The Challenge
What PurpleX Was Facing
PurpleX needed to deliver a unified attribution model across clients whose marketing data lived in twelve different channel platforms — Google Ads, Meta, LinkedIn, TikTok, programmatic DSPs, email platforms, and affiliate networks — each with different API rate limits, different attribution windows, different event schemas, and different latency characteristics. Building a meaningful multi-touch attribution model required combining events from all channels into a single user journey, which was impossible without a cross-channel identity resolution layer.
The Solution
What We Built
We built a multi-source marketing data pipeline with individual channel connectors managing API authentication, rate limit backoff, and incremental sync for each platform. A probabilistic identity resolution layer matched user events across channels using available identifiers — email hashes, device IDs, click IDs — to construct cross-channel journey sequences. The attribution engine computed configurable multi-touch models (linear, time-decay, data-driven) over the unified event stream. A connector scheduler managed the different data latencies per channel — adjusting when each source was re-queried to balance freshness against API quota consumption.
Results