design: more taste, not more noise

AI doesn't replace judgment. It gives judgment more options to choose from, faster. How we use it, and where we don't.

// logs · archive

All logs

Every note from the studio in one place: data, branding, ai and the way we work.