Data & Analytics · Media & Publishing
For an established independent book publisher, a decade of accumulated ETL, warehouse and dashboard tools replaced with one governed cloud platform, without breaking a single number the business depended on.
Repository not shown as per company policy
0+
ETL pipelines reverse-engineered and migrated off the legacy tool
0
enterprise warehouse in place of four disconnected tools
0
operational applications unified behind one portal
0
environments with CI/CD: development, UAT and production
01Challenge
Four disconnected systems had accumulated over a decade: an ETL tool running more than two hundred largely undocumented pipelines, a warehouse with business logic buried in stored procedures, dashboards isolated from daily operations, and a mountain of claims, allocations and reconciliations run by hand in spreadsheets over email. The same question could be answered four different ways, and access was managed separately in every system.
02Approach
Undocumented logic is still load-bearing logic. Each pipeline was mapped and its output captured before anything was rewritten.
New output was validated against the legacy system domain by domain, so each cutover was a small verified move rather than one large irreversible one.
Reports embedded in the portal the business already used, rather than a separate tool requiring a separate login and a separate license.
03Outcome
Scale had, over time, buried the reporting platform under four disconnected systems and a large manual process run over email and spreadsheets. The program replaced all of it, and the hardest constraint was that every number had to keep matching what the business already knew to be true.
Six layers from source system to end user, each with validation, logging and retry built in rather than bolted on afterward.
Analytics delivered inside the company portal, so reports sit beside the operational apps instead of in a separate tool nobody opens.
Claims, allocations and reconciliations moved out of email into auditable workflows with roles, approvals and a trail.
What changed
Where it started
Every pipeline was reverse-engineered, validated against legacy output, and only then re-platformed.
How it works
Data moves from source systems through orchestrated pipelines into the warehouse, then out through analytics and custom APIs into a single portal, the only place business users need to go.
01 / 04
Inside the system
The engineering was the straightforward part. Not breaking a number the business depended on was the hard part.
Pipeline, warehouse and report testing, with business-rule and financial reconciliation against legacy output.
GuardrailRegression testing and user acceptance testing before every release, not only at the end.
Cloud identity integrated with the corporate directory, role-based access across the portal, row-level security applied environment by environment.
GuardrailPermissions administered centrally, replacing per-system ad-hoc access.
Automated CI/CD across all environments with build validation and smoke tests.
GuardrailDevelopment to UAT to production, every time. No direct changes to production.
Built with
Roadmap
The architecture was built to take on new sources and capabilities with minimal structural change.
Four disconnected systems, replaced by one governed enterprise platform.
More work
A decision-support layer over a staffing platform, turning candidate, requirement and recruiter activity into KPIs that mean the same thing whoever is looking at them.
Reporting, prediction and document processing for a medical data platform, delivered with a cross-functional business intelligence, data engineering and machine learning team.
For an online retailer, an analytics platform that brings order, product and customer data into one governed view, so trading meetings start from agreed numbers.
If something here is close to what you need, tell us about your situation and we will walk you through how we would approach it.
Intelligence → Innovation → Automation → Growth