A mid-size manufacturer has run Power BI on a P1 capacity for four years. At renewal, its reseller recommends the obvious replacement: an F64 Fabric capacity, which matches P1 in compute. The finance team approves it, since the price looks similar. Nobody checks how the capacity is actually used. A later review shows that most of the workload consists of a few dozen reports refreshed overnight, with peak interactive use for two hours each morning. Only about 150 people view content regularly, and most of them already have Pro licenses through their Microsoft 365 plan. The company is paying for a large, always-on capacity it barely uses.
Stories like this are becoming common as organizations move from Power BI Premium to Fabric. The retirement of P SKUs forces a decision, and the easiest path is to buy the equivalent F SKU. That path is not always the right one. Power BI consulting has become a capacity-economics exercise. The key question is whether workloads truly need a Fabric F SKU, and which size, or whether they belong on Pro or PPU licenses. Consultants who default to like-for-like capacity without profiling workloads risk overbuying.
What Changed with Power BI Premium
Microsoft announced that Power BI Premium per-capacity SKUs would be removed from the purchase flow for new customers on July 1, 2024. Existing customers without an Enterprise Agreement could renew P SKUs until January 1, 2025, while Enterprise Agreement customers could continue renewing until the end of their agreement. At renewal after those points, customers transition to Fabric F SKUs.
Fabric capacities work differently from P SKUs in ways that matter for cost:
- More Sizes: F SKUs range from very small (F2) to very large, rather than the few sizes Premium offered.
- Flexible Billing: Capacities can be bought on a pay-as-you-go basis or reserved for a term at a lower rate.
- Pause and Resume: Pay-as-you-go capacities can be paused when not needed.
- Broader Workloads: F SKUs run Power BI and other Fabric workloads, such as data engineering, data warehousing, and real-time analytics.
- Viewer Licensing Threshold: On smaller F SKUs, users need a Pro or PPU license to view content, while larger capacities allow free-license users to view shared content.
These options make right-sizing possible. They also make it easy to buy the wrong thing.
Why Like-for-Like Is Often the Wrong Default
A P1 capacity maps to F64 in compute terms. Many transitions start and end there. The logic is simple: same size, similar price, minimal risk.
This approach ignores several questions:
- Is the capacity fully used? Many P1 capacities run well below their limits most of the day.
- Who views content? If most viewers already have Pro licenses, the free-viewer benefit of a large capacity may not matter.
- What workloads run? Some organizations use Premium mainly for larger dataset sizes or paginated reports, which smaller capacities or PPU may also support.
- When are peaks? If demand spikes briefly, smoothing and smaller capacities may handle it.
- What else will run on Fabric? If the organization plans to add data engineering or warehousing, capacity needs may grow; if not, Power BI alone may need less.
Without answers, a like-for-like purchase is an expensive guess. Sometimes it is right. Often it is not, and the cost of the error repeats every month.
Licensing Options Power BI Consulting Services Should Compare
Power BI consulting services should compare several paths, not just one. A Microsoft Power BI consultant who presents only a single option has usually skipped the analysis.
- Pro Licenses
Every user who creates or views shared content has a Pro license. This works well for smaller user bases and moderate workloads. Many organizations already have Pro through Microsoft 365 E5 or similar plans, which changes the math significantly.
- Premium Per User (PPU)
PPU gives individual users access to Premium features, such as larger model sizes, more frequent refreshes, and advanced AI features, without buying capacity. It suits organizations where a limited group needs Premium features.
- Small or Mid-Size F SKUs
A smaller Fabric capacity, such as F8, F16, or F32, can host specific workloads with Pro or PPU users viewing content. Pay-as-you-go capacities can be paused outside business hours.
- Large F SKUs
F64 and above allow free-license users to view content and suit large viewer populations, heavy workloads, and broad Fabric adoption. For organizations with thousands of casual viewers, this can be the most economical choice.
- Hybrid Approaches
Many organizations combine options: a large capacity for broadly shared content, smaller capacities for departmental or development work, and PPU for a group of power users.
Workload Profiling by a Microsoft Power BI Consultant
Right-sizing starts with data. A Microsoft Power BI consultant should profile current usage before recommending any capacity.
- Capacity Utilization: Use the Fabric Capacity Metrics app or equivalent to see compute use by hour, day, and workload.
- Refresh Patterns: Identify when refreshes run, how long they take, and which datasets are largest.
- Interactive Load: Measure query volume and concurrency during peak periods.
- User Counts: Separate creators, frequent viewers, and occasional viewers.
- Existing Licenses: Check which users already have Pro or PPU through other agreements.
- Feature Needs: List Premium features in use, such as paginated reports, large models, deployment pipelines, or specific AI features.
- Growth Plans: Estimate how usage and workloads will change over the next two to three years.
With this profile, the consultant can model costs for each option and show the trade-offs clearly.
Common Mistakes in F SKU Transitions
Several errors show up repeatedly when organizations move from Premium to Fabric.
- Ignoring Existing Licenses: Buying a large capacity to cover free viewers when most users already hold Pro licenses through their Microsoft 365 plans.
- Leaving Pay-as-You-Go Running: Choosing pay-as-you-go for flexibility, then never pausing the capacity, which can cost more than a reservation.
- Reserving Too Early: Committing to a large reservation before profiling, then discovering the workload needs less.
- Mixing Development and Production: Running test refreshes and experiments on the production capacity, which inflates peak use and triggers throttling.
- Opening Fabric Workloads Without Governance: Enabling data engineering and warehousing for everyone, so unplanned jobs consume capacity meant for reporting.
- Skipping Monitoring: Switching capacities without setting up the metrics app and alerts, so problems surface only when users complain.
Each of these is easy to avoid with a short planning phase. A few weeks of profiling and a clear governance plan usually cost far less than a year on the wrong capacity.
Modeling the Cost
A useful cost model compares, for each scenario:
- License costs for Pro and PPU users.
- Capacity costs for each F SKU, on pay-as-you-go and reserved terms.
- Savings from pausing capacities outside business hours, where practical.
- Costs of any additional capacity for development and testing.
- Expected growth in users and workloads.
The model should also consider performance risk. An undersized capacity may throttle during peaks, frustrating users. Fabric smooths usage over time, which helps absorb short spikes, but sustained overuse leads to delays. The goal is the lowest cost that still meets performance needs.
FinOps practices now extend well beyond cloud infrastructure. The FinOps Foundation’s State of FinOps survey for 2026 found that 64% of respondents now manage software licensing costs as part of their FinOps scope. Power BI and Fabric capacity decisions fit squarely within that trend.
Optimization Before Sizing
Capacity needs depend on how efficiently workloads run. Before settling on a size, it often pays to optimize the heaviest items:
- Reduce Model Size: Remove unused columns and tables, and use appropriate data types.
- Use Incremental Refresh: Refresh only recent data instead of full reloads.
- Stagger Refreshes: Spread scheduled refreshes to avoid overlapping peaks.
- Improve DAX: Rewrite slow measures that drive heavy query load.
- Retire Unused Content: Remove reports and datasets that nobody uses but still refresh.
These steps can lower the capacity required, sometimes by a full size or more. They also improve report speed for users, which makes optimization easier to justify even when licensing is not the main concern.
What to Expect from a Power BI Consulting Company
When evaluating a Power BI consulting company for a Fabric transition, ask how it approaches capacity:
- Do you profile current capacity use before recommending an F SKU?
- How do you factor in existing Pro and PPU licenses?
- Will you model several scenarios, including smaller capacities and hybrid options?
- What optimization work do you recommend before sizing?
- How do you monitor capacity after the transition?
A Power BI consulting firm that recommends F64 in the first meeting, without asking for usage data, may be following the default rather than the client’s needs.
Beyond Capacity: Planning the Transition
Capacity is the starting point, not the whole transition. Power BI development services also need to cover:
- Workspace Migration: Reassigning workspaces from P to F capacities and validating that content works.
- Governance: Reviewing tenant settings, especially as Fabric adds new workloads users can create.
- Security: Checking workspace roles, row-level security, and sharing settings.
- Monitoring: Setting up alerts for capacity use and throttling.
- Training: Helping administrators understand Fabric capacity management and billing.
Planning these steps with the capacity decision avoids surprises after the switch.
A Representative Right-Sizing
Return to the manufacturer. A workload profile shows that capacity use peaks at about 40% of F64 for two hours each morning and stays low for the rest of the day. Most regular viewers already have Pro licenses through Microsoft 365. The main Premium features in use are larger model sizes and paginated reports for a finance team of 20.
The consultant models three scenarios: F64 reserved, F32 pay-as-you-go paused overnight, and Pro for all viewers with PPU for the finance team. After optimizing two large models and staggering refreshes, the F32 scenario handles peak load comfortably. The company chooses a smaller capacity and a handful of PPU licenses, cutting its annual Power BI spend substantially compared with the like-for-like proposal. Capacity monitoring is set up to flag growth, with a plan to scale up if Fabric adoption expands.
Measuring the Outcome
After the transition, track:
- Capacity Utilization: Average and peak use against the chosen size.
- Throttling Events: Any periods when performance was limited.
- User Experience: Report load times and refresh success rates.
- Cost Per User: Total Power BI and Fabric spend divided by active users.
- Forecast Accuracy: How actual costs compare with the model.
Review these quarterly and adjust capacity as needs change, scaling up before throttling affects users.
Power BI consulting services should start with capacity planning, now that P SKUs are retiring and Fabric offers many more options. Workload profiling, licensing analysis, and optimization reveal whether an organization truly needs a large F SKU or would be better served by a smaller capacity, Pro, or PPU.
In fact, professional Power BI consulting services include capacity profiling and cost modeling for Fabric transitions. Before the next renewal, one question can save a great deal of money: how much of the current capacity is actually being used?
