Fragmented CDPs were doubling storage costs and capping growth with per-user pricing, while brittle Python scripts and SQL bottlenecks stalled every campaign. Replaced with a Snowflake + dbt + Hightouch stack that orchestrates 75 audiences across 15+ platforms — for under $100K a year.
Role Led the Composed CDP build — architecture, dbt consent layer, and Hightouch rollout, as consultant
Vendor-native audience tools are siloed—they can't seed lookalikes from unified identity, execute real-time exclusions, or enforce consent across platforms instantly. The enterprise needed a warehouse-native approach to orchestrate all three simultaneously.
Lookalike campaigns require rich behavioral and transactional data from top-tier customers. Vendor builders can't access unified identity across six touchpoints (web, mobile, CTV, billing, offline)—data remains siloed.
When users convert, they must be instantly removed from top-of-funnel campaigns across Facebook, Google, etc. Vendor tools require manual syncs or fragile integrations, causing ad spend leakage.
When users withdraw consent (GDPR/CCPA), they must be removed from all 15+ platforms instantly. Vendor-native audiences lack unified consent enforcement, forcing scattered compliance scripts across systems.
On top of this marketing imperative, three operational failures made the situation worse: broken cost economics, fragile pipelines, and velocity stalled by engineering bottlenecks.
Legacy CDPs like Segment and Tealium price by Monthly Tracked Users (MTUs). The enterprise paid twice—once in Snowflake, once in the CDP's isolated cloud.
Custom Python scripts pushed raw events via HTTP POST. When vendor APIs throttled, scripts collapsed completely—engineers debugged while marketing ran on stale data.
Marketers couldn't operate the warehouse directly. Every audience launch required an engineering ticket with 2–3 week turnaround, killing experimentation.
Build a warehouse-native Composed CDP using Snowflake, dbt, and Hightouch. Deploy Reverse ETL to activate audiences directly from the warehouse, eliminating vendor silos, API brittleness, and the data volume tax entirely.
Migrate segmentation logic and identity mapping into the warehouse using dbt. Bake GDPR/CCPA compliance directly into transformation layers—when users withdraw consent, the pipeline automatically strips them from all downstream cohorts at the source. dbt · Snowflake
Deploy Hightouch as a lightweight Reverse ETL engine on top of Snowflake. Extract only differential data rows and stream changes to 15+ destinations with native API rate-limit management and automated retries. Hightouch · Reverse ETL
Enable non-technical marketers to construct complex cohorts via visual, click-based canvas. Rules translate into optimized SQL running natively in Snowflake—deployment time collapses from weeks to under ten minutes. No-code · Governance
With clean, real-time profiles flowing reliably, build automated rules for smart frequency capping across CRM and paid social. Untargeted users with high-intent signals trigger dynamic paywalls and personalized promotions instantly. Real-time · Orchestration
Developers freed from managing 15+ ad network and CRM API connections. Schema updates, rate-limiting, and consent scripting all offloaded to Hightouch. Focus shifted back to core product work.
Activate data natively from Snowflake. Avoid linear CDP pricing increases. Flat-rate infrastructure costs under $100K annually with complete data ownership.
Marketing designs, validates, and launches multi-channel audiences independently—without engineering tickets or queue delays. Transforms to a truly agile, self-serve growth engine.
75 dynamic audiences across six touchpoints, finally unified. Enables accurate LTV, multi-touch attribution, and real-time behavior-triggered campaigns at scale.
Let's talk
Tell me your biggest data or AI bottleneck and I'll send you how I'd approach it — the same way I worked this problem. No résumé dump, no pitch. Just a useful conversation.
Start the conversation →