ARCHITECTURE. LEVEL 2.
The platform runs the campaigns. The warehouse remembers everything.
At some point your questions outgrow the platform UI: joining customer data with margin, comparing cohorts across years, feeding a model. The answer isn’t replacing the CDP. It’s the triangle: tracking flows into Bloomreach, raw data flows onward into BigQuery or another warehouse, and computed results flow back into Bloomreach for activation.
THE TRIANGLE
Three corners, each doing what it’s best at.
Corner 1: Tracking. Your site and apps send events to Bloomreach through its SDKs. Corner 2: Bloomreach Engagement. Profiles, segmentation, real-time triggers, and campaign execution. Corner 3: The warehouse. Raw event data flows from Bloomreach into the warehouse, where it joins your orders, margin, logistics, and offline data. The return path: computed attributes, scores, and audiences flow from the warehouse back into Bloomreach as imports, where campaigns and personalization act on them.
WHAT THIS SOLVES
Board-level reporting, cost control, and data science, all in one place.
Board-level reporting: marketing data joined with finance data, in the BI tool the company already uses. Cost control: full history in the warehouse at warehouse prices, not platform prices. Data science: churn scores, LTV models, and propensities computed on raw data, then activated through campaigns. AI context: the warehouse holds the full, joined history that AI needs, and where a large language model can answer plain-language questions across all your data. One version of the truth: the warehouse becomes the referee when platform numbers and finance numbers disagree.
WHAT WE BUILD AT LEVEL 2
The export pipeline, the warehouse model, and the return path.
The export pipeline, the warehouse model on the receiving end, the return-path imports, and the monitoring that tells you when any leg of the triangle stalls. Plus the governance: which system owns which attribute, so the triangle doesn’t degrade into two databases that disagree. We work with your data engineers or bring the pipeline work ourselves for BigQuery-based setups.
WHEN LEVEL 2 IS WRONG
A warehouse with no consumer is a standing cost.
If nobody in the company writes SQL and reporting needs fit the platform UI, the warehouse is a standing cost with no consumer. Stay at level one; revisit when the first analyst is hired. And if multiple departments beyond marketing need the customer data asset, look at level three instead.
Questions your platform UI can’t answer?
Book a discovery call. Bring one of those questions. We’ll sketch the triangle that answers it.