Carry the source, date and uncertainty into the next use

A point can identify a surveyed building position, a postal-area centre or an unverified coordinate, and those meanings are not interchangeable. I ask what the task requires before deciding how records should appear together. In the 2018 view, identical markers suggested that all 180 positions supported building-level screening. Source review showed otherwise. The correction needed both clearer representation and restrictions on use; a different colour alone would not make a coarse position suitable for a detailed assessment. I work with surveying and business professionals on the evidence required for the intended purpose. Data stewardship preserves the provenance and unresolved cases so that completeness of a coordinate field is not mistaken for accuracy of the property location or permission to draw an insurance conclusion from it.

A location may be unchanged while the boundary used to classify it changes. For an earlier reporting period, applying today’s map can therefore produce a different answer without any movement of the property. I ask which source version and effective period the reporting question requires, and preserve the relationship used to produce the original account. Reconstructing history is not an instruction to conceal corrections. It is a way to distinguish a later boundary change from a real change in the underlying population. Where provenance cannot establish the appropriate version, the answer should remain unresolved until the responsible reporting owner decides how to handle it. A current geographic convenience should not silently become a statement about what was true in an earlier period.

A modelled boundary can be drawn sharply even when input, location and modelling uncertainty affect how nearby properties should be interpreted. I work with relevant specialists to make that uncertainty visible rather than design a binary display that suggests more certainty than the evidence supports. A sensitivity exercise describes behaviour under its defined assumptions, not every possible future event. The resulting classes need a purpose and clear limitations. Data colleagues can preserve those definitions and identify unstable cases; we cannot certify hydrological behaviour or determine a claim from a map intersection. The important leadership decision is whether the proposed use remains justified, particularly when the useful output is a set of cases needing further assessment rather than a single confident classification for every location.

Location information supports decisions made by underwriting, actuarial, claims, legal and other professional owners. My function should describe what this information establishes, how it was obtained and which uses require further assessment. It should not turn a map category into a policy entitlement or a statement of expected financial loss. I make the operating expense and unresolved work visible in the recommendation so that the limitations do not disappear under pressure to deliver a finished dashboard. My own approval of an over-precise display showed why this responsibility belongs at leadership level. The CDO needs to ask what the user is likely to infer, not only whether the source fields arrived correctly. A technically populated dataset can still communicate a meaning that the available evidence cannot support.