Which ETL tools are there for SMEs?
ETL stands for extract, transform and load. What that involves is covered in what is a data pipeline (ETL). This article is about the next question: what do you do it with?
For an SME that works with Power BI, three routes remain in practice:
- Power Query. The transformation step already built into Power BI Desktop and Excel. You fetch data, transform it in a series of steps and load it into your report or worksheet.
- Make.com. An automation platform that fetches data at a fixed time and processes it record by record. It is built for processes, but you can also use it to put data into a database.
- Your own pipeline. A script or a data platform such as Data Factory that fetches data in bulk and writes it to a central database. Power BI then reads from that database.
When is Power Query enough as an ETL tool?
Power Query is often a fine start. You need nothing extra, the steps are easy to follow and for many systems there is a connector or an API you can call. For one or two reports on a few sources, it is usually the right choice.
You do need to know where the limits are. According to Microsoft's documentation (checked on 5 November 2026), these limits apply with a Pro licence:
- At most eight scheduled refreshes per day per semantic model. With Premium Per User or a Fabric capacity that is 48.
- A refresh must finish within two hours. With incremental refresh, which also works on Pro, you only fetch the latest period again. The source does need to be able to filter by date.
Those limits are rarely the real problem. The harder limits are in the setup:
- Power Query in Power BI Desktop or Excel does not write to a database. The result sits in your report or worksheet and is not easy to use outside it.
- You do not build up history by default. If your system only shows today's position, after a refresh you only see today's position too.
- The logic sits in the report. Build a second report and you copy the steps, leaving two versions to maintain.
Dataflows exist for that last point: Power Query online, so several reports can use the same transformed tables. With a Pro licence you can create the older type, which Microsoft now labels as legacy, with fewer features. Dataflow Gen2 can write to a database, but it requires a Fabric or Premium capacity.
When is Make.com suitable as an ETL tool?
Make works well when something has to happen per record, or when the numbers are small. Think of recording the status of a few hundred open deals every night, fetching a list from a system without a connector, or passing on an order while giving someone a task. It is quick to build, scheduling and error handling are included, and someone else can follow the scenario on screen.
The tipping point lies in how Make charges. With ordinary modules every action costs a credit, and a module runs for every record that comes in. How to calculate that is covered in Make.com automation: what does it cost. With reporting data this adds up quickly, because you are not processing ten events a day but thousands of rows a night.
Here too, there are limits to know upfront (checked on 5 November 2026):
- Run time. According to Make's help centre, a scenario stops after 45 minutes, or after 10 minutes on the free plan. The remaining records are then no longer processed.
- Data transfer. Every 10,000 credits in your plan include 5 GB of data transfer per month.
- Storage. Make's data stores are small: 10 MB per 10,000 credits. So they are no substitute for a database.
If you use Make as an ETL tool, write to a real database, fetch only what has changed and make sure you get an alert when a run fails. How to do that last part is covered in Make.com error handling.
When do you need your own pipeline?
Your own pipeline fetches data in bulk, puts it in your own database and records the transformations there. Power BI, Excel and automations then read from the same tables. That takes more work to set up, but it solves exactly the points where Power Query and Make get stuck: you build up history, your definitions are in one place and the number of records no longer directly sets your bill.
Here too you have a choice:
- A script on a scheduled task. For example Python and SQL. This gives the most freedom, especially with systems that have an idiosyncratic API. You are responsible for making sure someone can maintain it, and you have to arrange documentation and monitoring yourself.
- Azure Data Factory. A Microsoft service where you assemble pipelines from ready-made activities. You pay for what you use, among other things per 1,000 runs and for the time spent copying data. Check the current rates in Microsoft's pricing calculator.
- Data Factory in Microsoft Fabric. Similar, but the usage draws on a Fabric capacity. Interesting if you already have one or want to use the rest of the platform.
In every case you need a place where the data lands. Which database fits is covered in a database for SMEs. Whether you should call that a data warehouse straight away is covered in data warehouse for SMEs: when you need one and when not.
The licence or hosting is usually not the biggest cost here. That is the time for building and maintenance: APIs change, tokens expire and someone has to know what happens when the pipeline does not run for a night.
How do you choose between Power Query, Make and your own pipeline?
Go through these five questions. A clear yes on one of the first four is usually reason enough to look one step further.
- Do you need to look back in time? Do you want to know what your stock, pipeline or receivables looked like a quarter ago, and does your system not keep that? Then you have to record snapshots. Power Query alone cannot do that. For small numbers Make can, above that your own pipeline.
- Do several reports or systems use the same data? With two reports a dataflow can still get you there. If you also use the same figures in automations or a customer portal, the data belongs in a database.
- How many records change per day? Roughly count on one Make credit per row written. At tens of thousands of rows a month you will notice it in your plan. At hundreds of thousands, a bulk route almost always makes more sense.
- Does something have to happen, or does something only need to become visible? If a task, email or update has to come out of it, Make fits. If the goal is only a report, a workflow tool is often not needed.
- Who will maintain it? A simple solution your colleague understands beats a tidy pipeline nobody dares to touch. If you have nobody in-house, arrange maintenance before you start building.
Our rule of thumb: choose the lightest route that answers your current question and do not invent volumes you do not have yet. But do start recording history in time if you are going to need it, because you cannot fetch it afterwards.
An example: three phases in the same web shop
A fictional but recognisable example. A web shop with twelve employees uses a web shop platform and an accounting package, and management wants to see revenue and margin per product group.
- Phase 1: Power Query. One Power BI report fetches orders and invoices and refreshes four times a day. This works fine, takes little build time and management has its overview.
- Phase 2: adding Make. The buyer wants to look back at stock levels per week, but the platform only shows today's position. A scenario records about 400 changed stock rows in a small database every night. That is roughly 400 credits a night, so around 12,000 a month, plus a few credits for fetching. That still fits, although it is already more than the smallest credit tier of 10,000.
- Phase 3: your own pipeline. A second sales channel is added and the number of rows grows to 5,000 a night. Through Make that is about 150,000 credits a month. For this data the web shop moves to a script that fetches it in bulk and writes it away. Power BI now reads from the database, and the revenue definition lives there too.
Make does not disappear. It keeps sending the alert when a product drops below minimum stock, for example. Each tool then does what it is good at.
Help choosing your ETL tool
Not sure which route suits your situation, or is your current setup reaching its limits? Setting up pipelines and databases is what we mean by data engineering. The systems we can already connect are listed under integrations. For the processes around it, see workflow automation, and for the reports on top, dashboarding.
