Most data strategies and modernization programs begin with an inventory: what we have, what we're missing, what we should buy. It feels rigorous. It is backwards.
Data modernization should begin with the decisions the business needs to make, not the technology it wants to buy.
A decision is the unit of value. A readmission prevented, a supplier switched earlier, a price set with confidence. If a data investment doesn't change a decision someone makes, it doesn't create value, however elegant the platform.
Design top-down, build bottom-up
We use one picture to keep every program pointed the right way. Read it top-down when you design. Build it bottom-up when you deliver.
We start here: the decision
- WhyBusiness OutcomesRevenue, efficiency, risk, customer experience, decision speed
- WhatData & AI CapabilitiesStrategy, analytics, AI, automation, decision intelligence
- What it rests onTrusted Data FoundationGovernance, quality, ownership, security
- HowTechnology & ArchitecturePlatforms, integration, data products
Most programs start here: the technology
Business outcomes say why. Data & AI capabilities say what the business must be able to do. The trusted data foundation says what those capabilities rest on. Technology & architecture say how, and come last.
What changes in practice
The first workshop is about decisions, not systems. The modernization roadmap is ranked by value and feasibility, not by platform readiness, so the estate becomes AI-ready where AI will actually be used. And every phase has an outcome measure agreed before it starts, so the program can prove what it changed.
How this goes wrong
- The platform is chosen first. The decision it was meant to serve is discovered afterwards, and the architecture is retrofitted.
- Every layer is optimized alone. A strong platform team, a strong analytics team, and nothing that connects their work to an outcome.
- The roadmap is a document. No owners, no sequence, no measures: a report, not a strategy.
- Value is assumed. Nobody agreed the outcome measure before the work started, so nobody can prove it afterwards.