From manual ad ops to a platform behind 40% of revenue

The problem

Planning which ads play on which screens, tracking inventory and running campaigns were stitched together by hand across a disconnected ERP workflow.

There were few patterns to borrow. Nobody had a standard way to schedule ads across thousands of screens.

What I found

Media agencies, theatres and our operations team needed different depths of the same data.

The design was easy to expand and hard to stop. Our first complete version was too big to build.

Key decisions

Co-own the definition

With the PM I ran personas, market and competitor analysis, interviews with Qube's media agency wing, and wrote the user stories. Our roles overlapped on purpose.

Ship fewer things, well

We kept the workflows that had to work on day one and moved the rest to later releases.

Build the design system with it

Pixel was built alongside Slate, so the platform launched consistent.

How it's built

Designed with engineering

Scheduling logic, the inventory model, analytics and operations data were worked out with the engineers who built them, from the concept stage.

Stories as spec

The user stories I wrote covered the edge cases the scheduling engine had to handle.