Two big buttons: twice‑as‑fast scheduling in 3,000+ cinemas

The problem

The interface dated from 2005. For fifteen years engineers added features with no design oversight.

Projectionists ranged from tech-comfortable managers to operators in their 60s who had run projectors for decades. There was no budget for upfront research.

What I found

Five to six years of field feedback, and the support engineers who handled it, were research we already had.

After launch I got visits funded: about 20 projectionists in 8–10 theatres. People with 2–3 days on the new system disliked it. People with 1–2 months preferred it. Judge a change after it settles.

The things users never mention are often the things they love most.

Key decisions

Dark by default

The booth is dark. The old bright screen was a lamp in a dim room.

Status always visible

What's playing, time left, and whether everything is healthy, without opening a menu.

Format tags

Automatic 2D, 3D and Atmos tags, so nobody reads file names to find out.

Keep show times

Theatres plan intermissions, food sales and parking around end times. We kept approximate times despite time drift and dynamic ad playlists.

Recover from power cuts

An alert and an automatic seek to 5 seconds before the interruption. This was unsolved before.

How it's built

Built to survive

The project ran through COVID theatre closures, team attrition and a Windows to Linux migration that paused development. A component library, design tokens and documented user stories kept it moving.

Names that fit

Film package names differ by only a few characters, so they show full width by default, with resizable columns where they can't.

Small team, long project

One front-end engineer and one QA at a time. I carried the decision history for five years and wrote it down so others could.