Pay for the capacity you need
A capacity running at or over its limit shows up in different ways — reports that crawl, refreshes that fail outright, requests that get throttled. The instinct is to size up, and it does make the symptoms go away. But nothing underneath changed: you're paying more every month to absorb the same inefficiency. Optimisation is a one-off cost. An oversized capacity is a permanent one.
We help you find out which one you're dealing with, whether you're choosing a capacity for a new deployment or you've suspected for a while that the one you're on is larger than it needs to be.
The most common cause of a struggling capacity, and the hardest to spot from the outside. We profile which measures and queries actually cost you, and rewrite the ones that fight the engine.
Wide flat tables, snowflaked dimensions, bidirectional relationships, mismatched granularity. We move the model towards the shape the engine is built for.
Fabric already smooths background cost across the day, so what matters is total consumption, not scheduling: refreshing only what changed, and dropping refreshes nobody reads.
Import, DirectQuery or Direct Lake — each trades memory, query speed and source load differently. Choosing wrongly is expensive in both directions.
A page with twenty visuals fires twenty queries per open, per user. Layout is a capacity decision as much as a design one.
What your capacity actually consumes, how close it runs to throttling, and whether pausing is realistic for your users — bearing in mind a reserved capacity is billed whether it runs or not, so pausing and reserving cancel each other out.
Finding the expensive parts of a Power BI workload is a specialist job, and it goes much faster with the right instruments. We profile queries and measures, read VertiPaq statistics to see where memory and query time actually go, hunt down measures and columns nothing references any more, and check models against established best practice — increasingly with AI and MCP-based tooling alongside the standard instruments.
We have been doing this for years, across many deployments. That experience is mostly worth speed: the patterns that cause trouble repeat, so we tend to find them quickly rather than working through everything from first principles. Quick wins often surface in the first pass — and where they let you drop a capacity tier, the work can pay for itself within a few months.
A capacity recommendation is worth more when the people making it have nothing riding on the answer.
Your portal runs in your own Azure subscription, so your capacity bill goes to Microsoft — not to us. We earn nothing from you buying a bigger SKU.
Capacity metrics, query profiling and load testing — so the recommendation comes with reasoning attached instead of a number pulled out of the air.
If the savings wouldn't justify the work, or your capacity is already right-sized, we'll tell you that rather than invent a project.
Every environment is different — some have a large capacity and a long list of problems, some have no capacity yet and a decision to make. That's why we start by talking rather than by promising a number.
Half an hour with whoever knows the environment day to day — with them on the call there's nothing to prepare. What you're running, what's failing, what you're paying, and whether a capacity decision is coming up.
Sometimes it's clear there's real money on the table. Sometimes the answer is a quick win your own team can handle, or that your setup is already fine. We'll say which — including when the honest answer is that you don't need us.
Then we scope it properly and quote it before anything starts. The optimisation work itself is paid consulting: profiling and rewriting the measures that cost you, reshaping the model, reworking refresh strategy and storage mode, and re-testing until the numbers move.
Want the technical detail first? Read our write-up on what actually makes a capacity struggle — what to measure, and how to size deliberately instead of defensively.
A few questions so the first conversation is a useful one. We'll come back with our initial thoughts and whether it's worth talking further.
Would rather just grab a slot? Book half an hour directly.