Data Normalisation

Overview

Data normalisation is the discipline of structuring data so that it is clean, consistent, and free of duplication. Done well, it is invisible — the business simply trusts what the system tells it. Done badly, it haunts a system for its entire lifetime in the form of reconciliation errors, duplicate customers, contradictory reports, and slow, painful migrations.

One of Our Deepest Specialties

This is something we are genuinely best-in-class at. Two decades of designing transactional schemas, warehouse models, and reporting layers — we have seen every form of badly-normalised data and we know how to fix it.

Whether the data lives in a relational database, a document store, a spreadsheet, or scattered across half a dozen line-of-business systems, the same fundamentals apply — and we apply them rigorously.

What We Do

• Entity-relationship modelling that reflects the real business domain
• Third normal form (3NF) database design as the default starting point
• Denormalisation where genuinely required for performance — never by accident
• Master data management strategies across multiple systems of record
• Deduplication strategies, including fuzzy matching where exact keys do not exist

Common Problems We Fix

• Duplicated customer records across systems — the same person appearing three times
• Address fields that disagree between billing, shipping, and CRM
• Reference data hardcoded into multiple places, drifting apart over time
• "The same field but slightly different" — product codes, account numbers, status flags

Our Approach

We start by understanding the business entities — customers, products, transactions, properties, whatever your domain is — and then we model them in a way that is honest about what they actually are.

That honesty is the part most projects skip. It is also the part that pays back, every day, for the rest of the system's life.