How we work
From source material
to defined outcomes.
Data quality and commercial suitability cannot be established by slogans. We use a staged approach to clarify the purpose, investigate the records, agree the work and document the result.
Start with the purpose
A dataset being assessed for potential acquisition has different questions from a dataset being cleaned for system migration. Before work begins, clarify the audience, intended use, source formats, rights, practical constraints and expected deliverables.
A good first enquiry states the problem and approximate scale. There is no need to upload confidential records before information-sharing and commercial arrangements are discussed.
Four stages of engagement
- 01
Understand
Describe the collection or data problem, assess source definitions and identify the immediate commercial or technical objective.
- 02
Agree
Define scope, rights and permitted uses, outputs, review responsibilities, confidentiality and acceptance criteria before committing to work.
- 03
Shape
Profile data, map fields, review matching rules or acquisition suitability and document transformations and exceptions as relevant.
- 04
Deliver and explain
Handover agreed outputs with field descriptions, validation findings, any identifier mappings and important unresolved questions.
Quality assurance is contextual
A quality check must match the data's purpose. A product catalogue may need unit normalisation and consistent stock identifiers; a historical archive may need temporal context and non-overwritten source attributions; a migration dataset may require valid keys and destination-compatible structures.
For integration, report which sources were matched and which records remain unmatched. For cleaning, compare before-and-after profiling results against the agreed rules. For acquisition, document the rights and limitations on which commercial suitability depends.
Automated checks support human review but cannot settle every ambiguous record or contractual question. Retaining uncertainty is sometimes the most accurate result.
Useful handover materials
A project may result in a prepared data export, mapping tables, source-identifier crosswalks, field dictionaries, a quality summary, a decision log and an exception list. Deliverables differ by scope; not every engagement requires every document.
Browse the practical guides for examples of how a dataset-sale description, an integration plan or a cleaning brief can be structured.
Ready to discuss your data?
Start with a description of the records and what you want to achieve. We can then agree an appropriate next step.
Start an enquiry