← Selected Work
Global Music Label · NDAMedia & EntertainmentBuild · End-to-endPublic · 10,000+ staff

Fan-engagement intelligence for a global music label.

The first fully customisable campaign-intelligence platform tailored to every artist — segmenting 200M+ fans, generating brand-true content, and timing every message with ML.

Campaign intelligence platform
Client
One of the world's largest music groups
Ownership
End-to-end (UI→AI→data)
Fan profiles
200M+
Stack
React · BigQuery · Kafka · Feast
Models
K-means · LightGBM · LLM
Reach
15+ markets & timezones
The challenge

Fan engagement is not one-size-fits-all. Every artist builds their brand differently — some through email, others through social or live experiences — and key indicators vary by genre, market and audience. Existing tools couldn't adapt fast enough, forcing marketers into rigid dashboards and guesswork.

Our approach

We designed and delivered the first fully customisable campaign-intelligence platform tailored to the unique identity of every artist — combining ML fan segmentation, an LLM content engine, ML send-timing, and a human-in-the-loop review workflow.

What we built

Fan Intelligence — 200M+ fan profiles unified across social, email and SMS, with real-time streaming, social, purchase and direct-comms signals served from a feature store. AI Content Engine — LLMs auto-generate subject lines, email body and SMS, calibrated to artist brand voice, fan segment, language and market; human-reviewed before every send. Smart Delivery — ML predicts optimal send time per fan, event-triggered across 15+ markets, with A/B gating and fatigue suppression. Closed feedback loop — interactions re-score fan profiles daily.

Tech stack
AI / ML
K-means + RFMLightGBM propensityLLM content (multi-agent)ML send-time model
Data
Feast feature storeBigQueryApache Kafka200M-record ingestion
Application
React web appPer-artist campaign dashboardDrag-and-drop builder + AI assistant
What we delivered
  • A customisable campaign-intelligence platform per artist
  • ML fan segmentation — 30+ segments, re-scored daily
  • An LLM content engine, brand-calibrated & multilingual
  • ML send-time orchestration across 15+ markets
  • A human-in-the-loop review & A/B-testing workflow
  • End-to-end: UI/UX, frontend, backend, AI/ML, data & handover
Timeline
Discovery
Per-artist journeys & KPIs
Build
Segmentation · content · timing engines
Delivery
Platform + clean handover to client DevOps
Data at work
200M
Fan profiles unified
30+
Behavioural segments, daily
15+
Markets & timezones
Outcomes
200M
Fans personalised
30+
Daily ML segments
NDA
Impact values withheld
Next case →Vietnam's largest streaming service — 0 to 10M+.

Have something
that needs to ship?

Tell us the problem. We'll tell you honestly what makes sense — even if that's not us.

Start a conversation