Data Platform Modernization

Lakehouse rebuild ended weekly reporting disruptions and took core ETL from daily to every four hours.

Fourteen key tables re-architected on Databricks / AWS — adopted across five business units; legacy warehouse fully retired in ~2 years.

Setup

Client: National consumer referral service
2-year engagement
3 consultants in yr 1, ~1.5 in yr 2
Scope: ETL logic, semantic layer, use-case migration

Situation

All company data lived on a legacy warehouse carrying years of undocumented ETL and semantic-layer logic. On a once-daily ETL, failures came roughly weekly — blocking reporting until midday — with multi-day outages every few months. With a CRM replacement underway, the client needed its data foundation rebuilt, not lifted and shifted.

Approach

Year 1

Untangle & Rebuild

Traced years of layered ETL and semantic logic relied on by stakeholders in every business unit; rebuilt 14 key tables — spanning marketing and sales funnel, partner network, and sales-force management — governed and documented on Databricks / AWS.

Year 2

Mitigate & Retire

Partnered on adoption across five business units, migrating the long tail of use cases — operational files, reporting, real-time models — until the legacy warehouse was fully deprecated.

Impact

24h ⇨ 4h

core ETL refresh cadence — a 6x fresher foundation for reporting and models

1 / Week ⇨ ~0

reporting disruptions — no full-pipeline failures on the new foundation

5 BUs

adopted firm-wide; years of technical debt retired with the legacy warehouse

Hours between core ETL refreshes (lower = fresher data)

Legacy

24h

Lakehouse

Adopted by Marketing, Sales, Finance, and Partner Network — later extended to Product — with governed, documented tables replacing years of undocumented intermediate logic.