The problem
Every report tells a different story.
The same question returns different figures depending on who asks it.
History is lost when master data, price lists or structures change.
Every new analysis needs another extract from the operational systems.
Our contribution
A data warehouse needs a precise definition of what each row represents. We establish the grain of each fact — an order line, a transaction or a period balance — and the dimensions used to interpret it. Customers, products, organisation and calendars need compatible definitions across processes. Design starts with analytical questions and available sources, avoiding joins that multiply values and aggregations that conceal important differences in the underlying records.
History is part of the model. We distinguish when an event occurred, when information was valid and when it reached the system. We decide which changes to retain and how to handle corrections, late data and changing hierarchies. Mappings, keys and reconciliations document the path from sources to the analytical structure. These make it possible to check totals, trace measures to their origin and explain how a change affects historical views and comparisons.
What the work covers
Facts and grain
We define events, balances and levels of detail, specifying which measures can be added and across which dimensions.
Shared dimensions
We organise common keys, master data and hierarchies, handling unmatched records and differences between system codes.
History and validity
We define history and late-data rules, distinguishing business changes from corrections to the information describing them.
Source reconciliation
We check counts and values through the model, making filters, transformations and exceptions visible.
What takes shape
Practical work products, scoped to the engagement.
- The problem
- The analysis
- The result
| # | Work product | What it contains |
|---|---|---|
| 01 | Analytical model | A fact and dimension schema with defined grain, relationships, keys and measures. |
| 02 | Mappings and history | Source mappings and rules for validity, historical changes and corrections. |
| 03 | Consistency checks | Repeatable integrity, count and total checks, with differences traced back to source records. |
Six stages
The assessment path
Check, with Kimball’s dimensional approach, that the analytical data foundation holds up to different questions over time.
A bounded engagement, with scope and fee agreed before work begins. If you go ahead with the project, the assessment fee is deducted from the follow-on engagement.
| # | Stage | Current-model map | Model and load findings | Initial action plan |
|---|---|---|---|---|
| 01 | Requirements: business processes, recurring questions, users. | yes | no | no |
| 02 | Source-system analysis: keys, history, rate of change, quality. | yes | no | no |
| 03 | Model review: facts, dimensions, grain, history handling, conformed dimensions. | yes | yes | no |
| 04 | Load review: lineage, reconciliation checks, performance, recovery. | no | yes | no |
| 05 | Shared definitions and ownership: glossary, metadata, data owners. | no | yes | no |
| 06 | Priorities and an initial action plan. | no | yes | yes |
What you receive
- A
- Current-model map
- B
- Model and load findings
- C
- Initial action plan
Two ways to start.
01 · Free data check-up
Start with a data check-up.
About an hour online, a written summary within five working days. No cost, no obligation.
- 01RequestAsk for the check-up through the contact form.
- 02Online meetingAbout an hour to trace systems, sources, reports and weak points.
- 03Written summaryWithin five working days: current situation, issues and recommended path.
- The situation
- The issues
- The path
02 · Service assessment
Go straight to the assessment.
If you already know this service is the one, we start with the six stages below.
- 01Requirements
- 02Source-system analysis
- 03Model review
- 04Load review
- 05Shared definitions and ownership
- 06Priorities and an initial action plan