← Back to work

Pivot Table, Facilio CMMS

Redesigning a powerful analytics tool from the ground up so people would actually reach for it.

Role
UX Designer, end to end
Domain
Enterprise SaaS / CMMS
Platform
Web application
Year
2025
Status
Live in production
Read
7 min
The redesigned Pivot Table with data: Site and Storeroom rows, Item type and Item columns, Tasks and Closed tasks values, and grand totals
Problem
Controls were scattered across V1. Nobody could predict where an option lived.
My role
UX end to end, from domain learning and research to QA support.
Outcome
Usage up, fewer how-do-I questions, and three new features adopted without docs.

The full story

Fold any chapter once you have read it.

01Problem

What was the problem?

V1 worked but was fragmented: formatting, filters and value controls lived in different corners. Conditional formatting and column reordering did not exist.
My role

Domain learning, interviews, competitive analysis, information architecture, interaction design, a UI audit and QA support.

V1 · Before
V1: filters, themes and conditional formatting sit in a toolbar, away from the fields.
Core friction

Nobody could build a pivot view without trial and error, or the implementation team.

02Research

How did I research it?

Learn what a pivot table is, how it's used, and what was wrong with ours.
Fields, the Row, Column and Value zones, and the table they make.
ToolWhat I borrowed
Formatting where the user works.
Conditional formatting and cell visuals, without clutter.
Year → Quarter → Month date hierarchies.
V1 · Before
V1: conditional formatting hides in a column header menu.
V1 · Before
V1: a second home for it, in a Customize drawer.
V1 · Before
V1: filters in a modal, away from their fields.
Existing design

The audit and the interviews agreed on four problems.

4 problems found

  • Pivot controls were scattered, so key features were hard to find.

  • Filters were hard to reach, for quick or advanced analysis.

  • Formatting and customisation were inconsistent for non-technical users.

  • Widespread accessibility issues were identified.

    • Text too small in data-dense views.
    • Button colours failed WCAG 2.1 AA contrast.
Each problem, and the move that answers it.
03Ideas

Which ideas did I try?

The directions, and the one chosen.
  1. Rejected

    Centralised settings panel

    Everything in a side drawer. Rejected: it pulls settings away from the columns they change.

  2. Rejected

    Toolbar-based approach

    Formatting in a toolbar above the table. Rejected: long, and per-field options hid.

  3. Chosen

    Inline field pills with overflow menus

    Each field pill carries its own menu. It matches the mental model and mirrors Power BI.

Final direction

Every field dropped into a zone becomes a pill that shows only its own options. Less to scan, less to remember.

04Solution

What did I design?

The screens, the features and the components.

A value opens its own menu: aggregation, then formatting.

A date expands into Year, Quarter, Month and finer levels.

A field lifts as a chip, and the zones light up as targets.

Applied filters stack as tags, with Clear all.

Quick filter: search, tick, apply.

Formatting opens from the value's own toolbar.

Alignment from the same toolbar.

Sub totals per row and column, each with its position.

View, grid, density and theme in one menu.

The new information architecture.
Quiet on purpose: light borders, row tints, totals at every level.
Design approach

Every control is scoped to its zone, under a fixed filter bar. The table stays the main surface.

05Outcome

What changed after launch?

Discoverability was the biggest shift: with controls beside their context, people explored without guidance. Conditional formatting and date hierarchy, new to V1, became part of regular reporting.
Observed, no V1 baseline
Pivot usage rose after launch
Reported
Fewer how-do-I questions to the implementation team
  • Worked
    • Domain learning first. A real model of pivots grounded every decision.
    • Selective borrowing. The best of three tools beat copying one.
    • The implementation team as research partners. They knew V1's failures.
  • I'd change
    • Test earlier. Concept tests on the pills and filter bar would have shortened QA.
    • Baseline the metrics. Without a V1 figure, the growth is hard to size.