A CSRD platform consists of six layers, each with its own complexity. The ingestion layer pulls data from ERP (SAP, Oracle, Microsoft Dynamics), HR (Workday, AFAS, ADP), energy management (sub-meters, energy suppliers, building management systems), procurement, logistics and external sources such as EcoVadis scores or CDP disclosures. We use dbt as standard for transformations, Airflow or Prefect for orchestration, and Fivetran or Airbyte for the SaaS sources.
The warehouse layer is an append-only data warehouse, such as Snowflake, BigQuery or hardened PostgreSQL, in which raw measurement data, its provenance and its timestamp are recorded immutably. On top sits a calculation engine that applies emission factors (DEFRA, IPCC, sector-specific databases), performs unit conversions and runs allocation models for Scope 3 categories. In addition, a materiality engine captures double materiality: financial materiality and impact materiality, with threshold logic and review workflows.
The narratives layer supports the writing that ESRS requires: policy statements, transition plans, and TCFD-aligned climate scenario analyses. LLM augmentation helps teams update last year's narratives with this year's data without the author starting over. Finally, the output layer delivers EFRAG-compliant XBRL tagging, a PDF report for stakeholders, and an API for your investor relations website or investor portal. A data engineering platform as the foundation ensures the same source data can also feed other analytics and compliance questions.