Deep dive
Start from the content you can license
A general music service needs licences for both the recordings and the underlying songs in every country it serves, along with reporting obligations and minimum guarantees that make launch costs high. For most new products, a narrower catalogue is far more realistic: podcasts you produce or host, audiobooks, guided meditation, language or exam audio courses, devotional music, regional labels or independent artists who upload directly.
The content model shapes the software. Owned or directly licensed content needs simpler rights logic and lighter DRM. Label catalogues need territory rules, detailed play reporting and royalty accounting. Decide this in discovery, because it changes the admin, the reporting and the budget.
It also changes the launch plan. A podcast or course platform can launch globally on day one, while a label catalogue may be limited to the territories you have cleared. Independent artist uploads bring their own work: identity checks, content ID matching to catch re-uploads of copyrighted tracks, and payout onboarding. Each path is valid, but each needs a different first release, so the product and legal plans should be written together.
Building a player that never stops
Listeners judge an audio app on one thing: does it keep playing. That means a foreground service on Android with the right notification controls, proper audio session handling on iOS, ducking for navigation prompts, resuming after calls, handling Bluetooth disconnections and preloading the next track for gapless playback. Cross-platform frameworks wrap the native players, but the edge cases still need native knowledge.
Test on the phones and headphones your audience uses, including budget Android devices with aggressive battery management. Our mobile app testing device lab covers exactly these scenarios.
- Cache the next track and the current one so short network drops are invisible.
- Offer a data saver setting with a lower bitrate rendition.
- Record play events locally and sync them later for accurate reports.
Recommendations without a giant data team
Early on, editorial playlists and simple rules work well: new releases from followed artists, more from the same genre, popular in your city. As listening data grows, collaborative filtering and embeddings produce personalised mixes. Spotify's scale of recommendation research is not needed to delight a niche audience; clean play events and a sensible model go a long way.
Store every play, skip, save and search as an event in a warehouse from the first release. The same events feed royalty reports, creator dashboards and recommendations, so the investment serves three purposes. Our data engineering team usually sets this pipeline up during the MVP build.
Subscriptions and store rules
Premium plans can be sold on the web, through the App Store and through Google Play. Each channel has its own fees, trial mechanics and refund flows, and rules about telling users about cheaper web prices have changed in some countries since 2024. Keep a single entitlement service as the source of truth and use a subscription platform such as RevenueCat to reduce store integration work.
Family and student plans add membership management and eligibility checks, which is why they sit in the scale tier. Plan pricing by region early, because purchasing power differs widely between markets. In price-sensitive markets such as India, short plans, UPI payments and occasional top-ups often convert better than long subscriptions, so make plan length a configuration rather than a fixed assumption.
Running costs and growth
Audio files are much smaller than video, so delivery costs are modest compared with a video service. The main running costs are content licences or revenue shares, storage and CDN, search hosting, subscription fees from stores and payment providers, and maintenance at roughly 15-20% of the build cost per year.
Growth investments usually follow listening time: car and speaker integrations, offline improvements, better recommendations and social features. Each new surface adds maintenance, so add them when data shows your audience uses them. Our application maintenance and support team can run the platform while you focus on content.