Fix slow Power BI reports at the cause, not with a bigger capacity
Most slow Power BI reports are not a capacity problem. They are a model or DAX problem that more capacity only postpones. I find the specific columns, relationships and measures responsible, fix them, and show the before-and-after numbers.
Why Power BI reports get slow
A report is only as fast as the queries its visuals send. Those queries are answered by the VertiPaq storage engine, which is fast and multi-threaded, and the formula engine, which is single-threaded. Slow reports almost always spend their time in the formula engine or scanning far more data than they need.
The usual culprits are predictable: high-cardinality columns such as GUIDs and datetime stamps bloating memory; bidirectional or many-to-many relationships widening every filter; Auto Date/Time creating hidden tables; and measures that force row-by-row evaluation — iterators calling other measures, FILTER over whole tables, or nested CALCULATE with ALL.
Refresh problems follow the same pattern: queries that do not fold back to the source, full refreshes of tables that should be incremental, and too many refreshes scheduled at the same time.
Signs this applies to you
- Visuals take more than a few seconds, or show “visual has exceeded the available resources”
- The model grew past a few GB and keeps growing
- Refreshes time out or have to be retried
- Capacity was upgraded and reports are still slow
- Matrix visuals with totals are noticeably slower than the same data in a table
How I tune a model
- 1
Measure
Performance Analyzer and DAX Studio Server Timings for the slowest visuals; VertiPaq Analyzer for memory by table and column.
- 2
Find the cause
Separate storage-engine scans from formula-engine time, and model size from query shape. Each slow visual gets a specific root cause.
- 3
Fix in order of impact
Remove or split high-cardinality columns, simplify relationships, rewrite the worst measures, enable incremental refresh where it matters.
- 4
Prove it
Re-run the same measurements and report before/after timings and memory.
What you get
- Ranked list of the slowest visuals and measures with root causes
- Memory profile by table and column with removal/split recommendations
- Rewritten measures with before/after timings
- Refresh design changes (folding, incremental refresh, schedule)
- Before/after summary you can share with stakeholders
Power BI Health Check — Scoped and priced after a short call. Start with the fixed-scope Health Check to find and rank the causes from model metadata. If you want the fixes implemented, a performance sprint is scoped and priced separately. See all pricing.
Related reading
Frequently asked questions
- Will moving to a bigger capacity make my reports faster?
- Only if the capacity is saturated. A single slow visual is usually limited by the single-threaded formula engine or by an oversized model, and a bigger SKU does not change either.
- What does “visual has exceeded the available resources” mean?
- The query behind the visual hit a memory or time limit — commonly a measure that materialises a very large intermediate table, or a visual with too many fields at a high granularity. It is fixed by changing the measure or the visual, not by retrying.
- Do you need access to our data?
- For the audit, no — model metadata (a .bim, PBIP/TMDL folder or VertiPaq Analyzer export) is enough. For implementation I agree minimal access with you up front.