Why would an SME want a central database?

Your business data is probably scattered right now: in your accounting software, your CRM, your planning tool, your web shop and a handful of Excel lists. Each system has its own database, but that database is built to make the system work, not to build reports on or to combine data.

A central database is where you bring that data together. Connections fetch it at fixed times, you record definitions once and you build up history your systems do not keep themselves. Reports in Power BI and automations then read from that single source.

Whether you need such a central place yet is covered in data warehouse for SMEs: when you need one and when not. This article is about the next question: if you want one, which database do you choose?

Which databases are realistic options for an SME?

There are dozens of databases, but for an SME that works with Power BI, three routes remain in practice.

  • PostgreSQL. An open-source database that has been developed for decades and is used very widely. There are no licence fees for the software itself. You can run it yourself or use it as a managed cloud service, such as Azure Database for PostgreSQL or a comparable service from another cloud provider.
  • SQL Server and Azure SQL Database. Microsoft's database. You can run SQL Server on your own server, or use Azure SQL Database as a managed cloud service. It naturally fits well with the rest of the Microsoft world, including Power BI and Microsoft 365 accounts.
  • Microsoft Fabric. Not a single database but a data platform, which now also includes Power BI. Within Fabric you can use a lakehouse, a warehouse and, since late 2025, also a SQL database built on the same technology as Azure SQL.

All three can store ordinary tables, run SQL queries and easily handle the amount of data an SME has. So the difference is not whether it can, but in management, cost and how it fits with what you already have.

Which questions should drive your choice of database?

Rather than comparing the technology, you are better off answering these five questions.

  1. Who manages it? A database on your own server needs updates, backups and monitoring. A managed cloud service takes most of that off your hands. For a company without its own IT department, a managed service is almost always the wiser choice.
  2. How does Power BI reach it? Power BI has standard connectors for PostgreSQL, SQL Server and Azure SQL, and Fabric is directly connected to it. If the database sits in your own server room, you need an on-premises data gateway for scheduled refresh in the Power BI service. With a cloud database that is usually not needed.
  3. Which Microsoft licences do you already have? If you already use Power BI Pro or a Fabric capacity, that changes the sums. How the licences fit together is explained in Power BI licensing: what does an SME need?
  4. What does the bill look like? With a managed database you pay for compute and storage. With Fabric you pay for a capacity that is billed by the hour while it is running. Do not just look at the monthly price, but also at what happens when usage grows.
  5. Who will work with it next? Choose something your current or future partner knows. A database only its builder understands is a risk.

When do you choose PostgreSQL, SQL Server or Fabric?

With those questions in mind, a fairly clear picture emerges.

PostgreSQL: flexible and without licence fees

PostgreSQL is a good fit if you do not want to be fully tied to Microsoft, if other applications run on the same database too (a web application or a customer portal, for example) or if your partner already works with it. Many automation and data platforms have a standard connection for it. Do choose a managed version, unless you have someone who wants to maintain a server.

SQL Server or Azure SQL: the logical Microsoft choice

If you already work with Microsoft 365 and Power BI, Azure SQL Database is often the most obvious choice. You can sign in with the same Microsoft accounts, the connection to Power BI is simple and many people know the language. At the time of writing (checked on 28 October 2026), Microsoft offers a free version of Azure SQL Database with a monthly allowance of compute time and 32 GB of storage per database. According to Microsoft it is intended for development and proof-of-concept work and has no SLA, but it is fine for getting started. Running your own SQL Server in the server room is also possible, but then the management is back with you.

Microsoft Fabric: if you are already heading towards a data platform

Fabric becomes interesting when you want more than one database: fetching data with Data Factory, storing it in a lakehouse or warehouse and reporting in Power BI, all in one environment. You do pay for that with a fixed capacity, and on capacities below F64 the people viewing the reports still need a Power BI Pro or Premium Per User licence. For an SME with a few sources and a handful of reports, Fabric is often more than you need right now. For a growing company with several sources and its own data ambitions, it can actually bring clarity.

Our rule of thumb: start with a managed Azure SQL or PostgreSQL database, unless you already know you will use the rest of Fabric. A well set-up database can always be moved or grown later.

An example: three sources, one database

A fictional but recognisable example. A wholesaler with twenty employees uses an accounting package, a CRM and a web shop. Everyone has a Microsoft 365 account and three people already have a Power BI Pro licence. There is no in-house IT department.

The trade-off:

  1. Management. Nobody wants to maintain a server, so it will be a managed cloud database.
  2. Licences. Pro is already there, and with three sources and two reports a Fabric capacity is not needed yet.
  3. Choice. An Azure SQL database, set up with separate schemas for raw and transformed data. Sign-in uses the existing Microsoft accounts.
  4. Loading. A data pipeline fetches customers, invoices, deals and web shop orders every night. How such a pipeline works is explained in what is a data pipeline (ETL).
  5. Reporting. Power BI reads only from the database, no longer directly from the systems.

If the wholesaler uses Cirqll as its CRM, Cirqll data in your own database explains how to keep the history of your pipeline there as well.

How do you set up a database for an SME properly?

Whichever database you choose, these agreements make the difference between a cluttered attic and a foundation:

  • Separate raw and transformed. Store fetched data unchanged in its own part of the database, and put the transformed tables that reports read from somewhere else. That way you can always go back to the source.
  • Record definitions in the database. What revenue is, when a customer counts as active and how you treat VAT live in one place, not separately in every report.
  • Give people only the rights they need. Reports get read rights, connections get write rights on their own tables, and nobody works with the admin account.
  • Arrange backups and monitoring. Managed cloud databases usually include backups as standard. Check how long they are kept and agree who gets an alert when a connection stops.
  • Document briefly. One page listing which tables exist, where they come from and when they are refreshed is enough.

Help choosing and setting up your database

Choosing, setting up and loading a central database is what we mean by data engineering. The systems we can already connect are listed under integrations, and for the reports on top, see dashboarding. How such a project runs in phases is described in our approach.