A governed, multi-account Data Mesh Lakehouse on AWS turns fragmented retail data into automated, self-serve
Customer-360 and RFM segmentation analytics.
About the Customer
The customer is a leading retail and consumer-goods group in the Dominican Republic, operating multiple
commerce banners with a broad national footprint. As an Enterprise-segment retailer, it generates high transaction volumes across diverse operational systems, including Magento, SAP, Zendesk, Instaleap, and Icomm.
Customer Challenge
The retailer lacked a unified, scalable view of customer lifecycle dynamics, unable to distinguish new, recurrent,
paused, and reactivated customers across its commerce banners, which reduced the precision of marketing and
loyalty initiatives. It also had no scalable insight into cross-product purchase behavior, preventing its commercial
teams from reliably designing bundling, product recommendations, or cross-sell strategies. Compounding both
issues, data was dispersed across multiple heterogeneous source systems and formats (CSV, JSON, Parquet, GZIP), and propagating fine-grained governance permissions across accounts added further operational complexity. Left unaddressed, these gaps translated directly into imprecise marketing spend, missed loyalty and retention opportunities, and lost revenue from underdeveloped cross-sell and bundling programs.
Partner Solution
Strata Analytics designed and implemented a multi-account, Data Mesh Lakehouse on AWS following the AWS
Serverless Data Lake Framework (SDLF), organized across four purpose-built AWS accounts. An
Exchange/DevOps account serves as the entry point, where AWS Lambda and AWS Step Functions collect raw data from the retailer’s operational systems into an Amazon S3 sources bucket. Separate Transformation accounts for development and production process this data under the SDLF framework, giving Strata’s team CI/CD-driven agility to onboard new datasets quickly.
A dedicated Domain account custodies datasets organized by business domain (Sales, Customers, Logistics, and
others), with each domain functioning as an independent data product. An Amazon EventBridge rule triggers a Step Functions-orchestrated crawler that verifies dataset completeness before invoking AWS Glue to catalog new data, and AWS Lake Formation governs this Domain account, applying column- and row-level security and creating resource links to consumer accounts.
A Consumption account hosts Amazon QuickSight, with Amazon Athena serving as the serverless SQL engine that lets business users query shared data without provisioning dedicated compute. On top of this foundation, Strata delivered lifecycle-aware RFM (Recency, Frequency, Monetary) segmentation per business unit and a
Customer-360 view, together with product-association insights for cross-sell and bundling, built on Amazon S3 with Apache Iceberg staging tables for ACID guarantees, time-travel, and low-cost incremental data handling.
Results and Benefits
The retailer’s commercial teams moved from manual, fragmented reporting to automated, self-serve dashboards in Amazon QuickSight, with direct access to per-business-unit RFM segmentation and a unified Customer-360 view for the first time. Product-association insights now surface commercially relevant cross-sell and bundling
opportunities at scale, without requiring manual rule design.
Marketing campaigns built on the platform’s RFM segmentation were measured against a control group across
mailing campaigns launched in the first half of 2026: segmented campaigns converted at a 15% rate, compared to 5% for the unsegmented control group, a threefold improvement directly attributable to RFM-based targeting.
Before this project, the retailer also had no integrated view across its five commerce banners, so any cannibalization
between them, a promotion or campaign in one banner drawing sales away from another rather than generating net-new revenue, could not be detected or measured. Because the Lakehouse now maintains shared customer segments and a Customer-360 view across all five banners, commercial teams can identify overlapping segments and cross-banner purchase patterns for the first time, and design marketing and loyalty initiatives that account for cannibalization risk instead of being blind to it. The solution is live and in production use today.
About the Partner
Strata Analytics is a cloud consulting and systems integration partner delivering data engineering, analytics, and
AI/ML solutions on AWS for enterprise retail clients across Latin America. Strata holds the AWS AI Agents
Competency, awarded March 19, 2026.
