Over the years I have been involved in numerous initiatives that were described, at their outset, as data problems. Data quality issues. Reporting inconsistencies. Integration challenges where different systems produced different answers to the same question. In each case, the initial assumption was that this was a technical matter — a problem with data models, data cleanliness, or tooling. And in almost every case, that assumption was wrong.
Data problems are almost never about data. They are about the organisation that produces it.
The Pattern That Repeats
The symptoms are always recognisable: inconsistent reports, missing information, conflicting figures for the same business metric across different parts of the organisation. The initial response is also consistent: improve the data model, introduce better tooling, clean up the datasets. All logical steps — and all, in isolation, insufficient.
In one organisation, multiple reports showed different figures for the same business metric. The obvious assumption was data inconsistency somewhere in the pipeline. We investigated the systems, the interfaces, the transformation logic. The data itself was not fundamentally incorrect. What we eventually found was that different teams were defining the same metric differently. Same name, different calculation — each defensible in isolation, none aligned with the others. No amount of data cleansing would have resolved it, because the problem was not in the data. It was in the organisation's failure to agree a shared definition.
The data was accurately reflecting the organisation. The organisation just did not like what it saw.
The Real Causes
The same underlying causes appear across sectors, organisation types, and data maturity levels. None of them are data problems. All of them produce data problems.
The Technology Trap
The most consistent mistake I have seen in response to data quality problems is the assumption that technology will resolve them. New data platforms. New governance tooling. New architectures. Data lakes, data warehouses, data meshes — each generation of technology promises to solve the problem that the previous generation did not.
Technology can make good data governance easier to implement and easier to sustain. It cannot create the ownership, definitions, and organisational alignment that good data governance requires. A data catalogue populated with assets whose ownership is unclear does not become useful because the catalogue is well-designed. A data mesh with poorly defined domain boundaries does not produce coherent data products because the platform is well-built. The technology exposes the organisational problem — it does not solve it.
The organisations that invest heavily in data technology before resolving the underlying ownership and definition questions consistently find that they have built sophisticated infrastructure for managing data that is still inconsistent at its source.
What Actually Makes the Difference
The improvements I have seen that were both significant and durable shared a common characteristic: they addressed the organisational questions before, or alongside, the technical ones. Not instead of them — the technology matters — but in the right sequence.
Data is not the problem. It is a reflection of the organisation that produces it. If the organisation is unclear about ownership, inconsistent in its definitions, fragmented in its processes, and siloed in its decision-making — the data will be too. With great fidelity.
Fixing the data without fixing the organisation is cleaning a mirror rather than improving what it reflects.
That observation — along with the twelve that preceded it in this series — comes from fifty years of watching organisations struggle with the same underlying challenges: complexity, alignment, decision-making, and the ability to execute consistently over time. The technology changes. The challenges persist. I hope these reflections have been useful to those navigating them.