Two sides, one ecosystem

Redacted operates a two-sided marketplace built for both artists and fans. Creators pay for premium hosting and analytics, and use Redacted to distribute music to rival platforms like Spotify, YouTube, Apple Music, TikTok, keeping 100% of external royalties.

On the listener side, a massive ad-supported free tier funnels fans into paid subscriptions. Binding the whole ecosystem together is their Fan-Powered Royalties model, where a subscriber's money flows directly to the artists they actually stream.

A bold promise that needed bulletproof infrastructure

Redacted was preparing to roll out a landmark 100% distribution royalty plan. This was a bold commitment to creators that demanded a bulletproof delivery pipeline.

An audit of the existing infrastructure revealed five critical gaps that made such a rollout impossible without intervention.

Zero observability

No centralized tracking for the delivery lifecycle from track creation to DSP arrival.

Blind telemetry

No visibility into success and failure rates, end-to-end speed, or delivery retry efficiency.

Reactive operations

DSP acknowledgments weren't integrated, so the Ops team learned of failures only after the fact.

YouTube sequencing errors

Content ID and Art Track deliveries suffered from processing-order failures at scale.

Spotify batch failures

Large batch architecture meant one bad track could fail an entire delivery batch.

From reactive systems to predictable delivery

Redacted partnered with Tarka Labs to:

1. Achieve full end-to-end transparency in the DSP pipeline for engineering and Ops teams.
2. Enable proactive Ops workflows to resolve issues before creators noticed.
3. Boost delivery reliability for key DSPs (Spotify, Apple, YouTube).
4. Ensure platform confidence for the 100% royalty launch.

We didn't just fix the pipeline. We rebuilt confidence in it.

We built a comprehensive observability + reliability framework while re-architecting critical delivery components.

Core focus areas are:

1. Full pipeline visibility
2. Real-time monitoring and alerting
3. DSP-specific reliability improvements
4. Failure isolation and recovery

From bottlenecks to breakthroughs

Five targeted interventions

End-to-end pipeline instrumentation

Mapped and instrumented the entire delivery pipeline to capture granular metrics like success/failure rates, delivery speeds, and retry efficiency across all DSPs.

DSP acknowledgment integration

Engineered a system to ingest delivery acknowledgments directly from DSPs, feeding this data into observability dashboards so Ops could see external delivery states proactively.

Real-time alerting framework

Implemented live alerts on DSP delivery metrics to instantly notify the team the moment latency spikes or failure thresholds were breached.

YouTube sequencing fix

Investigated and resolved the ordering logic for YouTube Content ID and Art Track deliveries, ensuring proper, deterministic sequencing under high volume.

Spotify single-item batch architecture

Decoupled Spotify's large batch model into single-item batches, eliminating the cascading failure risk so one broken track no longer takes down an entire delivery job.

The results speak - so do the creators

1. Achieved 99%+ delivery success rate across DSPs.
2. Enabled 100% observability across all integrations.
3. Successfully launched the 100% distribution royalty plan.
4. Strengthened creator trust and platform reliability.
5. Shifted Ops to a proactive monitoring model.
6. Isolated failures and prevented cascading breakdowns.
7. Reduced latency and retry inefficiencies.
8. Improved scalability for high-volume pipelines.