The data conversation has been going on long enough that most organizations think they know how it ends. They do not. The top-line ambition keeps changing — reporting, self-service analytics, machine learning, generative AI — but the actual constraint stays in the same place: the data itself, and the discipline around it. A modernization roadmap that does not start with a clear-eyed read of where data readiness actually stands is a roadmap that overruns.
The Five Dimensions That Matter
Gartner's research on data and analytics maturity converges on a five-dimension frame that holds up in practice: quality, lineage, accessibility, governance, and architecture. Each one is a real constraint. Each one is independently solvable. And each one has a measurable maturity curve that an executive team can read and prioritize against.
Quality Is The Floor
MIT Sloan Management Review's long-running data and analytics research is a useful corrective to the ambition-led roadmap. The productivity ceiling for any downstream system — a BI dashboard, a customer 360, a machine-learning model — is set by the quality of the source data. Teams that skip the quality work and chase the use case end up debugging numbers for months instead of shipping. Quality is not glamorous, but it is the floor everything else stands on.
Lineage Is The Audit Trail
Lineage answers the unglamorous-but-load-bearing question of where a number came from. The Data Management Association's DMBOK is the canonical reference, and the implementation playbook is clear: instrument the pipeline, surface lineage in the catalog, and treat lineage as a product surface that analysts and auditors actually use.
Governance That Does Not Slow You Down
Most data governance programs fail by being either too heavy or too light. The pattern that holds up is the federated one: a small, opinionated central team that owns the platform, definitions, and policies, with domain teams owning the correctness of their own data. McKinsey's research on data operating models traces this pattern across multiple industries and concludes the same thing: governance works when accountability is local and platform is shared.
Architecture That Compounds
The architectural question that matters is not warehouse vs. lakehouse, or Snowflake vs. Databricks. It is whether the architecture compounds: whether each new ingestion lowers the cost of the next, whether each new model reuses the work of the last, and whether the platform can carry both BI and ML workloads without forking. Treat architecture decisions through that lens and the technology choices become much less fraught.
The One-Week Assessment
A useful self-assessment, runnable in roughly five business days. Day one: pick three high-value use cases and trace the data dependencies for each. Day two: score each underlying dataset against the five dimensions on a 1–5 maturity scale. Day three: identify the two or three constraints that are gating multiple use cases. Day four: outline a six-month remediation plan for those constraints. Day five: socialize the findings and reach alignment on what gets funded next.
The output should be a clear answer to the question executives keep asking themselves and never quite getting: are we ready, and if not, what would it cost to be ready?
Key Takeaways
- Data readiness has five real dimensions: quality, lineage, accessibility, governance, architecture
- Quality is the floor — every downstream system is bounded by source quality
- Lineage is a product surface, not a backend artifact
- Federated governance — central platform, local accountability — is the pattern that holds up
- Judge architecture decisions by whether they compound, not by vendor
- Run the five-day readiness assessment before committing the roadmap