How a data team stops being a report factory
Most data teams spend their days not on modelling but on ten variations of the same report. What changes that is not more people — it is less repetition.
There is a shortcut for judging a company’s data maturity: ask how many report requests the data team gets per week, and how many of those are small variations of one another. The ratio is usually surprisingly high.
Removing repetition is cheaper than adding headcount
The same question asked ten different ways is a sign that the business cannot reach the data. Open that access and the request queue shortens; what remains are the genuinely hard ones: lifetime value, churn prediction, price elasticity.
A data team’s success is measured not by the reports it produces, but by the questions it no longer has to be asked.
Our role here is to be the bridge: gather the data into one profile, open it to the business in language it understands, and leave the data team solid ground to work on.
What we do in the first three months
Month one is inventory: which report is fed by which table, and how many different ways each definition has been written. Seeing that "active customer" means three different things to three teams is the first shock in most companies.
Month two is the dictionary: we write a single definition for the thirty most used metrics and bake it into the panel. Once the definition lives in code, the argument ends and everyone sees the same revenue.
Month three is access: we open the screen where the business can ask its own questions, and we watch the data team’s inbox. If the number of requests does not fall, we set something up wrong — and what is usually missing is not a screen but trust: if you cannot see where a number came from, nobody believes the answer to their own question.
