Unified Governance Across 3,000+ Retail Touchpoints

Situation

L’Occitane operates a global luxury retail network spanning over 3,000 points of sale across more than 90 countries. The Salesforce footprint had grown organically over years: Sales Cloud managing B2B wholesale accounts, Service Cloud handling customer support across regions, and Marketing Cloud powering loyalty and CRM campaigns. Each cloud operated as a standalone system with no unified view of the customer.

The result was three separate identities for the same person. A customer who purchased in-store in Paris, bought online in Tokyo, and contacted support in New York existed as three unconnected records — with different IDs, incomplete histories, and no cross-channel context. Marketing campaigns were built on 30-40% of available customer data. Store associates had no visibility into online purchase history. Service agents resolved cases without knowing a customer’s product portfolio or recent touchpoints.

Across 12 APAC markets, the situation was further complicated by regional Salesforce implementations that had diverged from the global architecture. Each country had developed their own integration patterns, field customizations, and data models. The result was a multi-cloud environment that was technically connected but operationally fragmented.

Diagnosis

The core architectural problem was an absence of a single customer data layer. The three Salesforce clouds used different identifiers for the same customer — email in Marketing Cloud, account ID in Sales Cloud, case contact in Service Cloud — with no systematic identity resolution between them.

Integration architecture was batch-based and point-to-point: Marketing Cloud synced with Sales Cloud nightly via a scheduled job; Service Cloud pushed case data to Sales Cloud via a custom-built REST API; Commerce data sat entirely outside the Salesforce ecosystem. This approach created data latency (campaigns worked on 24-48 hour old data), brittle dependencies (when one integration broke, downstream clouds went stale), and no foundation for real-time activation.

The multi-cloud data model had also accumulated years of inconsistency. Field naming conventions differed across clouds. The concept of “customer” was modeled differently in each system. There was no canonical data dictionary, no data ownership framework, and no governance process for schema changes.

For Agentforce readiness — a strategic priority on the roadmap — this data foundation was insufficient. Intelligent agents require a complete, current, unified view of the customer. You cannot build AI that actually performs on fragmented, 24-hour-delayed data.

Action

Phase 1: Data Cloud as the Customer Data Hub

The architecture decision was to implement Salesforce Data Cloud as the central customer data layer, rather than building yet another point-to-point integration between clouds. Data Cloud would ingest from all three Salesforce clouds plus external sources, resolve customer identities, and provide a unified profile as the system of record.

Data streams were configured from Sales Cloud (account and contact data), Service Cloud (case history, interaction logs), Marketing Cloud (campaign engagement, email behavior, preference data), and the e-commerce platform via MuleSoft-managed API connections. Ingestion frequencies were set by data type: high-frequency streams for transactional events (purchases, case opens), daily batch for historical records.

Identity resolution required designing a multi-attribute matching ruleset. Customer matching used a hierarchy of: email address (deterministic, primary), phone number (deterministic, secondary), loyalty card ID (deterministic, tertiary), and behavioral clustering (probabilistic, for anonymous web profiles). The resolution model reduced 3+ million duplicate customer profiles to a unified identity graph.

Calculated Insights were built on top of the unified profile: customer lifetime value (combining online + in-store + wholesale), churn risk score (engagement decay + support frequency), product affinity model (category-level purchase clustering), and NPS trajectory (linking survey scores to behavioral signals).

Phase 2: Multi-Cloud Activation

With the data foundation in place, the activation layer was architected across three surfaces:

Marketing Cloud was reconnected to Data Cloud for real-time segmentation. Segments previously built on Marketing Cloud’s local contact data were rebuilt on Data Cloud Unified Profiles, providing 2.5x more complete customer attributes for targeting. Activation latency dropped from 24 hours to near-real-time.

Experience Cloud powered a new store associate mobile application, delivering a Customer 360 view at the point of sale. Associates could see the customer’s complete purchase history (in-store and online), active loyalty tier, recent support interactions, and personalized product recommendations — all drawn from the unified Data Cloud profile.

Service Cloud was enhanced with Data Cloud context cards, surfacing purchase history and product ownership directly in the case management interface. Agents no longer needed to request customer history manually or switch between systems.

Phase 3: Governance and APAC Harmonization

A data governance framework was established to prevent re-fragmentation. This included a canonical data dictionary (field naming standards, object ownership, definition of customer, household, and account), a change control process for schema modifications, and data quality dashboards measuring completeness, accuracy, and freshness across clouds.

For the 12 APAC markets, a rationalization process aligned regional field customizations to the global data model, established regional data stewards, and implemented monitoring to detect drift from the canonical model.

Result

The multi-cloud architecture delivered a unified customer view across all 3,000+ L’Occitane retail touchpoints — in-store, online, and service channels — for the first time. Identity resolution reduced customer duplication by 85%, creating a single customer graph from which all activation could operate.

Marketing campaign performance improved materially. With 2.5x more complete customer data and near-real-time segmentation, conversion rates on targeted campaigns increased by 40% in the first two quarters post-launch. Segment build time dropped from hours to minutes.

Store associate adoption of the Customer 360 application was high. Associates with access to complete customer history closed 15% more incremental sales through relevant product recommendations. Service case resolution time dropped 25% as agents had full context without escalation requests.

The architecture also established the data foundation required for Agentforce deployment — a planned next phase. With a unified customer profile, complete purchase and service history, and real-time data activation, L’Occitane’s Salesforce environment is now positioned to support AI agents that can make context-aware decisions across the customer lifecycle.

Technologies used: Salesforce Data Cloud, Sales Cloud, Service Cloud, Marketing Cloud, Experience Cloud, MuleSoft (e-commerce integration), Identity Resolution Engine, Calculated Insights, Data Cloud Activation