Data modelling

Data modelling is the work of turning raw, messy source data into clean, defined tables that everyone can query and trust. Raw events and exports are full of duplicates, odd field names, and inconsistent values; a model resolves all that into a tidy layer, a 'customers' table, a 'subscriptions' table, an 'active user' definition, that means the same thing every time it's used.

For a founder, this is where 'what's our MRR?' stops getting three different answers. The model is where you encode the business logic once: what counts as active, how you handle refunds, where a trial becomes a customer. Tools like dbt make these definitions versioned and testable. Skip it and every dashboard reinvents the logic slightly differently, and you lose hours arguing about which number is right. Model the handful of core entities your whole business runs on, and let everything downstream read from them.