The situation
Data quality was handled reactively, incident by incident, and was invisible to the stakeholders who depended on it. There was no shared definition of what "good" meant, so there was no way to say whether it was improving, and no way for anyone other than the person handling the incident to assess it.
What I did
- Authored a data quality strategy across nine dimensions: accuracy, completeness, consistency, timeliness, validity, uniqueness, reliability, relevance and accessibility.
- Attached KPIs to each dimension so quality became a tracked number rather than a subjective judgement.
- Wrote a communications plan, because a quality framework nobody outside the data team sees does not change stakeholder trust.
- Productized a repeatable AI-assisted assessment workflow and documented it so any teammate could run the same assessment and get comparable results.
- Added freshness, volume and schema monitoring on the critical models so degradation was detected rather than reported by a stakeholder.
What it means for you
Data quality cannot be improved before it is defined. A framework with named dimensions, KPIs against each, and a process someone else can run turns quality from one person’s vigilance into something the function owns.
Context
These are from eight years owning the data warehouse and reporting function of a consumer marketplace platform, in-house rather than as an outside consultant. The employer and the internal system names are withheld; the numbers are the real ones.