Getting Business Central data into Microsoft Fabric

Business Central is an excellent transactional system and a frustrating one to report on at scale. Once you're across multiple companies, or you want history and cross-entity analysis, querying it directly stops being viable. Getting BC's data into Microsoft Fabric is what changes that — and the pattern for doing it well is now fairly settled.

Why not just hit the APIs?

For a handful of records, BC's APIs are fine. For analytics they aren't: you're limited by what the endpoints expose, you carry the query load on your production environment, and joining across a dozen companies for a group view is painful. You want the data lifted out, once, into a place built for analysis.

The pattern: export, then consolidate

The approach I use is the bc2adls / ADLSE pattern. A lightweight BC extension exports changed records to a data lake on a schedule — incrementally, tracking what's new or modified rather than dumping everything each time. A notebook on the Fabric side then consolidates those exports into clean Delta tables in a Lakehouse. BC stays the source of truth; Fabric becomes the analytical layer sitting behind it.

Incremental is the whole game

Full reloads don't scale. Relying on change tracking so you only move deltas — including handling deletes properly — is what keeps the pipeline cheap and quick enough to run every few minutes. Get this right and near-real-time group reporting stops being a fantasy; get it wrong and you're reloading millions of rows to catch a handful of changes.

Consolidating many companies

The real payoff arrives with multiple entities. Landing every company's tables into one Lakehouse model, keyed consistently, gives you a single group-wide view that BC itself will never show you — one place to see revenue, cash or project position across the whole business.

Keep the lake tidy

Delta tables need housekeeping. Running OPTIMIZE to compact small files and VACUUM to clear old versions keeps queries fast and storage costs sane. It's easy to skip and expensive to ignore — small files quietly strangle query performance over time.

Then, and only then, Power BI

With clean Delta tables in place, Power BI becomes almost boring — in the best way. The hard work has already been done in the pipeline, so reports are fast, consistent and trustworthy, and you're modelling over a proper star schema rather than wrestling live BC data.

The mental model that helps: BC runs the business, Fabric answers questions about it. Keep those jobs separate and each does its own far better.