Case study 05 · Fabric data engineering

HR Data Platform & 12-Month Workforce Cost Forecasting in Microsoft Fabric

Automating workforce data ingestion and cost forecasting through a Bronze/Silver/Gold Lakehouse.

Domain
HR & Finance
Platform
Microsoft Fabric Lakehouse
Planning horizon
12-month rolling forecast

The challenge

HR and Finance needed a reliable way to forecast employer costs across companies, functions and future time periods. Manual reporting and source-system extracts made it difficult to maintain history, model joiners and leavers consistently, and produce a repeatable 12-month cost view.

Data goal

Automate ingestion of employee and vacancy data from an HR REST API into a structured, reusable Fabric platform.

Planning goal

Create forward-looking salary and employer-cost forecasts while preserving historical salary changes for temporal analysis.

What was delivered

  • An end-to-end Microsoft Fabric data platform using a Bronze/Silver/Gold Lakehouse architecture.
  • Incremental REST API ingestion with watermarking, retry logic and Azure Key Vault integration.
  • Scalable transformation pipelines using Fabric Data Pipelines, PySpark, Spark SQL, T-SQL and Delta Lake.
  • SCD Type 2 historisation to preserve salary and employee changes over time.
  • Forecasting logic for salaries, bonuses, statutory contributions, vacancies, joiners, leavers and multi-currency payroll.
  • Business-ready outputs supporting 12-month rolling workforce cost forecasts.

Business value

  • Enabled HR and Finance to forecast employer costs 12 months ahead.
  • Removed manual HR cost reporting through automated ingestion and transformation.
  • Preserved full salary history, enabling more accurate trend and temporal analysis.
  • Created a reusable Fabric foundation for workforce analytics rather than a one-off report.

Technology

  • Microsoft Fabric
  • Lakehouse
  • Delta Lake
  • PySpark
  • REST API
  • Azure Key Vault

Commercial takeaway

A source-to-outcome Fabric story: API ingestion, medallion architecture, historisation and planning logic all connect directly to a measurable business use case.

Planning data you can forecast from

We build the ingestion, historisation and modelling layers that turn operational HR and finance data into forward-looking plans.