If you went looking for a Power Apps licence in the last few months and found the option you'd been told to buy had quietly disappeared, you're not imagining it. As of 2 January 2026, Microsoft closed the long-standing Power Apps per-app plan to new customers. That SKU was the cheap, sensible default for years: licence one app for one user, pay a small per-app fee, done. With it gone for new buyers, most Australian organisations now face a straight two-way choice between the Power Apps Premium (per-user) plan and the pay-as-you-go meter. The two price very differently, and the wrong call can mean paying two or three times what you need to.
This piece lays out the AUD numbers, the break-even point that actually decides it, and the Dataverse implication that catches people out after they've signed up. We work with this licensing model every week for Brisbane and east-coast clients, so the guidance below is the call we'd make in a real engagement, not a feature comparison lifted off a vendor deck.
What actually changed in January 2026
The per-app plan let you licence access to a single application per user at a low monthly rate. It suited the very common pattern where a frontline worker only ever touches one custom app. From 2 January 2026 that plan stopped being sold to new customers. There's important nuance: if you're on an Enterprise Agreement or buying through a Cloud Solution Provider (CSP) partner, existing per-app entitlements can generally still be renewed or purchased, and Microsoft restored CSP availability after the initial announcement. But if you're a new buyer without one of those channels, per-app is effectively off the table.
That leaves two products doing the heavy lifting, and the decision between them is almost entirely about how many of your licensed people actually use the apps in a given month.
The two options and what they cost in AUD
Power Apps Premium (the renamed per-user plan) is a flat seat licence. You pay per assigned user per month whether they open an app once or fifty times, and that user gets unlimited custom apps plus access to Dataverse and premium connectors. List price is roughly USD $20/user/month, which lands at approximately AUD $33.60 per user per month (indicative AUD list — confirm at purchase; AUD ex GST).
Pay-as-you-go is consumption billing through an Azure subscription. You're charged per active user, where an active user is anyone who opens an app at least once in the calendar month. Open it zero times that month, pay nothing for that person. The meter sits at roughly USD $10 per active user per app per month, around AUD $16.80 (indicative AUD list — confirm at purchase; AUD ex GST). There's no upfront commitment and no seat assignment — usage just appears on your Azure bill.
The break-even that decides it
Strip away the noise and it's one ratio: what share of your licensed population actually opens the app in a typical month. Premium costs about AUD $33.60 per user regardless. Pay-as-you-go costs about AUD $16.80 only for the people who show up. Divide one by the other — 33.60 ÷ 16.80 — and you get a break-even at a 50% monthly active rate.
- Below ~50% monthly active users: pay-as-you-go is cheaper. The dormant half of your population costs you nothing.
- Above ~50% monthly active users: the per-user Premium plan is cheaper and far more predictable.
- Right around 50%: pick on predictability, not pennies — see below.
A worked example. Say you roll an app out to 100 staff but only 30 open it in a given month. Premium would cost 100 × $33.60 = $3,360 if you licensed everyone, or you'd be fiddling with assigning seats to a rotating cast. Pay-as-you-go costs 30 × $16.80 = $504 that month, and it self-adjusts as usage moves. That's the classic case for the meter: broad, shallow, unpredictable usage — field staff, seasonal workers, an app that only fires during an annual process.
Flip it. An app 90 of those 100 staff use daily costs 90 × $16.80 = $1,512 on the meter versus $3,360 on Premium seats — the meter still looks cheaper until you remember Premium covers unlimited apps per user and the meter charges per app. Give those 90 people a second and third app and the meter overtakes the flat seat fast.
The Dataverse implication people miss
Both plans include Dataverse entitlements, but the way you pay for what you consume differs, and this is where a tidy licensing decision turns into an unexpected line item. Premium seats come with a per-user Dataverse database and file capacity allowance that pools across your tenant. Pay-as-you-go bills Dataverse database, file, and log capacity as separate Azure meters on top of the per-active-user app charge.
So if your app is data-heavy — lots of stored records, attachments, audit logging — a pay-as-you-go environment can quietly accrue Dataverse storage charges that the headline 'AUD $16.80 per active user' figure never hinted at. We've seen the app meter come in low and the storage meter be the thing that actually moves the bill. Before you commit to pay-as-you-go for anything data-heavy, model the Dataverse consumption separately, not just the active-user count.
Practical setup notes
Pay-as-you-go requires linking the Power Platform environment to an Azure subscription with a billing policy — that's an admin and finance job, not something a maker self-serves, and the charges land on your Azure invoice rather than your Microsoft 365 bill. Premium is assigned like any other seat licence in the Microsoft 365 admin centre. If your finance team needs charges to sit in one predictable place each month, that operational difference matters as much as the per-unit price.
One more channel point: if you have an existing Enterprise Agreement or a CSP partner relationship, ask explicitly whether per-app is still available to you before assuming it's gone. For a single-app, single-user scenario it can still be the cheapest path where you're entitled to it — the 2 January change closed it to new buyers, not to everyone.
How to choose without overthinking it
- 1Estimate the monthly active ratio: of everyone who could use the app, how many realistically will in a typical month?
- 2Under ~50% active and one app per person → pay-as-you-go.
- 3Over ~50% active, or people using multiple apps → Premium per-user seats.
- 4Data-heavy app → model Dataverse storage meters separately before trusting the per-user figure.
- 5New or uncertain workload → launch on pay-as-you-go, watch the real Azure metering for a quarter, then graduate steady users to seats.
The maths isn't complicated; the discipline is in measuring real usage rather than assuming it. Most of the overspend we see comes from buying flat per-user seats for a population that only half-uses the app, or from running a data-heavy app on the meter and being surprised by Dataverse charges. Get the active ratio right and the Dataverse footprint modelled, and the rest of the decision falls out of the 50% break-even on its own.