The data journey, from source system to a number you can trust
· Data
Every figure in a planning model has travelled a long way. Map that journey well and the rest of EPM gets easier.
Before anyone budgets, forecasts or reports a single number, that number has to get into the system. In my experience this data journey is where most EPM projects succeed or fail, long before anyone notices. A beautiful cube on top of messy, untrusted data is just a faster way to be wrong.
Here is how I like to think about the journey, stage by stage.
1. Source
It starts in the operational systems: the ERP, the CRM, payroll, spreadsheets that someone maintains on a Friday afternoon. These systems were built to run the business, not to plan it, so the data rarely arrives in the shape finance needs. The first job is simply to know your sources: what they contain, who owns them, and how often they change.
2. Extract and stage
Pulling data straight into a planning cube is tempting and almost always a mistake. I prefer a staging layer, a place where raw extracts land untouched so you can always trace a number back to exactly what the source sent. In Jedox this is the Integrator's job: connect, extract, and stage before anything is transformed.
3. Transform and validate
This is where raw data becomes meaningful: mapping account codes to a common chart, aligning entities and currencies, splitting or combining dimensions, and applying business rules. Just as important is validation. Do the totals reconcile to the source? Are there nulls where there should not be? A load that fails loudly is far better than one that silently loads half the data.
4. Load
Only now does data enter the model. A good load is repeatable, scheduled, and logged, so that next month it runs the same way without a person nursing it. The goal is boring reliability.
5. Trust
The final stage is the one people forget: trust. A number is only useful if the people using it believe it. That comes from lineage (you can show where every figure came from), from consistency (the same question gives the same answer), and from transparency (the logic is documented, not hidden in someone's head).
Why it matters
When the data journey is solid, everything downstream gets lighter. Budgeting conversations stop being arguments about whose number is right. Forecasts can be rerun in minutes instead of days. Reports reconcile on the first try. The platform finally pays for itself.
So before reaching for the clever modelling, spend the time on the plumbing. Map the journey, stage the raw data, validate hard, and make the whole thing reproducible. Get that right and trust follows, and trust is the real product.