Moz IM: shake-to-action audio ads with an AI voice assistant

Audio ads were the one format nobody could interact with. I designed a SaaS platform that lets advertisers add Shake-to-Action and an AI voice assistant right inside a VAST file.

Role
UX/UI Designer
Industry
AdTech / MarTech
Platform
Web
Year
2023

TL;DR

Problem
Audio ads had no clicks, so advertisers couldn’t prove ROI.
Solution
A self-serve campaign builder for interactive audio: shake triggers, voice Q&A, real-time stats.
Impact
campaign setup went from a dev ticket to a 10-minute self-serve flow.

Context & problem

Business goal

Change the audio ad market with Shake-to-Action: interactive triggers embedded straight into VAST files, so a listener can act on an ad by shaking the phone.

User pain

Traditional audio ads are a dead end: people hear them but can’t respond, and generating the tracking files behind an interactive ad is deeply technical.

Why it’s hard

The interface had to make that technical generation simple and, at the same time, give detailed analytics on a physical user interaction nobody had measured before.

My role

UX/UI designer on the platform. I designed the Ad Unit builder, the configuration interface for VOX AI and the analytics dashboard. The voice assistant itself is a product feature; my part was how marketers set it up and measure it.

Discovery & research

Product model

A clear hierarchy: Advertisers → Campaigns → Ad Units. Each Ad Unit links to one of 12 action types, from “Add to Apple Wallet” to “Open website”.

The VOX AI evolution

The product grew to include VOX AI, a conversational voice bot that engages people right after they shake the phone and guides them through complex service scenarios. That added a new layer to configure and a new step to measure.

Design strategy

Hide the technical complexity

The strategy was to mask technical complexity behind a clean, linear workflow: managers move step by step and never have to touch code.

Measure the new journey, not old metrics

Standard ad metrics can’t describe a shake. The analytics had to follow the actual path a listener takes.

Key decisions

1. Making an invisible interaction, a shake, obvious to advertisers who’ve never seen it

A step-by-step Ad Unit builder lets managers set VAST parameters and pick a media action without writing code, with a live preview of the interactive ad in the creative editor.

2. A voice assistant setup flow non-technical marketers can finish alone

Vox Configuration is a sandbox for the AI agent. I designed controls for the bot’s persona, its permissions (for example, calendar or location access) and custom instructions, so the AI behaves predictably in an ad context.

3. A funnel built for shake-to-action

Beyond standard metrics, the dashboard visualizes a unique conversion funnel: listen → physical shake → AI dialog → final action. Custom data visualizations show clearly where the audience drops off at each step.

Final design

Design system & UI

A clean, linear UI runs through the Ad Unit builder, the VOX AI Playground and analytics, with custom data visualizations designed for the shake funnel. The CMS editor works on iPad too.

Outcome & impact

  • Campaign setup: from a dev ticket to a 10-minute self-serve flow
  • The VOX AI Playground let marketing managers deploy complex voice bots in minutes and upsell through dialog instead of static links
  • Advertisers got unprecedented transparency: Shake Rate and the follow-up actions triggered via API are tracked precisely
  • That transparency proved the interactive audio model and took the product from an ambitious MVP to a full tool for high-engagement campaigns with measurable ROI

What I learned

  • A new interaction needs a new funnel: old metrics can’t show where people drop off.
  • Making AI behave predictably is a design problem: persona, permissions and instructions need clear controls.
  • A linear flow is the fastest way to make a technical product feel simple.

Next case

0-network: traffic distribution for media buyers

Next case

Want ad-tech UX that marketers actually understand? Let’s talk.

Get in touch