Tithi MondalBlog

Article6 min readUpdated

Data Table Design for Web Apps: Sorting, Filters, Bulk Actions

Data table design for web apps: how to plan columns, sorting and filters in tables, bulk actions and tables on mobile so admin users find rows fast.

Read inहिन्दीবাংলাमराठीதமிழ்తెలుగుગુજરાતી

Data Table Design for Web Apps: Sorting, Filters, Bulk Actions: cover illustration

A school office has 800 students in one table. The clerk needs to find every student whose fees are pending for April and send a reminder. In a badly designed table, she scrolls for ten minutes. In a good one, she filters, selects all and sends in under a minute. Data table design is the work of making rows of information quick to scan, find and act on.

Who uses a data table, and what are they doing?

Table users are working, not browsing. They are clerks, support staff, accountants and managers doing the same job many times a day.

Before choosing columns, write down the three questions the user asks most. For the school clerk: who has not paid, who paid today, and which class is this student in. Each question suggests a default sort, a filter and a column.

Most table screens are used on a desktop or laptop, often with a small screen and a mouse that is not always good. Some are used on a phone by field staff. I plan both from the start. For the wider context of admin screens, read admin dashboard UI UX design and B2B SaaS product design basics. This article goes deeper on the table itself.

How many columns should a table UI design have?

A table should have 5 to 7 visible columns, ordered by importance. More columns can exist, but they should be hidden by default and available through a column picker.

Column rules:

  1. First column is the identity. The name, order number or invoice number. It is the thing the user clicks to open the full record.
  2. Group related columns. Put date and status together, amount and balance together.
  3. Align by type. Text to the left, numbers to the right, status labels with fixed width.
  4. Use clear headers. "Fees due" is better than "Amt".
  5. Show units once. Put the rupee sign in the header, not in every cell, when all values are rupees.

Freeze the first column and the header row, so they stay in view during scrolling. For a table with 20 rows, any layout works. For 2,000 rows, a frozen header is necessary.

Row height is a real decision. A compact row shows more data, a comfortable row is easier to click. Offer both as a density setting for heavy users, with comfortable as the default.

How should sorting and filters in tables work?

Sorting changes the order of rows by one column, and filtering reduces the rows to the ones that match. Both should be visible, instant and reversible.

For sorting:

  • Clicking a column header sorts it. A small arrow shows the direction.
  • Only sortable columns show the arrow on hover. A name, date or amount is sortable, a notes column is not.
  • Set a sensible default sort, such as newest first.

For filters, place them above the table, not hidden in a menu. Use these patterns:

NeedControl
A few known valuesDropdown or chips, such as Status
DateDate range with presets
Free textSearch box over name and ID
Many valuesMulti select list with its own search

Show active filters as removable chips and a clear "Reset all" link. A hidden filter is the top cause of the complaint "my data is missing". Also show the result count: "42 of 800 students". For the broader principles, see search and filter UX design.

How do bulk actions in a table work?

Bulk actions let the user change many rows at once. They appear only after at least one row is selected, and they must say how many rows will change.

The flow:

  1. A checkbox sits at the start of each row, and one in the header selects the visible rows.
  2. After selecting, a bar appears above the table: "12 selected" with actions such as Send reminder, Export, Delete.
  3. If the table has pages, offer "Select all 342 matching rows" as a clear second step. Do not select hidden rows by accident.
  4. Risky actions, such as delete, ask for confirmation that repeats the count: "Delete 12 students?".
  5. After the action, a message confirms what happened and offers Undo when possible.

Keep per-row actions in a "more" menu at the end of the row, and keep the most common action visible. Do not show 5 icons on every row. The user cannot tell what they mean, and the table looks heavy.

How do tables work on mobile?

On a phone, a wide table cannot be read. The reliable answer is to convert each row into a card that shows the 3 or 4 most important fields, and open the full record on tap.

The student row becomes a card: name and class on top, fees due in large type, a status label and a call button. Sorting and filters move into a bottom sheet opened by a single Filter button. Bulk selection can be done with a long press, but offer a visible Select button as well, because long press is not discovered by many users.

When data truly needs columns, such as a stock comparison, allow horizontal scrolling with the first column fixed. Show a small shadow at the edge so the user knows more columns exist. Check the result on a 360 pixel wide screen, which is common on mid-range Android phones. My article on mobile first website design explains the thinking, and screen sizes and resolution helps with the sizes.

What about empty tables, loading and errors?

A table has four states besides data: loading, empty, no results and error. Each needs a clear message and a next step.

  • Loading: show grey row shapes where data will appear, not a spinner over a blank page.
  • Empty: explain what the table will hold, and give an "Add first student" button.
  • No results: say "No students match these filters" and offer Reset filters.
  • Error: say what failed in plain words and offer Try again.

I cover these in detail in empty states, loading and error screens.

Pagination or infinite scroll is a choice with trade-offs. Pagination lets users reach a position and return to it, so it suits work screens. Infinite scroll suits feeds. For admin tables I prefer pagination with a page size option.

What data table mistakes cost users time?

The costliest table mistakes make the user lose their place or their data. These are the ones I see most.

  • Too many columns, no priority. Everything looks equally important.
  • Filters reset after opening a row. Keep filters when the user returns.
  • Colour only for status. Add a word, such as "Overdue", along with the colour.
  • Truncated text with no way to read it. Show the full value on hover or in the detail view.
  • Bulk delete without a count. This causes serious accidents.
  • Dates in mixed formats. Use one format everywhere, such as 14 Nov 2026.
  • Export that ignores the filters. Users expect the file to match what they see. See my reports and analytics screen design article.

On admin panels I have designed, the table was usually the screen used the most, so it received the most design time. A table that looks plain but works fast is a good result.

If you need an admin panel or internal tool where tables carry the daily work, see my UI/UX design service or reach me at the contact section.

Frequently asked questions

How many columns should a data table have?

+

Show 5 to 7 by default, with the most important first. Offer a column picker for the rest so the table stays readable.

Should I use pagination or infinite scroll in an admin table?

+

Pagination is usually better for work screens, because users can return to a position and select rows reliably. Infinite scroll suits feeds.

How should a table look on mobile?

+

Turn each row into a card with the 3 or 4 key fields, and move filters into a bottom sheet. Use horizontal scroll with a fixed first column only when columns must be compared.

Where should table filters go?

+

Directly above the table, with active filters shown as removable chips and a reset link. Hidden filters make users think data is missing.

Working on something like this?

Tell Tithi what you are planning. You get a clear scope, a fixed price and a timeline before any work begins. I reply to every enquiry within 24 hours.

Your details go only to Tithi and are used only to reply to you. No newsletters, no sharing.