Modern Measurement Architecture

Modern measurement spans collection, processing, warehouse, activation, reporting. This module covers the full architecture, the customer-ID strategy, and the modernization path from legacy analytics.

1. The 2026 architecture

The modern measurement architecture has these layers: event collection (client + server), event processing (server-side container, CDP), warehouse (BigQuery / Snowflake), activation (CDP, reverse ETL), reporting (GA4, looker, internal BI). The trend: consolidate around the warehouse as source of truth.

2. Customer ID strategy

• First-party user_id (created at signup or anonymous-to-known transition).
• Hashed PII (email, phone) for vendor matching.
• Device IDs (mobile).
• Click identifiers (FBCLID, GCLID).
• Persisted in first-party cookie and server-side store.

3. CDP integration

Modern CDPs (Segment, mParticle, RudderStack, Tealium, Adobe RT-CDP) sit alongside server-side tagging. The relationship: CDP collects events, server-side container forwards to ad/analytics destinations, warehouse stores everything for analysis and modeling.

4. Real-time vs batch

• Real-time: clicks, page views, conversions (sub-second to seconds).
• Near-real-time: enrichments, derived events (seconds to minutes).
• Batch: warehouse loads, modeling, reporting (minutes to hours).

5. Event schema

Standardized event schema across all sources is foundational. Components:

• Event name (e.g., page_viewed, product_added).
• Standard user properties.
• Standard contextual properties.
• Event-specific properties.
• Versioning for schema changes.

6. Reverse ETL

Reverse ETL moves data from the warehouse back to operational systems (ad platforms, email, CDP). Examples: Hightouch, Census. Use cases: warehouse-derived audiences sent to Meta and Google, computed traits sent to email, modeled scores activated in CRM.

7. Privacy at architecture level

• Consent enforcement at the data layer.
• Differential treatment by jurisdiction.
• Data retention policies.
• Customer-data deletion workflows.
• Audit logs for compliance.

8. Warehouse integration

BigQuery, Snowflake, Databricks as the analytics source of truth. Direct event-streaming into the warehouse via the server-side container, CDP, or dedicated pipeline tools (Fivetran, Airbyte). Modeling, attribution, and audience-building happen in the warehouse.

9. Cross-domain and cross-device

Stitching: cross-domain (multiple owned domains), cross-device (web + mobile + offline). Strategies: first-party ID propagation, customer-match in destinations, device graphs (LiveRamp, Tapad).

10. Build / buy / hybrid

11. Modernizing legacy

1. Audit current stack and data flow.
2. Define target architecture.
3. Set up parallel infrastructure.
4. Migrate one event stream at a time.
5. Validate data parity.
6. Decommission legacy.

Part of the GTM Server-Side series · RGM Training