Turning a music recognition app into a growth engine
Owned mobile roadmap for the SoundHound music recognition app. Built a continuous research engine and PLG experiments that lifted MAU 12% YoY and quarterly revenue 15%.
+12%
YoY+15%
QoQ+20%
DownloadsProblem
The SoundHound music recognition app had a loyal base but flat growth. Roadmap decisions were driven by feature requests and feel, not a steady stream of customer insight, and the app had no real product-led growth motion — monetization sat behind a generic paywall with limited experimentation.
Context
- Two platforms (iOS, Android), managing both the core music recognition app and the voice AI assistant app under a shared core team.
- Mature category with a dominant competitor — net-new install growth was expensive.
- Inherited a roadmap heavy on stakeholder asks, light on instrumented bets.
Team & scope
- Reported to
- Director of Product Management
- Team
- Led a single core cross-functional team (1 designer, 1 researcher, 2 testers, 1 data analyst, 5 backend engineers) with dedicated, separate frontend iOS/Android engineers for each of the two apps.
- Directly owned
- End-to-end product ownership of both the music recognition app and the voice AI assistant mobile app.
- Influenced
- Cross-team alignment with SoundHound's broader NLU platform bets and enterprise integrations.
Approach
Built a weekly research engine
Partnered with a dedicated UX researcher to run a continuous discovery cycle — interviews, surveys, and usability tests every week. Insights flowed directly into roadmap reviews, so prioritization arguments started from evidence instead of opinions.
Launched In-App Purchases as the PLG foundation
Replaced the static paywall with an IAP architecture that supported pricing, packaging, and placement experiments. Wired up an experimentation framework and BI tooling so every change had a measurable hypothesis.
Ran a steady cadence of growth experiments
Tested onboarding flows, paywall placements, value-prop copy, and surface ordering against MAU and revenue. The bets that won got codified; the losers shipped learnings, not noise.
Connected mobile work to the Voice AI platform bet
Supported translating restaurant menus from Toast, Clover, and Square into SoundHound's NLU engine, keeping the mobile roadmap aligned with the broader Voice AI restaurant platform story.
Research insight
Continuous weekly research kept surfacing jobs-to-be-done that didn't yet exist in the category. Several of those bets shipped on my roadmap 4+ years ago — and have only recently appeared in competitors' apps. The research was sound; the constraint was sequencing, not signal.
Decisions & tradeoffs
- Held the line on shipping fewer, better-instrumented experiments instead of a long feature list.
- Prioritized retention surfaces before acquisition spend — the leak in the bucket had to close first.
- Said no to a 'redesign the home screen' push until research validated the underlying jobs to be done.
The tradeoff I defended
What I'd do differently
Sequenced the big research-backed bets first instead of stacking smaller wins. The competitive validation eventually came in — but leading with the headline bets would have proved the research and the roadmap much faster.
Tools & methods
- User interviews
- Usability testing
- SQL analytics
- Experimentation framework