Bluekens: Fabric and Databricks, Each Doing What It Does Best

Bluekens sells and services Volvo cars across several locations in the Netherlands. Every day brings new cars sold, cars delivered, service appointments planned and work carried out in the workshop, and each of those runs on its own operational system.

Those systems did their jobs, but they did not talk to each other. Sales, after-sales and the workshop each had their own view of the business, and combining them meant exports and spreadsheets. Bluekens wanted one platform for all of it, and one that its own people could run.

We did not choose between Microsoft Fabric and Azure Databricks; we combined them. Fabric became the core of the platform. Its OneLake keeps all the data in one place, Power BI sits right next to it, and its low-code tools let an analyst build and change reports without writing Spark. It runs on a steady, fixed capacity, which suits the daily rhythm of a dealership well.

Databricks does what Fabric is less suited for: heavy data engineering at scale. Its compute scales up when a big job arrives and back down when it is done, so a heavy load does not have to fit inside a fixed capacity. When demand spikes, Databricks takes the load; the rest of the time, Fabric does the work.

Data engineering demand is not flat. Most days are routine, but a historic reload, a new source system or the month-end close can need far more compute than an ordinary day. With Fabric alone you have two choices: keep the capacity sized for an ordinary day and let the spikes queue, or size it for the biggest peak and pay for capacity that sits idle most of the month.

Combining the two removes that trade-off. Switch between the three setups below and watch one illustrative month.

One result the business noticed straight away was in the workshop. The maintenance system and the accounting system used to be two separate worlds: a technician booked a part on a job, and the finance team found out later, after a manual entry or an overnight batch. Until then, stock, margins and open jobs in the books were out of date.

The platform made the two work together seamlessly. Any part booked in the maintenance system is now logged in the accounting system straight away. Book a few parts on the job below and switch between before and after.

Beyond the workshop, we connected the operational systems across Bluekens' sales, after-sales and workshop workflows. Each feeds the platform, and together they give one picture of the business: from a car sold, to its first service, to every part fitted along the way.

A platform only lasts if someone inside the company owns it. We designed this one to be run by Bluekens' in-house business analyst, and delivered it with a clean handover: documentation, a walkthrough of every part, and the analyst working alongside us before we stepped back. Today the analyst owns the platform and extends it day to day.

That is where the combination pays off again. The analyst works mostly in Fabric, close to the data and the reports, while the heavy engineering runs quietly on Databricks in the background.

• Microsoft Fabric deployed and configured as the core data platform.
• Azure Databricks integrated alongside Fabric, with both platforms working in tandem: Fabric for the daily rhythm, Databricks for heavy and spiky data engineering.
• The maintenance system and the accounting system working together: every part booked in the workshop is logged in accounting at once.
• Multiple operational systems connected across sales, after-sales and workshop workflows.
• A clean handover to Bluekens' in-house business analyst, who now owns and extends the platform.