AI, data and governance consulting
Data governance
The state of the data feeding any AI initiative: origin, quality, ownership and traceability.
Why it comes first
Until now data informed human decisions, and an error was filtered in the reading. A system that decides and executes has no such step. If an organisation does not trust its data for a board report, it cannot trust it to a system that acts on its own.
Most AI projects that fail do not fail because of the model. They fail because nobody could say where a figure came from or who was accountable for it.
What we organise
The inventory of sources, the owner of each domain, quality rules, and traceability from origin to use. Also which data should not be feeding a system at all, which tends to be the uncomfortable part.
We work from DAMA and whichever data protection requirements apply, and the result is embedded in the procurement and development cycle: a rule that does not change how things are bought changes nothing.
How it is measured
With verifiable evidence: an auditable register of sources, owners designated on record, and quality controls that leave a trace when they run. Not with a maturity statement.
Frequent questions
- Do we have to fix all our data before starting with AI?
- No, and waiting for that is the most common way never to start. You organise the domain feeding the prioritised use case, and extend from there.
- Is this a technology project?
- Rarely. The hard part is agreeing who is accountable for each domain. The tool comes later and is the easy part.
- Who should lead it?
- Someone from the business with authority over the domain, supported by technology. Led by technology alone, it ends up as a technical inventory nobody uses.
- Can it be done without a CDO?
- Yes. The function is required, not the job title: someone who decides on the data and answers for it.