What exactly is a Power BI integration?

By a Power BI integration we mean a fixed connection between a source, such as your accounting, CRM or point-of-sale system, and Power BI. Your numbers then refresh without anyone making an export every week.

Technically, such an integration has two parts:

  • Fetching. Power BI, or a tool in front of it, fetches the data from the source and turns it into tables. In Power BI this is done with Power Query.
  • Refreshing. In the online Power BI service you set how often that data is fetched again. In practice, this is the part where an integration most often breaks.

So an integration is more than the first report. The point is that the numbers are still right next month.

What are the ways to connect your software to Power BI?

Broadly there are four routes. They are not mutually exclusive: in many companies one source runs directly and another through a layer in between.

  1. A standard connector. Power BI has built-in connectors for many sources, such as SQL databases, SharePoint, Excel and some of the well-known SaaS packages. Some vendors offer their own connector or a ready-made template. This is the quickest start, but you get the tables the way the connector provides them.
  2. Directly through the API. Almost every modern package has an API. You can call it yourself with Power Query. That gives you full control over which fields you fetch, but you have to handle authentication, paging through hundreds of pages of results and what happens when the API briefly does not respond.
  3. Through a file. An export that lands in SharePoint or OneDrive automatically, for example through a scheduled export or a Make scenario. Power BI reads that file. Simple and cheap, but only as reliable as the process that puts the file there.
  4. Through a database of your own. A data pipeline fetches the data from all your packages and puts it in one database. Power BI only reads from that database. It is more work upfront, but your calculation rules live in one place and you can keep history.

Which tools you can use for that last route is covered in ETL tools for SMEs.

How often does a Power BI integration refresh?

That depends on how Power BI uses the data. In most SME reports the data is imported: Power BI makes a copy, which you have to refresh to see changes. With DirectQuery, Power BI queries the source again with every click. You then do not need to refresh, but the report is only as fast as the source, and not every source supports it.

For imported data these limits apply (source: Microsoft documentation, checked on 10 November 2026):

  • Power BI Pro: up to 8 scheduled refreshes per day.
  • Premium Per User and Fabric capacity: up to 48 per day.
  • Duration: a refresh on shared capacity must finish within two hours. On Premium the limit is five hours.
  • Failed refreshes: after four consecutive failures, or when credentials have expired, Power BI switches the schedule off. By default the owner of the semantic model receives an email about it.

For most management information, one refresh per night plus a few during the day is plenty. If a number really has to be right to the minute, Power BI is probably not the place to answer that question. Which licence you need is covered in Power BI licensing in 2026.

Do you need a gateway for your Power BI integration?

A gateway is a piece of software that gives the online Power BI service access to data that cannot be reached from the internet. According to Microsoft you need one if your source:

  • sits on your own server or PC, such as a local database or an Excel file on a network drive;
  • sits in a private network, even if that network runs in the cloud;
  • needs software that has to be installed on a machine of your own.

If you only work with cloud packages that can be reached over the internet, you usually do not need a gateway. If something does sit locally, use the gateway's standard mode on a machine that is always on. According to Microsoft, personal mode is meant for a single user and not for reports a team relies on. A gateway on a laptop that gets closed in the evening is a common reason why Monday morning's refresh fails.

How do you choose the right route?

Go through these five questions. They are based on what we see in practice, not on a fixed rule.

  1. How many sources do you want to combine? One source can easily be direct. From three sources on, it becomes hard to keep track inside Power BI of which customer has which number in which system. That is when a database becomes attractive.
  2. Is there a good connector or template for your package? If so, start with it and see whether the tables answer your question. If not, it will be the API or a layer in between.
  3. Do you need history? Many packages only show the current state. If you want to know what your pipeline or stock looked like a month ago, you have to store those snapshots yourself. A direct integration cannot do that.
  4. How large is the data? Thousands of rows are no problem. If you pull hundreds of thousands of rows through an API every night, you run into the API's limits or the maximum duration of a refresh.
  5. Who will manage it? Someone with some Power BI experience can maintain a direct integration in Power Query. A pipeline needs someone who knows where to look when something stops.

For accounting packages we have worked out the choice separately in Connecting Power BI to your accounting software.

An example: three sources at a wholesaler

A fictional example. A wholesaler with twenty-five employees uses Moneybird for accounting, HubSpot for sales and an Excel file for purchase planning. The managing director wants to see per customer what has been sold, what is still outstanding and what is in the pipeline.

  • Step 1: one question. "Which customers are growing and which are declining, and does that match what sales expects?" That gets one dashboard, not five.
  • Step 2: review the sources. Moneybird and HubSpot both have an API. The Excel file is on SharePoint. Nothing is stored locally, so no gateway is needed.
  • Step 3: the route. Because there are three sources and customers have different names in Moneybird and HubSpot, they choose a small database. A nightly pipeline fetches invoices and deals and matches customers on their Chamber of Commerce number.
  • Step 4: Power BI. The report only reads from the database and refreshes at seven in the morning, after the pipeline.
  • Step 5: check. Revenue per customer is compared with the figures in Moneybird. Twelve customers turn out to be duplicated in HubSpot. Those are cleaned up first.

Had they only used Moneybird, a direct integration would have been the logical choice. The third system made the difference.

Where does a Power BI integration usually go wrong?

  • Signing in as one person. If that colleague leaves or the password changes, the refresh stops. Use a dedicated service account or an API key that belongs to the company.
  • Nobody reads the error. The email about a failed refresh goes to the owner of the model. If that is the consultant who built it, nobody inside the company hears about it. Add a fixed address.
  • Fetching everything. Pulling five years of invoices through an API every night is slow and fragile. Fetch only what you need, or only what has changed since the last run.
  • Calculation rules in three places. If revenue is calculated differently in Power Query, in an Excel list and in the package itself, the meeting turns into a debate about which number is right. Define it in one place.
  • A gateway on a laptop. See above: it works fine until someone goes on holiday.
  • An export as a "temporary" solution. A manual export put in SharePoint "for now" is still somebody's Monday task a year later. Automate putting it there, for example with a Make.com integration, or choose another route.

Help with your Power BI integration

We connect business software to Power BI for SMEs, directly where possible and through a data layer where needed. We start with a free introductory conversation about your question and your sources. You then receive a proposal with a fixed price per phase, so you can start small.

See the Power BI integrations we have worked out per package, such as Moneybird and HubSpot. Read more about data engineering if you want to bring several sources together, or about dashboarding if the data is already there.