Business Central analytics: the three routes in 2026

Three routes to Business Central analytics: Microsoft apps, a semantic model of your own, or Fabric. Five questions decide which you need.

Three routes, not thirty

Search for Business Central analytics and you will find thirty tools, each with a reason it is the one. Underneath, there are three routes, and the decision between them is made by five questions about your business rather than by any feature list. This post is the version of that conversation we have with every Business Central client, written down.

Route one: Microsoft's Power BI apps

Microsoft ships nine Power BI apps for Business Central, 116 reports across finance, sales, purchasing, inventory, projects, manufacturing and more, maintained by Microsoft and free beyond the Power BI Pro licences your users need. They are installed per company, they follow the standard data model, and they are the right answer for a single-company business with standard questions. Our post on the apps covers what they do and where they stop.

Route two: a semantic model of your own

The second route is a Power BI semantic model built over Business Central, reading from the API pages or a direct extract, with the measures defined the way your business defines them. This is where most mid-market firms end up once they have more than one company, a definition of margin that is not Microsoft's, or a second system whose data has to sit beside the ledger. It is also where tools like Jet and Cosmos live, offering a pre-built model and finance-friendly reporting on top; our reporting options explained post sets out how they compare with a build. The route is a matter of weeks, and its limits are refresh frequency, history, and how many other sources you want in the same model.

Route three: a Fabric data platform

The third route puts Business Central data into a Fabric lakehouse alongside everything else, with a semantic model on top. Today that means an extraction such as bc2adls or the standard connectors into OneLake; Microsoft announced Business Central mirroring at FabCon Europe in October, with public preview expected in a 29.1 update, which will turn that extraction into configuration. This is the route for groups, for businesses with a CRM, a warehouse system or an e-commerce platform that need to be reported with the ledger, for anyone who wants history that outlives an ERP migration, and increasingly for anyone who wants AI agents to reason over governed data. When Business Central to Fabric is worth it and the medallion architecture blueprint go deeper.

The five questions

How many companies? One company and the apps may do. Two or more and you need route two or three, because consolidation is the first thing the apps cannot do, as our consolidation post explains.

Whose definitions? If finance accepts Microsoft's definition of gross margin, ageing buckets and utilisation, route one. If the business has its own, and most do, the definitions have to be modelled once in a layer you control, which is route two at minimum.

What else needs to sit beside the ledger? Nothing: route one or two. A CRM, a stock system, an e-commerce platform, a payroll system, spreadsheets of budgets and forecasts: route three, because joining those to Business Central in a Power BI model alone becomes fragile fast.

How much history, and what happens at the next migration? If you need five years of trend and you expect to change ERP or reimplement within that time, the data has to live somewhere that is not the ERP. That is route three.

How fresh, and how many people? A nightly refresh for a finance team of ten suits route two. Hourly for two hundred people across sites, or a shift wallboard, points to route three on a capacity.

What usually happens

Most businesses start on route one, move to route two within a year because of the definitions question, and move to route three when a second company or a second system arrives. There is nothing wrong with that sequence as long as the move to route two is built so that route three can reuse it: the same measure definitions, the same dimension handling, the same report designs. The expensive version is rebuilding reporting three times. The cheap version is building the semantic layer once and changing what sits underneath it, which is how we design the route two model from the start. Five numbers every Business Central user should see every morning is a good example of a report that should survive all three routes unchanged.

Where to start

Answer the five questions honestly, with finance in the room. If every answer points to route one, install the apps and run the dimension job. If any answer points past it, the right next step is a short Establish phase that settles the definitions and the pattern before anything is built, which is how our Business Central analytics engagements start.

Business Central analytics in 2026: the three routes to reporting and how to choose between them