Learn
What is Data Modelling?
The process of designing how your data is structured and related – setting up tables, fields, and connections so your warehouse answers business questions quickly and accurately.
What it is
Data modelling is the blueprint for how information is organised in your warehouse or database. It defines what tables exist, what columns each table has, and how tables relate to each other. Good modelling means your finance team can answer "what was gross margin by product line last quarter?" with a simple query. Bad modelling means that same question takes three joins, a spreadsheet, and a prayer. Think of it like designing the floor plan of an office. You could throw desks everywhere and let people figure it out, or you could plan departments, meeting rooms, and corridors so that people who work together are near each other and everything flows logically.
Why it matters for your business
Poor data modelling is the hidden reason many reporting projects disappoint. The dashboards look fine, but the numbers are wrong because the underlying tables were not designed to handle edge cases – refunds counted twice, customers appearing in multiple segments, revenue recognised in the wrong month. A $40M SaaS company we worked with had their MRR calculation off by 12% because their data model did not properly separate new revenue from expansion revenue. It took three months of board reports before anyone caught it – and by then, the forecast built on those numbers had driven two bad hiring decisions.
How we approach it
We model data around business concepts, not source system structures. Instead of mirroring the tables in your CRM or ERP, we build models around the entities your team thinks in: customers, orders, products, revenue. We use dimensional modelling for reporting (fact tables for transactions, dimension tables for attributes) and document every business rule – what counts as a "customer," when revenue is "recognised," how returns are handled. Every model is version-controlled in dbt so changes are tracked and reversible.
Related Services
Key Takeaways
- •Data modelling is the blueprint that determines whether your reports are fast and accurate or slow and wrong.
- •Model around business concepts (customers, revenue, orders) – not around how your source systems happen to store data.
- •Every business rule should be defined once, in the model, not reinvented in each spreadsheet or dashboard.
- •Poor modelling is usually the root cause when dashboard numbers do not match the finance team's figures.