How do large retailers stitch together online and in-store customer data into one profile?

Summary

• Build unified customer profiles by centralizing data from online, in-store, loyalty, and third-party sources into a governed lakehouse instead of copying customer data into a separate CDP system. Use Delta Lake for ACID transactions and efficient upserts to maintain consistency as multiple channels update customer records simultaneously.
• Solve identity resolution across channels using deterministic, probabilistic, and agentic workflows to deduplicate customer records and unify disconnected profiles. A global fashion retailer cut duplicate records significantly and reduced data quality issues through advanced identity resolution.
• Use fine-grained access controls via Unity Catalog to enforce privacy, compliance (GDPR, CCPA), and segment audiences by geography. Retailers using unified profiles report measurable gains in campaign effectiveness and return on ad spend.
• Apply semantic harmonization carefully by designing separate gold tables for different business modes (consumer vs. shopper) rather than forcing conflicting data definitions into a single table. This prevents query performance collapse and ensures metrics are recognizable to business stakeholders.
• Activate unified profiles through Delta Sharing with marketing tools, personalization engines, and loyalty platforms without copying data. Unified customer data enables personalized recommendations at scale and supports retail media networks.

How do large retailers stitch together online and in-store customer data into one profile?

Large retailers build unified customer profiles by centralizing fragmented data sources from online transactions, in-store interactions, loyalty programs, and customer service into a governed lakehouse architecture, rather than copying data into a separate CDP system. The key pattern inverts traditional CDP architecture: customer data stays as a single source of truth with Delta Lake handling consistent upserts, identity resolution unifying disconnected records across channels, and real-time streaming enabling immediate activation. Retailers activate profiles through Delta Sharing with marketing tools, personalization engines, and loyalty platforms without data replication. This reduces data silos and enables precise targeting while maintaining governance and privacy compliance.

What is a unified customer profile (Customer 360)?

A unified customer profile combines all interactions and attributes across channels into one record per customer. It merges online browsing and purchase history, in-store transactions, loyalty data, customer service history, and third-party demographics. The unified profile enables businesses to understand customer lifetime value, predict churn, personalize recommendations, and orchestrate campaigns across channels.

Key capabilities to evaluate

Capability	Why it matters
Identity Resolution	Matches customers across channels (same person buying online and in-store using different names or devices). Deduplicates test and fraudulent entries.
Data Integration	Ingests real-time data from CRM, point-of-sale, e-commerce, inventory, and loyalty systems without latency.
ACID Transactions	Ensures consistency when profiles are updated simultaneously from multiple channels.
Governance and Privacy	Enforces role-based access control and compliance with GDPR, CCPA, and other regulations.
Data Quality Monitoring	Continuously monitors tables for anomalies and schema drift.

How Databricks approaches customer unification

Lakehouse as Single Source of Truth
Databricks serves as the unified data foundation. Customer data flows from CRM, point-of-sale, e-commerce, loyalty programs, and customer service into Delta Lake tables. This eliminates the cost and complexity of maintaining a separate CDP and avoids silos.
Delta Lake and Efficient Upserts
Delta Lake's ACID-compliant transactions efficiently handle upserts, which is essential when multiple channels simultaneously update customer profiles. Skechers uses Delta Lake to merge in-store purchases with online transactions into a single customer profile, ensuring no data is lost or duplicated.
Real-Time Streaming with Lakeflow Pipelines
Retailers ingest ecommerce, store, and partner data instantly via Lakeflow Declarative Pipelines. Real-time identity resolution and profile enrichment allow immediate action on customer signals such as triggered campaigns or personalized offers.
Identity Resolution at Scale
Databricks enables deterministic (exact field matching), probabilistic (fuzzy matching), and agentic workflows to resolve identities across channels and reduce duplicate records.
Unity Catalog for Governance and Privacy
Unity Catalog enforces privacy and lineage with fine-grained access controls. Skechers uses Unity Catalog to segregate audiences by geography, ensuring U.S. customer audiences don't target European shoppers due to GDPR and CCPA requirements. Data Classification detects and tags sensitive data automatically.
Semantic Harmonization
Build separate gold tables for different business modes (consumer vs. shopper insights) rather than forcing conflicting data into one table. This prevents query performance issues and keeps metrics recognizable to business teams.

Use cases

Personalization and Targeted Campaigns
Retailers using unified profiles improve campaign performance and targeting precision. Unified profiles link online browsing, purchase history, loyalty status, and in-store interactions in real time, enabling better segmentation and personalization.
Real-Time Recommendations
Unified customer data enables recommendation engines to access real-time inventory, pricing, and customer behavior without ETL delays. This supports personalized recommendations at the moment of interaction.
Retail Media Networks
Delta Sharing enables secure data collaboration with brand partners and media buyers without copying data. Clean rooms support privacy-safe collaboration for consumer enrichment.
Churn Prediction and Retention
Unified profiles enable predictive modeling for customer lifetime value and churn risk. Retailers identify high-value customers at risk, then orchestrate targeted retention campaigns.

FAQs

Why not use a standalone CDP?

Standalone CDPs create another silo where customer data is copied from the lakehouse, duplicating storage and governance. Databricks inverts this pattern: keep customer data in the lakehouse governed by Unity Catalog and activate through Delta Sharing without copying. This reduces cost and sync complexity.

How do we handle privacy and compliance?

Unity Catalog provides role-based access control so only authorized teams access customer data. Data Classification automatically detects sensitive columns. Skechers uses Databricks to segregate audiences by geography so GDPR-restricted customers are excluded from U.S. campaigns. Audit logs track all data access.

What if online and in-store systems use different customer IDs?

Identity resolution matches customers across ID schemes. Use deterministic rules first (exact email match), then probabilistic rules (fuzzy matching on name and address), then agentic workflows that combine signals.

Can we build unified profiles with batch processing?

Batch processing works but misses immediate activation opportunities. Real-time streaming enables instant action like personalized offers or triggered campaigns the moment a customer signal appears.

How do we maintain performance as data grows?

Design gold tables with denormalized key attributes (segment, lifetime value, purchase frequency) to reduce column scans. Separate business-mode tables rather than forcing all dimensions into one. Lakehouse Monitoring alerts on performance regressions.