Power Apps licensing, the connectors are the bill
Power Apps licensing looks simple, two paid plans, seeded rights, a pay as you go meter, until premium connectors, Dataverse capacity, and the seeded boundaries collide. The per user license is rarely the real cost driver: the connectors and the storage are, and governance is what keeps both honest.
Prepared by Redress Compliance · August 6, 2026 · Microsoft advisory. Based on 30 to 40 Power Platform engagements reviewed 2024 to 2026.
Executive summary
The plan structure is the easy part. The per app plan, near $5 per user per app per month, covers one app; the premium plan, near $20 per user, covers unlimited apps; and a pay as you go Azure meter bills per app per active user monthly with no upfront commitment. Microsoft 365 and Dynamics 365 seed limited Power Apps rights inside their own applications only, and the seeded boundary is where the licensing actually gets decided.
Premium connectors are the boundary trigger. One premium connector in an app pushes every user of that app onto a paid plan, and across our engagements premium connector usage pushed 20 to 35 percent of makers onto paid plans they did not need, where a redesign, the same data through a seeded path, would have kept them on rights already paid for.
Dataverse is the hidden line. Storage and premium connectors drive most surprise overages, and Dataverse storage overages added 10 to 20 percent to the annual bill on estates that never set a capacity alert. The database, file, and log capacities pool at the tenant level and grow with every environment nobody cleans.
Sprawl hides the demand signal. Environment sprawl of 50 plus environments hid duplicate apps and made true demand impossible to measure until consolidated, which is the governance finding under everything: unmanaged maker sprawl is the top reason Power Apps spend runs ahead of value, and the levers, plan right sizing and environment consolidation, cut spend without blocking a single maker.
The licensing models, and where each fits
| Model | The construction | The honest fit |
|---|---|---|
| Seeded rights | Limited Power Apps use inside Microsoft 365 and Dynamics 365 apps, standard connectors only | The default home for most internal apps, until a premium connector evicts them |
| Per app plan | About $5 per user per app per month, one app | Focused populations on one or two premium apps |
| Premium plan | About $20 per user per month, unlimited apps | Genuine multi app power users, a smaller population than provisioned |
| Pay as you go | Azure metered per app per active user monthly | Spiky and seasonal usage where commitment wastes |
The connector boundary, and the redesign dividend
The 20 to 35 percent of makers on unneeded paid plans got there the same way: an app needed data, the maker grabbed the premium connector that had it, and the licensing consequence, every user of the app onto paid plans, was invisible at build time. The redesign dividend is real because the seeded paths often reach the same data: standard connectors, the Microsoft 365 surfaces, or a flow restructured to keep the premium call server side. Not every app redesigns, and the ones that genuinely need premium connectivity should pay for it, but the review that asks the question app by app recovered the difference across our engagements.
The governance layer is what makes the review repeatable rather than heroic: connector policies by environment, a default environment locked down, and the maker path routed through environments whose licensing posture is deliberate. The license optimizer handles the seat side of the same estate, and the EA guide the agreement the plans ride inside.
The Microsoft EA renewal playbook
The agreement architecture the Power Platform lines ride inside: the seven levers, the true up mechanics, and the persona based mix the plans should follow.
Get the white paper →Dataverse capacity, the overage nobody alerts on
Dataverse capacity pools at the tenant, database, file, and log, growing with every environment, every table, and every attachment nobody lifecycle manages. The 10 to 20 percent annual overage on unalerted estates is the predictable result of a meter nobody watches: capacity alerts unset, growth unattributed, and the true up arriving as a surprise the platform recorded all along. The controls are unglamorous: the alert set, the growth attributed by environment, attachments routed to SharePoint where they belong, and the archive posture decided before the overage prices it.
- Percentile standing for your exact deal size and industry, from real closed transactions
- Scenario simulation before the call: test alternative terms and see the financial impact of each
- A negotiation playbook, talking points, and a two page executive brief on day one
What we saw across Power Platform engagements, 2024 to 2026
Across roughly 30 to 40 Microsoft Power Platform engagements Fredrik Filipsson reviewed between 2024 and 2026, the per user license was rarely the real cost driver; Dataverse and connectors were:
Pushed onto paid plans by premium connectors where a redesign kept the seeded rights doing the work.
Annual bill growth from storage nobody alerted on, attributed, or lifecycle managed.
The sprawl finding framed the rest: at 50 plus environments, duplicate apps multiplied, demand became unmeasurable, and every licensing decision was made blind until consolidation restored the signal. The recovery sequence was always the same, consolidate, then review connectors, then right size plans, because each step makes the next one measurable, and none of it blocked a maker from building anything.
Your first five moves
- Consolidate the environments first; at 50 plus, demand is unmeasurable and every other decision is blind.
- Review connectors app by app, and take the redesign dividend where seeded paths reach the same data.
- Set the Dataverse capacity alerts today, attribute growth by environment, and route attachments where they belong.
- Right size the plans to measured use: per app for focused populations, premium for genuine multi app users, pay as you go for the spikes.
- Put connector policies in the governance layer so the review repeats instead of reoccurring. The Microsoft practice runs the estate with you.
Frequently asked questions
How is Power Apps licensed?
Through four routes: seeded rights inside Microsoft 365 and Dynamics 365 covering their own apps with standard connectors, the per app plan near $5 per user per app per month, the premium plan near $20 per user for unlimited apps, and a pay as you go Azure meter billing per app per active user monthly.
When do Power Apps users need a paid plan?
When an app crosses the seeded boundary, most commonly by adding a premium connector, which puts every user of that app onto a paid plan. In our engagements, 20 to 35 percent of makers on paid plans got there through connectors a redesign would have avoided, keeping them on rights already paid for.
What are Power Apps premium connectors?
The connector tier outside the standard set, covering most third party and many Azure data sources, whose use in an app triggers paid licensing for all its users. The connector list, app by app, is the estate's real bill driver, which makes the licensing review a connector review in practice.
Why is our Dataverse bill growing?
Storage pools at the tenant and grows with every environment, table, and attachment nobody lifecycle manages, and estates without capacity alerts saw 10 to 20 percent added annually in overages the platform recorded all along. The alert, growth attribution, and attachment routing are the controls.
How many Power Platform environments should we have?
Few enough to measure: at 50 plus, duplicate apps hide and true demand disappears, making every licensing decision blind. Consolidation is the first move in every recovery sequence because it restores the demand signal the plan right sizing and connector review depend on.
How do we cut Power Apps costs without blocking makers?
The sequence from our engagements: consolidate environments to restore measurement, review connectors app by app for the redesign dividend, set the Dataverse alerts, and right size plans against measured use. None of it removes a capability makers use; all of it removes the spend running ahead of value.
