Power BI Licensing Explained - Subscriptions, Trials and What Australian Businesses Actually Pay For
Almost every Power BI project I have been part of hits the licensing question at an awkward moment. Usually it is right after someone has built a report they are proud of and wants to share it with the wider team, and the answer comes back that half the intended audience cannot open it without a paid license nobody budgeted for. The report is fine. The licensing was never thought through. And now there is an uncomfortable conversation with finance about a monthly bill that was supposed to be zero.
Power BI licensing is not actually that complicated once you understand the shape of it, but Microsoft does not explain it well up front, and the names have shifted over the years as Fabric absorbed Power BI. So let me lay it out the way I wish someone had laid it out for me the first time, based on what Australian businesses actually end up paying and where they get caught. Microsoft's subscriptions, licenses and trials guidance is the reference if you want the official version, but the official version reads like a pricing sheet, not advice.
The free tier is a trap if you misunderstand it
Anyone in your organisation can get a free Power BI license. With it, they can connect to data, build reports in the service, and use Power BI Desktop, which is free to download regardless. So people naturally assume Power BI is free, build something, and then discover the catch.
The catch is sharing. A free license lets you build and view your own content, but you cannot share it with colleagues in the normal collaborative way, and they cannot share theirs with you, unless paid licensing is involved. So the free tier is genuinely useful for a single analyst poking at data on their own machine, and close to useless the moment reporting becomes a team sport, which it almost always does. I have watched a team build out a whole reporting layer on free licenses, feel great about it, and then hit a wall on the day they tried to actually roll it out. Understand this early. The free tier is for exploration, not distribution.
Pro is the workhorse, and the per-user cost adds up quietly
Power BI Pro is the per-user license most organisations start with. It lets a person publish reports, share them, and consume shared content, within workspaces. For a long time this was simply "the Power BI license" and for many small and mid-sized Australian businesses it still is the right answer.
The thing to watch with Pro is that it is per user, and everyone who needs to either publish or view shared content needs one. That second part is what surprises people. It is not just your report builders who need Pro, it is your report viewers too, unless you move up to capacity, which I will get to. So a modest team of five analysts publishing to two hundred viewers is not five Pro licenses. It is potentially two hundred and five. At a per-user monthly rate that is a real number, and it is the calculation that quietly reshapes a lot of licensing decisions.
My honest advice: before you commit to Pro across a large viewing audience, actually count the viewers and multiply it out for a year. If that number makes you wince, you are the exact profile that capacity licensing exists for, and you should read the next section carefully rather than defaulting to Pro because it is the familiar option.
Premium Per User and capacity are where it gets interesting
Above Pro there are two directions you can go, and picking the wrong one wastes money in different ways.
Premium Per User, usually written PPU, is still a per-user license but unlocks the heavier features: larger datasets, more frequent refreshes, paginated reports, advanced dataflows and the like. It costs more per head than Pro. It suits a smaller group of power users who genuinely need those capabilities, where paying a premium for a handful of people is cheaper than jumping to capacity.
Capacity, which now lives under the Fabric umbrella, is the other model, and it changes the maths completely. Instead of licensing every viewer, you buy a pool of compute capacity, and content published to that capacity can be viewed by anyone in the organisation with just a free license. Read that again, because it is the whole point. With capacity, your two hundred viewers do not each need a paid seat. You are paying for the engine instead of the seats.
The trade-off is that capacity has a meaningful base cost and you need enough viewers to justify it. There is a crossover point where capacity becomes cheaper than buying Pro or PPU for everyone, and finding that point is genuinely one of the more valuable things we do early in an engagement. Get it right and you can save an organisation a large recurring sum. Get it wrong in either direction and you are either overpaying for capacity you do not use, or drowning in per-user licenses you did not need to buy. This is exactly the kind of question our Power BI consultants get pulled in to answer, because the honest answer depends on your specific viewer count, and there is no generic right choice.
One more thing on capacity now that Fabric is the container for all of this: the capacity you buy is not just Power BI anymore, it also powers data engineering, warehousing and the rest of the Fabric workloads. If your organisation is heading towards a broader data platform rather than just reports, that changes the capacity conversation again, and it is worth bringing our Microsoft Fabric consultants into the planning rather than sizing Power BI in isolation and redoing it a year later.
Trials are useful, but mind the cliff
Power BI offers trials, and they are a genuinely good way to test the paid features before committing budget. A free user can typically start a Pro or Fabric trial and get the fuller experience for a limited window. Use this. It is the right way to prove out whether capacity features actually solve your problem before you sign a purchase order.
The thing to plan for is what happens when the trial ends. When a trial expires, the elevated capabilities go away, and content that depended on them can stop working the way people expect. I have seen a team build a whole workflow during a trial, get everyone excited, and then have it quietly break when the trial lapsed and nobody had arranged the paid license to replace it. Treat a trial as a time-boxed experiment with a decision at the end, not as a free extension of the product. Diarise the expiry, decide before you hit it, and have the purchase lined up if the answer is yes.
Tenant-level settings sit underneath all of this
Licensing does not happen in a vacuum. Who can start a trial, who can be assigned which license, and whether self-service purchasing is allowed at all, are governed by tenant and admin settings. In a lot of organisations, self-service purchase is left on by default, which means individuals can buy their own Pro licenses on a company card without central oversight. Sometimes that is fine. Often it leads to a scattered, unmanaged license estate that finance cannot account for.
My general steer is to make a deliberate decision about self-service purchasing rather than leaving it at whatever the default is. If you want central control over spend and governance, turn it off and route requests through whoever owns the tenant. If you want to move fast and trust your teams, leave it on but monitor it. Either can be right. Drifting into it by accident is what you want to avoid.
How I would actually approach it
If you are starting out and small, begin with Pro, count your viewers honestly, and keep an eye on the total. If your viewing audience is large or growing fast, model capacity against per-user cost properly before you commit, because the crossover arrives sooner than people expect. Use trials to prove features, but treat the expiry as a hard decision point. And make the tenant-level choices about trials and self-service purchase on purpose rather than by default.
The deeper point is that licensing should follow your reporting strategy, not the other way around. When licensing gets decided reactively, in a panic, the day someone wants to share a report, you end up with a patchwork that costs more and governs worse than a plan made up front. Working out the right shape is part of a broader data and AI direction, which is what our business AI strategy work covers.
None of this is hard once someone lays it out plainly, which Microsoft mostly does not. If you are staring at a Power BI bill that seems too high, or a rollout that stalled on licensing, that is a very fixable problem and a common one. Have a look at what our Power BI consultants do, or just get in touch and we will help you work out what you should actually be paying.