Data model & metrics
We define your business metrics once, so everyone is talking about the same number.
The challenge
Two people ask for "this month's sales" and get two different numbers. Everyone has their own formula, nobody knows which one is right, and meetings are spent arguing about the data instead of deciding. How each figure is calculated is in three people's heads, and leaves when they do.
The solution
We build a data model people can actually read, and every metric is calculated on top of it: one documented, auditable definition per metric. Calendars, hierarchies and role-based permissions so each person sees what belongs to them. When a calculation rule changes, it changes in one place and every report follows.
It's for your company if…
- The margin sales calculates and the one finance calculates do not match, and each department stands by its own.
- Metrics are worked out in a spreadsheet full of formulas only its creator understands.
- Every new figure has to be requested from the person who runs the data and depends on their availability.
- Comparing this year with last year is hard because the criteria have changed over time and nobody recorded when or why.
What's included
Metric dictionary
Every metric with its formula, its data source and the owner of its definition, in a document that stays with the company and is revisited when the business changes.
Calculations you can audit
Every calculation can be opened and read: which tables it starts from, what it filters and what it aggregates. An auditor or a new manager can follow it end to end without asking for help.
A model named after the business
Tables and fields carry business names, not technical codes: customer, order, margin. Whoever arrives later understands what everything is without depending on whoever built it.
Hierarchies, calendars and comparisons
The groupings the business actually uses come ready-made, by area, by product family, by salesperson, with the fiscal calendar and period-on-period comparisons already solved.
Visibility per profile
Each profile sees its own scope: an area manager, their area; management, the whole company. The rules are defined once and apply to every report at the same time.
How we do it
Definition workshop
We bring together the people who use each metric and close its definition: what counts, what does not and since when. Disagreements are settled here, not in the reports.
Building the model
We build the tables and calculation rules on top of the integrated data, and every calculation decision is recorded in the dictionary.
Checking against today's numbers
We compare the model's results with the spreadsheets the company relies on today and explain every difference until there is agreement.
Release and change process
The model goes into production with an agreed process for changing definitions: who can request it, who approves it and how it gets recorded.
Key benefits
One figure, one definition
Auditable calculations
Built to scale
FAQ
We already have working reports. Why would we need a data model?
Because when every report recalculates the metrics on its own, two screens can show different figures. With a shared model, each metric is calculated once and every report reads it from the same place.
Who decides the right definition when two departments disagree?
The company does, not us. We bring the disagreement to a table with all sides, show the consequences of each option, and the decision goes into the dictionary in writing, with a date and an owner.
What happens when we change how something is calculated?
The rule is changed in the model, in one place, and the dictionary records from when it applies. If needed, both versions are kept so periods calculated under different criteria can still be compared.
How much of our time does it take?
The part that cannot be delegated is the definition workshops: the people who know the business need to be there. The rest we do. The assessment sets out how many sessions are needed and with whom.
Related success stories
Manufacturing
Industrial components manufacturer
15% rejection rate with no idea which shift, machine or product it came from
- Issue detection time
- −80%
- Rejects over six months
- −34%
- Quality report for the whole plant
- 1
Financial services
Business group with four companies
A seven-day monthly close and figures that never matched across companies
- Days to close the month
- 7→1
- Board report every month
- D+2
- Margin definition across all four companies
- 1