What success should mean

A platform should not be judged mainly by terabytes stored, connectors deployed, or dashboards produced. More meaningful indicators include:

• time required to discover and understand a relevant asset;
• proportion of critical attributes with explicit lineage;
• percentage of recurring questions answered from governed products;
• mean time to detect and resolve data incidents;
• effort required to onboard a new source into an existing semantic model;
• number of conflicting definitions for the same business concept;
• ability to replay a computation and obtain an explainable result.

We can summarise platform maturity as a reduction in expected answer cost:

where Q is the distribution of legitimate future questions. The expectation is crucial. A platform is an investment in classes of future inquiries, not merely in today’s reporting backlog.

The phrase “too much data, too few answers” describes a structural failure to convert inscriptions into warranted, reproducible claims. The remedy is not indiscriminate centralisation. It is the construction of a computational and institutional layer in which identities, meanings, time, quality, provenance, and permitted use are explicit.

A cultivated Data Platform does not abolish interpretation. It disciplines interpretation. It does not promise that every question has an answer. It makes visible what evidence exists, how it has been transformed, what uncertainties remain, and which conclusions are justified. In this sense, the deepest product of a Data Platform is not a dataset or dashboard. It is an organisational capacity for accountable inquiry.

*Conceptual references

Bateson, G. Steps to an Ecology of Mind (1972). Bowker, G. C., and Star, S. L. Sorting Things Out (1999). Floridi, L. The Philosophy of Information (2011). Shannon, C. E. “A Mathematical Theory of Communication” (1948). Wiener, N. Cybernetics (1948).*