Skip to main content
← Back to selected work

03 / Internal tools

One planning tool replacing nine spreadsheets

Quantitative task analysis, then a design system the team could extend alone.

Case study cover

Context

Internal tooling / operations planning

Manufacturing operations team, ~200 people using the tool

Company name withheld under NDA.

Role

Me — Task analysis, information architecture, design system

Team — 1 engineer (part-time), 1 ops lead as domain expert

The constraint

Nine separate spreadsheets, maintained by different planners, had quietly become the operation's real system of record — each with its own conventions, and no single one anyone fully trusted. There was no dedicated engineering team; this had to be buildable by one part-time engineer.

What we tried

01

Tried — Interview each planner individually to map their spreadsheet's structure, then unify them into one schema afterward.

Learned — The nine spreadsheets weren't nine versions of the same thing — several tracked genuinely different decisions that planners had bolted onto a shared format out of convenience. A single schema would have lost real distinctions.

Did — Ran a quantitative task analysis across all nine planners' actual usage logs to find which fields were load-bearing versus incidental, before designing any schema.

02

Tried — Design a fully custom system tailored exactly to the identified task structure.

Learned — The scope matched a team of five, not one part-time engineer — the 'correct' design wasn't buildable in the available time.

Did — Cut the design to a smaller core system of record plus an extensible tagging layer, so planners could still capture team-specific distinctions without needing new schema changes from engineering.

03

Tried — Require all nine planners to migrate on the same launch date.

Learned — One planner's team ran on a different reporting cycle, and a hard cutover would have broken their monthly reporting mid-cycle.

Did — Built a CSV import path so any planner could migrate on their own cycle boundary, with the last spreadsheet retiring six weeks after the first.

Outcome

9 → 1

All nine spreadsheets were retired within two months of launch, and the ops lead reported it as the first planning tool the whole team actually trusted as the source of truth.

Arun Jolly Joseph