Microsoft Power Platform spend is growing faster than any other line in most enterprise Microsoft estates. Power Apps, Power Automate, Power BI, Power Pages, Copilot Studio, and the Dataverse storage that underpins them now exceed Microsoft 365 commercial cloud add ons in many estates.
The growth is the right kind of growth: business outcomes from citizen development, automation, and analytics. The growth is also the wrong kind of growth: uncontrolled licensing sprawl, capacity allocation that does not match consumption, and audit exposure that few organizations have measured.
This CIO playbook is the buyer side framework for licensing Microsoft Power Platform with discipline, in 2026 pricing, with clear governance, audit defense, and a renewal stance that holds value as the platform expands. Pair this with our Microsoft services overview, the Microsoft Knowledge Hub, the Microsoft EA renewal playbook, and the 2025 to 2026 Microsoft pricing CIO playbook.
The strategic question for the CIO is not whether to use Power Platform. The platform's value has been demonstrated across thousands of enterprise estates. The question is how to license it economically while preserving the speed and flexibility that justify its existence. The buyer side answer addresses both at once.
The Power Platform landscape in 2026
Microsoft Power Platform is six commercially distinct product families that share a runtime, a data layer, and an admin center.
- Power Apps. Low code application development.
- Power Automate. Workflow automation, including the desktop RPA flows previously branded as Power Automate Desktop.
- Power BI. Business analytics.
- Power Pages. Low code public facing portals, formerly Power Apps portals.
- Copilot Studio. Custom Copilot agents, formerly Power Virtual Agents.
- Dataverse. The underlying data platform that supports them all.
Each family is licensed independently. Each has its own SKU structure, its own metering, and its own renewal characteristics. Customers commonly buy three to five families together. The interactions between the families produce most of the licensing complexity. Power Apps that consume premium connectors create per app license requirements. Power Automate flows that move data into Dataverse consume Dataverse capacity. Power BI dashboards that use Power Apps embedded objects require Power BI capacity in addition to Power Apps license. The interactions are not always documented in Microsoft's licensing guide.
Licensing models
Each Power Platform family offers two or three licensing models. The choice between them is the most consequential licensing decision the CIO makes for the platform.
Power Apps licensing
Power Apps is licensed three ways.
Power Apps licensing models
| Plan | Approx price | Suitable for |
|---|---|---|
| Per app plan | $5 per user per app per month | A small number of named applications used by named populations. |
| Per user plan | $20 per user per month | Unlimited Power Apps access. |
| Included in M365 plans | Bundled | Standard connectors and limited Power Apps use within standard M365 data sources. |
Premium connectors, custom connectors, and Dataverse use require a paid Power Apps license.
The most common cost trap is the assumption that Microsoft 365 entitled users can run Power Apps freely. Once the Power Apps in question use any premium connector (a SQL Server connection, an Azure DevOps connection, a third party SaaS connector that is on the premium list), the user crosses into paid licensing territory. Customers without governance discover this at audit, when Microsoft's compliance check identifies users running premium connector apps without Power Apps premium licenses.
Power Automate licensing
Power Automate is licensed in four ways.
Power Automate licensing models
| Plan | Approx price | Use case |
|---|---|---|
| Per user plan | $15 per user per month | Standard cloud flows. |
| Per user with attended RPA | $40 per user per month | Adds attended desktop automation. |
| Per flow plan | $100 per flow per month | Service account or system flows that run independently of any named user. |
| Process plan for unattended RPA | $150 per process per month | Licenses the unattended robot. |
The most common cost trap is the per flow plan. Customers who run dozens of system flows on per flow plans accumulate cost faster than they realize. The negotiation lever is to consolidate flows, to convert per flow to per user where the user count is small, and to negotiate enterprise pricing on the per flow plan when the count exceeds 50.
Power BI licensing
Power BI is licensed four ways. Power BI Pro is roughly $14 per user per month. Power BI Premium per user is roughly $24 per user per month and adds advanced features such as paginated reports and AI features. Power BI Premium per capacity (P SKUs) is roughly $5,000 per capacity per month for P1 and is suitable for organizations with more than approximately 500 Power BI users. Microsoft Fabric is the next generation analytics platform that bundles Power BI capacity with data engineering, data warehouse, and real time analytics workloads in a single SKU.
The most common cost trap is sticking on Power BI Premium per capacity (the P SKUs) when Microsoft Fabric would deliver the same Power BI value with broader analytics inclusion. The migration to Fabric is generally cost neutral to positive, but Microsoft's framing depends on the customer's negotiation. Customers who do not raise the Fabric question pay for legacy capacity SKUs at full price.
Copilot Studio licensing
Copilot Studio is licensed by message bundle. The standard SKU is roughly $200 per tenant per month for 25,000 messages per month, with overages priced per thousand messages. Copilot Studio's metering is the most opaque in the Power Platform family. The message count includes user messages, AI responses, and certain background processing, with the exact accounting differing between deployment patterns. Customers who deploy Copilot Studio without metering controls discover the message economics at the end of the first quarter, when the overage charges arrive.
Capacity vs per user: the strategic choice
The strategic licensing choice is between per user models and capacity models. Per user models scale with named users and are simpler to administer, but they are economically inefficient at scale. Capacity models pool resource across the tenant and are economically efficient, but they require capacity planning and can produce throttling under burst load. The right choice depends on user count, workload pattern, and the customer's appetite for capacity management.
| Pattern | Per User | Capacity |
|---|---|---|
| Small named user count | Recommended | Over provisioned |
| Large named user count, predictable load | Expensive | Recommended |
| Bursty load, citizen development | Recommended for protection | Throttling risk |
| Embedded apps for external users | Not licensable | Required |
The crossover point between per user and capacity for Power Apps is roughly 1,000 users. Below 1,000, per user is generally cheaper. Above 1,000, capacity becomes attractive. The crossover for Power BI is roughly 500 users. The crossover for Power Automate is more complex because flow consumption can be bursty. Customers should model both options with their actual user and workload data, not with Microsoft's reference scenarios.
Dataverse and storage
Dataverse is the data layer that underpins the Power Platform. Every Power App that uses Dataverse, every Power Automate flow that writes to Dataverse, and every Copilot Studio agent that reads Dataverse, consumes Dataverse capacity. Dataverse storage is metered in three units: database storage (rows and tables), file storage (blobs), and log storage (audit logs). Each is metered separately.
The cost trap is that Dataverse capacity is included in the Power Apps per user plan in modest quantities, but exceeds the included capacity in most enterprise estates. The overage pricing is per gigabyte per month, which adds up. Customers who deploy Power Apps with Dataverse heavy data models without capacity planning find that Dataverse overage exceeds the Power Apps license cost within the first year.
Dataverse capacity controls
The buyer side controls are: design Power Apps with Dataverse storage in mind, archive old data on a schedule, segregate audit log retention from production data retention, and run quarterly capacity reviews. Customers who treat Dataverse as a free SQL alternative pay for that mistake in capacity overage. Customers who treat Dataverse as a managed data platform with explicit capacity control achieve cost stability.
Copilot Studio economics
Copilot Studio is the newest commercially significant Power Platform family. The economics are still evolving in 2026. The message based metering produces unpredictable run rates. The customer's defensive posture should include three controls: message rate limiting at the bot configuration level, monitoring of message volumes per agent per day to identify anomalies, and the contracted right to convert message overages into pre commit at renewal rather than paying overage at peak rates.
The Copilot Studio licensing also intersects with Microsoft 365 Copilot. Customers running Microsoft 365 Copilot have access to a Copilot Studio capability set within the Copilot platform. The boundary between standalone Copilot Studio and Copilot Studio in Microsoft 365 Copilot is contested in many estates. The negotiation lever is to clarify the boundary in the EA, with explicit pricing for each scope.
Governance and the admin center
Power Platform governance is the most underdeveloped area in most enterprise Microsoft estates. The platform is sold as a citizen development platform. The pitch is speed and democratisation. The reality, in many estates, is hundreds of unmanaged environments, thousands of unmanaged apps, and untracked premium connector usage that produces audit exposure.
The Microsoft Power Platform admin center provides the controls. The controls work when they are used. Three controls matter most. Environment strategy, with a defined set of environments per business unit per workload type and a default environment that is locked down. Data loss prevention policies, applied per environment, that constrain which connectors can be used in which environment. Capacity allocation, with named capacity per business unit and quarterly review. The customers who run these three controls have measured Power Platform spend. The customers who do not have run rate that grows by 30 to 50 percent per year without explanation.
Shadow citizen development
The unmanaged side of Power Platform is the shadow estate. Apps built by individual contributors in the default environment, not catalogd, not maintained, not documented. The shadow estate has three risks. The first is audit risk: shadow apps using premium connectors create license exposure that the central team has not measured. The second is operational risk: the shadow estate becomes business critical without the support that business critical applications need. The third is data risk: shadow apps move data outside the controlled boundary, creating compliance exposure.
The buyer side response is not to stop shadow development. The response is to give it a managed channel. Citizen development environments with defined connector policies, capacity allocations, and a graduation path from shadow to managed. Customers who run this discipline preserve the speed of citizen development while controlling the licensing, operational, and data risks.
Audit risk and defense
Microsoft's audit motion against Power Platform has accelerated in 2025 and 2026. The audit follows the standard Microsoft audit clause in the EA or MCA. The Microsoft audit team has tooling that scans the customer's tenant for Power Apps, Power Automate flows, and Power BI usage that exceeds the licensed entitlement. The most common finding is premium connector usage by users who hold only Microsoft 365 entitlements without paid Power Apps or Power Automate licenses.
The defense framework is the same as the standard Microsoft audit defense framework. See our Microsoft 2025 to 2026 pricing playbook, the Microsoft EA renewal playbook, and the EA, MCA, CSP renewal proposal evaluation playbook. The Power Platform specific element is the connector inventory. The customer should maintain a documented inventory of which apps use which connectors, which users access which apps, and how the entitlement maps to the usage. Customers who hold this inventory walk into the audit prepared. Customers who do not hold it walk into the audit and produce a finding that becomes a settlement.
Negotiation levers in the EA
Power Platform pricing in the EA is negotiable on five dimensions. Customers who address all five preserve significant value. Customers who address one or none pay full list pricing.
- Volume discount.Power Platform list pricing is the starting point. Volume discounts of 15 to 35 percent are achievable for committed quantities, with the discount tied to a multi year commit and a documented adoption plan.
- Capacity bundle pricing.The Power Apps per user, Power Automate per user, Power BI Premium per user, and Copilot Studio bundle can be negotiated as a Power Platform suite at a discount to the individual SKU prices.
- Fabric conversion.The migration from Power BI Premium per capacity to Microsoft Fabric is a negotiation moment. The customer should secure pricing parity, conversion rights, and a multi year price hold.
- Copilot Studio overage.The right to convert message overage into pre commit at renewal, at a defined ratio, prevents punitive overage pricing.
- True down rights.The standard EA does not allow reduction of Power Platform license quantities mid term. Negotiated true down rights at annual anniversaries protect the customer when adoption falls short of plan.
Pattern study: a 35,000 employee insurer
A North American insurance group we advised was three years into Power Platform adoption when its central team commissioned a licensing review. The annual run rate had grown from $400,000 in the first year to $4.8 million by year three, against a forecast of $1.2 million. The growth was the result of three patterns. Citizen developers had built hundreds of Power Apps using premium connectors without licensing them. Power Automate per flow plans had proliferated. Power BI Premium per capacity sat at three P SKUs that were over provisioned for the actual workload.
The defense had four steps. We ran a tenant inventory using Microsoft's Center of Excellence toolkit to identify every app, every flow, and every connector. We rationalized licensing by converting 1,400 users from per app to per user, decommissioning 220 unused apps, and consolidating 80 per flow plans into 12 per user with attended RPA plans. We migrated from three Power BI Premium P SKUs to one Microsoft Fabric F64 SKU at lower cost with broader workload coverage. We renegotiated the EA at the next renewal, securing a Power Platform suite discount of 28 percent and a Copilot Studio pre commit at 65 percent of list. The customer's annualised run rate after the engagement was $2.7 million against an avoided trajectory of $7.2 million by year five.
For more Power Platform and Microsoft patterns see our case studies library and Microsoft EA negotiation case study.
The CFO view
From a CFO perspective the Power Platform is a fast growing line that must be measured at the unit level, not the contract level. The unit metrics that matter are licensed users per business unit per family, capacity utilization per family, premium connector usage per environment, and Copilot Studio messages per agent per month. CFOs who measure these units monthly catch growth before it becomes a renewal surprise. CFOs who measure only the EA total miss the per family growth until the renewal cycle exposes it. The CFO who funds the governance investment in year one captures the cost discipline in years two and beyond.
The Power Platform is a managed runtime, not a free addition to Microsoft 365. The buyer side discipline is to govern it as a managed runtime: environments, capacity, connectors, retention.
Closing thought
Microsoft Power Platform is one of the most strategic platforms in the Microsoft estate. It deserves CIO level attention to the licensing, the governance, and the renewal stance. The buyer side discipline is straightforward. Inventory the estate. Govern the environments. Control the capacity. Monitor the connectors. Negotiate the renewal. Customers who do all five preserve the value of the platform and contain the cost. Customers who do none of the five fund Microsoft's fastest growing product line at the expense of their own technology budget.
Redress Compliance is independent and 100 percent buyer side. We do not partner with Microsoft. We have advised on Power Platform licensing, governance, and audit defense across financial services, manufacturing, healthcare, public sector, and retail. If you are evaluating Power Platform licensing, planning a renewal, or facing an audit, the next step is a confidential briefing.
