Someone has suggested building an internal app on Power Apps. It sounds sensible until you open the licensing page, at which point the confidence drains away, because the page appears to be written for somebody who already knows the answer.
The underlying model is actually simple. It is the naming and the exceptions that are hard.
The one distinction that matters
Some things are included with the Microsoft 365 licences you already pay for. Some things require an extra licence per person, per month.
What separates them is where the data lives and what it talks to.
Build on things already inside Microsoft 365, SharePoint lists, Excel in OneDrive, Outlook, Teams, and you are generally covered by what you have. Reach outside that, into Dataverse or a third-party system with a premium connector, such as SQL Server or Salesforce, and you are into paid territory for every person who uses it.
That is genuinely the whole rule. Everything else is detail hanging off it.
Why the bill surprises people
The surprise is almost never the price of one licence. It is the multiplication.
A premium licence is per user, per month, forever. An app built for eight people is a modest ongoing cost. The same app rolled out to sixty is a different conversation, and the decision that caused it was made months earlier by someone choosing a connector without knowing it was a pricing decision.
This is the single most common way a small internal tool becomes an argument with finance. It is entirely avoidable, and only at the start.
The questions to ask before anyone builds
- Does this need a premium connector, and why? Sometimes yes, genuinely. Often it is habit, because the person building is used to a particular way of storing data.
- How many people will actually open it? Not how many are in the company. Multiply that number by the monthly cost by twelve, and look at the annual figure before agreeing to anything.
- Could this run on a SharePoint list instead? For a great many internal tools, the honest answer is yes, and that option is more capable than its reputation suggests.
- What happens when the person who built it leaves? Not a licensing question, but the one that decides whether this is an asset or a liability.
When paying is the right answer
None of this is an argument for always avoiding the premium tier. There are situations where it is clearly correct.
Real relational data, where records genuinely connect to other records and the relationships matter. Proper permissions, where different people must see different rows rather than different screens. Volume, where a list would strain. Anything where an audit trail is not optional.
In those cases the licence is buying capability you actually need, and building around it to save money produces something fragile that costs more in maintenance than it saved in fees.
The failure is not paying. It is paying without deciding.
A caution on the trial
Trials are generous and they end. An app built and demonstrated during a trial period will simply stop working for people who have no licence when it lapses, and it will do so on an ordinary morning with no warning to the people relying on it.
If a pilot is running on trial licences, know the date, and know what happens on it before you arrive there.
The honest summary
Ask one question early: does this need to reach outside Microsoft 365? If no, you are very likely covered by what you already pay for. If yes, work out the annual cost for the real number of users before agreeing to build.
Licensing terms change, so confirm the current position rather than relying on this or on what was true two years ago.
If a proposal is in front of you and you cannot tell which side of the line it sits on, describe what it is meant to do. That is usually answerable in a paragraph, and it is worth knowing before the build rather than after. How this work is approached here is set out separately.
Common questions
What decides whether we need a paid Power Apps licence?
Where the data lives and what it talks to. Build on SharePoint, Excel in OneDrive, Outlook or Teams and you are generally covered by Microsoft 365. Reach into Dataverse or a premium connector and it is per person, per month.
Why do the costs surprise people?
The multiplication, not the price. An app for eight people is modest. The same app rolled out to sixty is a different conversation, and the decision that caused it was a connector chosen months earlier.
Is it ever right to pay for premium?
Yes. Real relational data, row-level permissions, volume, or a required audit trail all justify it. Building around the licence to avoid the fee produces something fragile that costs more to maintain than it saved.