Journal

How to design data tables people can actually scan

Practical rules for alignment, density, headers and row actions that make data tables in dashboards and SaaS products easy to read.

How to design data tables people can actually scan

Tables are where a lot of SaaS products quietly lose their users. The marketing site looks polished, the onboarding is friendly, and then the customer lands on a screen with 14 columns, centred numbers and five icons on every row. They came to find one invoice or compare two campaigns, and the interface makes them work for it.

Good tables are not exciting to design, which is exactly why they are often left to defaults. But a few deliberate decisions make a big difference to how fast people find what they need. Here are the rules we follow.

Start with the job, not the data

Before choosing columns, ask what people do on this screen. Usually it is one of three things: find a specific item, compare items, or monitor for something that needs attention. Each job leads to a different table.

  • Finding: strong search and filters matter more than column count. The identifying column should be first and prominent.
  • Comparing: numbers need consistent alignment and units, and sorting must be obvious.
  • Monitoring: status needs to stand out at a glance, and the default sort should put problems at the top.

Then cut columns. Every column you add makes every other column harder to read. If a field is only needed occasionally, move it to the detail view or make it optional through a column picker.

Align text left and numbers right

This is the single most useful rule for tables. Text reads from the left, so left-aligned text forms a clean edge the eye can follow down the column. Numbers are compared by place value, so right-aligned numbers line up their units, tens and hundreds.

Use tabular (fixed-width) figures for numeric columns if your typeface supports them. Without them, a column of prices wobbles because a 1 is narrower than an 8. Align each column header the same way as its content, and never centre long text or numbers.

Choose a density on purpose

Row height is a trade-off between how much fits on screen and how easy each row is to read. Many design systems offer a comfortable and a compact density. Pick the default based on the job: comfortable for occasional users and simple lists, compact for power users who live in the table all day.

Whichever you choose, keep padding consistent and give cells enough horizontal space that values do not touch the column edges. Thin row dividers or very subtle zebra striping help the eye stay on the right line across wide tables. You rarely need both.

Make headers work hard

Headers should be short, specific and in plain language. "Revenue (USD)" beats "Rev." and tells people the unit once, so you do not have to repeat it in every cell.

  1. Make them sticky on long tables so people never lose track of which column is which.
  2. Show sort state clearly, with an arrow on the active column and a visible hover state on sortable ones.
  3. Freeze the first column on wide tables that scroll horizontally, so the row's identity stays visible.

Keep row actions quiet

A row with edit, duplicate, archive, share and delete icons repeated fifty times turns into visual noise. Show the one or two most common actions inline, put the rest in an overflow menu, and consider revealing actions on hover or focus on desktop.

For bulk work, add checkboxes and a toolbar that appears when rows are selected. Destructive actions such as delete should always ask for confirmation or offer an undo.

Use colour and status sparingly

Colour is the fastest signal in a table, so save it for things that need attention. A small status badge or dot for failed, overdue or at-risk rows works well. If every row has a coloured badge, nothing stands out.

Never rely on colour alone. Pair it with a label or icon so the meaning survives for colour-blind users and in low-contrast conditions, and check that badge text meets contrast requirements.

Design the empty, loading and huge states

Tables are rarely in their ideal state. Design for the edges as carefully as for the happy path:

  • Empty: explain why there is no data and offer the next action, such as importing or creating the first item.
  • No results: show which filters are active and offer a one-click way to clear them.
  • Loading: skeleton rows that match the real layout feel faster than a spinner in the middle of a blank area.
  • Very large: decide between pagination and infinite scroll based on whether people need to return to a specific position.

Rethink tables on mobile

Squeezing a 10-column table onto a phone rarely works. For most products, a better pattern is a list of cards that shows the two or three most important fields per item, with a tap to open the full detail. If users genuinely need to compare across columns on mobile, allow horizontal scroll with a frozen first column and make the scroll obvious.

A good table disappears. People should notice the information, not the grid around it.

Next time you design a dashboard, give the table the same attention as the hero section. Start from the user's job, cut columns, align with intent, and design every state. It is unglamorous work that users feel every single day.

Oybek Ruziev

Designer & Framer developer · AidanCreates

Keep reading

All posts