Data processing
Data processing
From the spreadsheets you have to the output you need, with every rule visible.

How a processing engagement runs
We start with an inventory: each source, its owner, how it is extracted, what each field means and how often it changes. We then agree one valuable output and build a pipeline for it — ingest, validate, clean, join, transform, check, publish — with tests for every rule.
Principles
- Flag, do not guess. Records that fail rules go to an exceptions list for a person to resolve.
- Rules in plain English. Each transformation has a one-line description your team can review.
- Repeatable. Running twice on the same input produces the same output.
- Minimal. Personal data only where essential, handled under a written agreement.
Where it runs
On your infrastructure or data platform wherever possible. We can recommend modest hosting options if you have none, but you remain in control of the environment and the data.
Limits
We do not provide ongoing data hosting as a service, and we do not buy or sell data. Quality of output is bounded by quality of source; we make that boundary visible.