Unifying customer data across a fragmented retail ecosystem

Business context

A national Australian retailer with both physical and digital channels had grown quickly through acquisition and organic expansion. The result was a technology landscape where customer data lived in at least twelve different systems — a legacy POS, two different e-commerce platforms (one being sunset), an email service provider, a loyalty programme database, a call centre CRM, and several marketing tools acquired by different teams at different times.

The business knew their customers were shopping across channels, but had no reliable way to understand that behaviour. Marketing was running campaigns based on incomplete data, the loyalty programme couldn’t recognise online purchases, and the executive team was making investment decisions based on channel-level reporting that didn’t reflect actual customer value.

The challenge

This wasn’t just a technology problem — it was an organisational one. Multiple teams owned pieces of the customer relationship, each with their own tools and data. Previous attempts to consolidate had failed because they’d been framed as IT projects rather than business transformation. The marketing team didn’t trust the data they had, the engineering team was stretched maintaining legacy integrations, and the analytics function was spending 80% of their time cleaning data rather than analysing it.

The brief was to design a customer data strategy that would give the business a unified view of its customers, enable personalised marketing at scale, and create a foundation for future AI-driven capabilities — all without disrupting existing operations during the transition.

My role

Engaged initially to deliver a Transformation Blueprint — a clear, actionable plan for the customer data platform strategy. When the blueprint was approved, extended in an advisory capacity to guide the implementation through its first phase.

What I did

Discovery and data mapping (weeks 1–3). Mapped every customer data source, identified overlap and conflict between systems, and documented the current data flows. This revealed that the business had roughly 2.4 million customer records, but only about 60% could be confidently deduplicated. The rest existed as fragments across systems with no shared identifier.

Strategy and architecture (weeks 3–6). Recommended Segment as the customer data platform, Snowflake as the analytical data warehouse, and Braze replacing the existing email and push notification tools. The architecture was designed to be event-driven — every customer interaction (online, in-store, email, loyalty) flowing through Segment as a canonical event stream, creating a single source of truth.

The key design decision was to implement identity resolution at the CDP layer rather than trying to clean up data at the source. This meant value could be delivered quickly while progressively improving data quality upstream — rather than waiting for a perfect data foundation that would never arrive.

Vendor selection and negotiation. Ran structured evaluations for both the CDP and marketing automation layers. For CDP, the shortlist included Segment, mParticle, and Tealium. For marketing automation, Braze, Iterable, and Klaviyo. Led the commercial negotiations, securing terms that included implementation support and data migration assistance.

Implementation guidance. During the advisory phase, provided architectural oversight, reviewed integration designs, and helped the engineering team navigate the complexity of connecting twelve source systems to the new CDP. Worked with the marketing team to define their first personalisation use cases, ensuring the technology investment was tied to measurable commercial outcomes from day one.

AI readiness foundation. A deliberate part of the architecture was establishing the data infrastructure needed for future AI capabilities. By centralising customer events in Segment and analytical data in Snowflake, the business now had the clean, structured data foundation required for product recommendations, churn prediction, and lifetime value modelling.

Systems and platforms

Segment as the customer data platform and canonical event stream; Braze for lifecycle marketing automation and cross-channel engagement; Snowflake as the analytical data warehouse; Looker for self-service business intelligence; Shopify Plus as the primary e-commerce channel feeding the event stream.

The architecture was event-driven by design — a deliberate choice that made the system extensible as new data sources or downstream tools were added.

Outcomes

The business achieved a unified customer profile across all twelve previously siloed data sources. Marketing campaign setup time fell from two weeks to two days as the team gained direct access to clean, segmented audiences without engineering involvement. Email revenue attribution improved 65% through better personalisation informed by cross-channel behaviour. The data team gained self-service access to customer insights for the first time — shifting from data cleaning to actual analysis. The architecture also established the foundation for AI-driven product recommendations, which the business subsequently built on in the 12 months following the engagement.