I started in a Belgian insurance group as a
data steward. For two years the work was
the records that wouldn't reconcile — fields that
disagreed across systems, customer histories that split
in two, cases where the data was technically valid and
still wrong. Alongside it: GDPR compliance on customer
data, and quarterly data quality audits across business
units.
Every figure in a database is the end of a chain. A rule
written a decade ago. A market with three players instead
of thirty. A person filling in a form who didn't quite
understand the question. Data quality work is where those
chains become visible, because it is where they break.
Two years as an analyst taught me the
other half: pulling the data the business needs —
SAS for commission and mass
payment runs, Salesforce for
customer data — and preparing the datasets that feed the
KPIs. That
side of the job is about deciding which parts of the
context a business can actually act on, and saying so
clearly enough that someone can make a decision from it.
A number is never only a number. It is the last step of
something that happened in the world.
My current work joins the two. I coordinate a
process consolidation project — routines
that had never been standardised are pulled into one
simple workflow — and I prepare the data that feeds the
KPIs. That means framing the requirement with
fifteen people from different functions,
then holding a data definition firmly enough that a
figure means the same thing in one department as in
another.
My strength isn't knowing everything in advance. It is
finding what needs to be known quickly, understanding the
context around it, and turning that into something
usable. That is also where
slim.lu and
Arbitoria come
from: I define the problem, the rules and the validation
criteria, and direct AI tools to build them.