Turn data into answers.
Your numbers live in ERP systems, specialist applications and Excel. We bring them together and build reports your business teams can work with. The architecture follows the task.
question How are costs developing per site? sources ERP system · business application · Excel lists volume manageable, refreshed daily path SQL → validated tables → report tooling chosen for the task, not the licence
Architecture that fits.
Good BI architecture starts with the task. Not every reporting project needs a platform built for 50 terabytes.
Often, the real work is bringing manageable datasets together, building clear models and making useful reports available with a reasonable amount of effort. We look beyond the Microsoft ecosystem and consider options such as DuckDB or PostgreSQL where they suit the requirements.
We do not sell licences, we do not resell cloud services and we are deliberately not a Microsoft partner. Our recommendations follow your data, your questions and the people who use and run the solution. The result is an architecture whose benefits and ongoing effort fit your project.
- Data sources and business questions
Where do the numbers come from, and which decision should they support? - Data volume and freshness
How much is it really, and how current does it need to be? - Concurrent use and permissions
Who works with the reports, and when? - Operations, team and future development
What can your team run and extend over time? - Total effort throughout the lifecycle
Building it is only part of the cost.
Small datasets can have demanding requirements. A large platform needs a clear purpose.
From the source to the report.
Architecture & assessment
Review goals, data sources, models and processes together. Decide between central and decentralised reporting, make bottlenecks visible and explain the next steps.
Reviewing an existing BI environment
Look at sources, loading processes, data model, reports, performance and maintainability together. The result is a documented recommendation: what works and what should change.
Data integration & warehousing
Load data from source systems regularly and with checks, keep its history and bring it together. Processes and structures your team can follow.
Reporting & analysis
Build reports around the questions business teams need to answer. Users keep working in the tools they know, often Excel or Power BI.
Self-service on a reliable foundation
Self-service tools make business teams more agile. Underneath, they need a maintained data model with history and checked data quality.
Improving existing solutions
Connect new sources and support system changes: when a source system changes, reports often stop fitting. Improve data processing and queries where it matters.
What we bring from BI projects.
- No one-size-fits-all BI
Every environment has its own sources, questions and teams. That shapes the solution, not a template. - Plan for system changes early
A new ERP or business application almost always affects reporting. Planning for it early saves rework. - Complement your team, not replace it
We contribute architecture experience and work alongside the people already on the project. - Documented from requirement to delivery
So the solution stays understandable without us.
booking lines in a central data warehouse, analysed by business teams directly in Excel
German Environment Agency (UBA) · 2009–2011instead of three days for recurring reports, thanks to several hundred automated rule sets
ZDL, data office of the German state finance ministers · 2012From published customer references, 2011 and 2012; the technology reflects its time. We have also presented our BI experience at the international SQLBits conference, for example on the BI stack, scale-out and Master Data Services (2010–2013).
Small components. Well-designed data flows.
Almost all of our own analytics tools use DuckDB in whole or in part. That practical experience also informs how we approach data preparation and reporting with manageable complexity.
- Explore files directly
Use SQL to query CSV, JSON and Parquet and combine data from different sources. Test a concrete business question against a defined dataset. - Make preparation repeatable
Express transformations as clear SQL steps and publish a validated dataset for reporting. Loading status, error handling and refresh are part of the process. - Choose the right frontend
Excel, Power BI or a focused web application: getting information to users is part of the architecture. For Power BI, we consider import or DirectQuery according to the connector, alongside refresh and shared-operation requirements.
01 sources business systems & files 02 preparation SQL & DuckDB 03 foundation validated tables & Parquet 04 use reports & analysis
The connection and operating model need to be validated for the specific project. Technical background: DuckDB & Parquet and the open Power Query connector.
Start with a useful question.
We do not start with a platform decision, but with the question your report needs to answer.
- 01
Understand the task
Which decision should the report support? Which numbers are missing or do not reconcile?
- 02
Test with real data
Trace a focused path from source to analysis and check that the data makes sense to the business.
- 03
Plan the next step
Agree what works, how the solution will be operated and which extension is worth making.
What question is behind your report?
Bring your data sources, your question or an existing solution. Together, we can find a sensible first step.