Studio, the proprietary platform from A Million Ads, lets users build dynamic digital audio ads tailored to listeners by factors like location, weather, date, and time.
But its Arrange View feature, which visualises ad variations on a timeline, was falling short:
A cluttered UI that confused new users
Hard to iterate on due to rigid architecture
Frequent bugs that undermined trust
Design Workshop
I ran a workshop with our Product Manager and engineers to map out assumptions about Arrange View, what users should be able to do, and how we could improve their experience.
The Solution
We uncovered several key challenges to address:
How to visualise audio with hundreds (or thousands) of unique variations
How to clearly show which version of the ad is playing
How to build user confidence that rules (e.g. Weather, Location) are working as expected
How to make testing different rule combinations quick and easy
How to let users edit individual audio blocks (e.g. trim, adjust volume)
I mapped out solution journeys in Sketch, readying them for a more rigorous second round of user testing.
The side-bar problem
Producers typically use Studio on large desktop screens, but adding controls to the new Arrange View proved challenging. My initial sidebar design took up space needed for the main audio timeline, and I also had to work around fixed global controls that weren’t in scope for redesign.
User Testing
I collaborated with engineers to build a lightweight, clickable prototype that was interactive enough to validate the concept without requiring extensive development time.
I tested it with five internal creative producers familiar with Studio. While I’d hoped to include client-side users, their limited availability made this difficult. Below is an example output from one of these sessions.
Testing Arrange View had its challenges. I needed feedback on key features without over-investing in ideas that might not land. To balance realism with efficiency, I mocked up high-fidelity individual screens instead of full flows, just enough to make the functionality feel believable.
For example, when exploring how to display audio variants, I needed to validate:
Was a dropdown the right interaction pattern?
Was it clear that variants could be repositioned via drag-and-drop?
Would users need to filter or search through variants?
Was limiting the number of visible variants the right approach?
Prototype
Selected screens from the prototype I tested with participants:
