THE PROBLEM
Records mixed operational meaning, organisational structure and technical implementation.
Similar words referred to different concepts; different words referred to the same concept. Roles, organisations, functional areas, applications and information products were mixed in free text, while referential integrity was weak.
A dashboard could visualise the ambiguity. It could not resolve it.
MY CONTRIBUTION
Preserve source truth while creating target governance.
- Profiled and classified the legacy ecosystem before redesign.
- Separated entities, relationships, roles, hierarchies and controlled concepts.
- Designed logical and physical 3NF relational models and semantic mappings.
- Built integrity checks, provenance and a migration approach focused on loss prevention.
- Produced end-to-end information-flow views, redesign documentation and handover material.
THE ARCHITECTURE APPROACH
Meaning first. Structure second. Migration third.
Treat semantics as architecture
Resolve what each term, role and information object represents before designing tables.
Separate source truth from target governance
Retain what the legacy record said while mapping it to governed concepts.
Make integrity testable
Express relationships and migration rules so missing, duplicate or contradictory mappings can be detected.
Design for repeatable analysis
Create structures that support trusted reporting and future analytics without reinterpreting meaning each time.
Build handover into the design
Document assumptions, mappings and decision logic for organisational continuation.
OUTCOME
A governed foundation other teams could continue.
- Controlled and traceable concepts across fragmented sources.
- Normalised relational structures with enforceable integrity.
- A repeatable migration path with provenance and loss checks.
- Logical, physical and end-to-end information-flow views.
- A maintainable redesign and handover package.
WHY THIS MATTERS ELSEWHERE
AI readiness starts with governed meaning.
The same pattern appears in acquisitions, regulatory reporting, data-platform consolidation, RAG initiatives and multinational programmes: information exists, but meaning and ownership are inconsistent.
Relevant capabilities: enterprise data architecture · semantic modelling · relational design · data profiling · controlled vocabularies · lineage · migration assurance.