Fabric × Business Central

Consolidating several Business Centrals: the difficulty is never the connection.

The APIs are documented and stable. What gets stuck are the charts of accounts that differ, the dimensions used differently from one entity to the next, and the FlowFields that return nothing.

Business Central SaaS instances consolidated onto one model

The number changes, the model stays one

The case

The same customer, several codes

Four companies in the same group, four Business Central instances, and the same customer coded differently in each one. We add an alignment step before any consolidation: the references come together, and the model returns a single figure.

Le même client, quatre codes différents

Brancher quatre bases est facile. Les faire tomber d’accord est le vrai travail, et c’est l’étape que nous ajoutons avant toute consolidation.

The most frequent case among our clients, and the most underestimated when quoting. We have taken up to seven Business Central instances and seven CRMs onto one model. At that volume we export with bc2adls rather than through the API: the load is incremental and does not consume your ERP quotas.

7
Business Central instances consolidated onto one model: 7
CRMs aligned in parallel, on the same project: bc2adls
Incremental export, without consuming API quotas: 11
Expert consultants

Our speciality

Three alignments that hold a consolidation together

The recurring difficulties of a multi-Business Central consolidation
Gap	What happens when it is neglected
Diverging charts of accounts	Each entity evolved its chart its own way. Consolidating without a mapping table produces aggregates that mean nothing, and nobody notices before the close.
Dimensions used differently	Dimension 2 is a cost centre in one subsidiary and a sales channel in another. The consolidated report then adds up quantities that have nothing to do with each other.
Item and customer reference data	The same customer exists under three different codes in three entities. With no reconciliation, any ranking by customer is wrong.

The technical point

FlowFields, rebuilt as measures

Business Central calculates certain fields on the fly rather than storing them: the FlowFields. A customer balance, a running total of entries, an outstanding amount. They do not exist in the tables, they are recalculated each time they are displayed.

As a result, a standard extract does not bring them back. They have to be rebuilt as measures in the semantic model, starting from the entry tables and reproducing exactly the original filtering logic.

It is precise work, and it is the leading cause of reports whose figures never match the ERP, the symptom that kills the adoption of a reporting project.

The technical choice

Getting the data out: APIs or bc2adls

On Business Central SaaS there are two ways to feed Fabric, and the choice matters as soon as volumes climb.

Standard connectors compared with bc2adls
	Standard connectors	bc2adls
Principle	Querying the Business Central APIs at every refresh.	An open source extension that exports the tables to Azure Data Lake in Delta format, directly readable by Fabric.
Volumes	Comfortable on a narrow scope; the API quotas become constraining as volume and the number of instances grow.	Incremental export: only the changes travel, which moves the volume constraint elsewhere.
History	Limited to what the API agrees to return.	History is landed once in the lake, then kept independently of the ERP.
To maintain	Nothing in particular: it is standard.	An extension to install in each instance, to follow when Business Central is updated.

What consolidating means

Eight systems, one figure

Your systems arrive as they are in a first layer, time-stamped. A second layer aligns the reference data, charts of accounts, dimensions and customers, and settles the duplicates. A third sets one definition per indicator. Reports, Excel and your applications then all read the same thing.

Eight systems, one revenue figure

Wiring the sources in is the easy part. Agreeing on what each figure means is the project.

We have reconciled up to seven Business Central instances and seven CRMs onto one model. The difficulty grows with the number of reference sets to align, more than with the volume of data.

Our method

How we run a multi-BC consolidation

An inventory of the gaps, entity by entity
Charts of accounts, dimensions, reference data. We produce the list of divergences before writing a single transformation. This is the step failed projects skip over.
Mapping tables validated by the business
These are not technical decisions: saying that account 706 in one subsidiary matches 704 in another commits the accounting. Your teams validate, we implement.
Ingestion and multi-company composite keys
Every record carries the entity it came from. Without that, two invoices numbered the same in two companies merge into one, and the bug is silent.
FlowFields rebuilt as measures
The original filtering logic is taken back up, measure by measure, with a check against the ERP.
Reconciliation figure by figure
We compare the aggregates of the model with those of each Business Central, period by period, down to the euro. A report that does not reconcile will never be used, whatever its graphic quality.

Frequently asked

What people ask about Business Central

Can several Business Centrals be consolidated into one reporting layer?

Yes, and it is a case we handle regularly. We have delivered platforms consolidating up to seven Business Central SaaS instances on a complete scope. The difficulty is not the connection, the APIs are documented and stable, but the alignment: charts of accounts, dimensions, item and customer reference data almost always differ from one entity to the next.

Do the setups have to be harmonised before consolidating?

No, and that is good news: waiting for the setups to be harmonised often means never starting. The alignment happens in the transformation layer, using mapping tables your business teams validate. Each entity keeps its own setup, and the consolidation sits above it.

What happens to Business Central FlowFields?

FlowFields are calculated on the fly by Business Central and are not stored: they do not come through in a standard extract. They have to be rebuilt as measures in the semantic model, starting from the entry tables. It is precise work, and forgetting it produces reports whose figures never match the ERP.

Should bc2adls be used to get data out of Business Central?

It is the approach we favour on large scopes. bc2adls is an open source extension, carried by the Microsoft community, which exports Business Central tables to Azure Data Lake storage in Delta format, so directly readable by Fabric. Its benefit on a multi-instance consolidation is decisive: the export is incremental, it does not go back through the APIs at every refresh, and so it does not hit the quotas that strangle standard extracts as soon as volumes climb. For a narrow scope, the standard connectors are enough and simpler to maintain.

Business Central SaaS or on-premises, does it change anything?

Yes, in how the data is reached. On SaaS you go through the APIs and the endpoints provided, with quotas to respect. On-premises, direct access to the database remains possible and is often faster, but the historical load and the availability are yours to manage. The Fabric target is the same in both cases.

What if we also run Dynamics NAV?

That is common: one entity migrated to Business Central, another stayed on NAV. The two consolidate onto the same platform. The table structures share a common heritage, which helps, but the version differences call for mapping work that should not be underestimated. We have delivered projects combining Business Central, NAV, SAP and Sage on one model.

How many instances, and how far apart are they?

Eleven expert consultantsSaint-Priest, near Lyon, France

Two hours with a consultant to inventory your entities, measure the gap between their setups and tell you what the consolidation would really cost. Free, no commitment.

Request a scoping workshop

Scoping workshop · 2 hours · free