Enterprise Power BI Performance and DirectQuery Tuning — a fixed-fee accelerator from EPC Group
Last updated by Errin O'Connor, Founder & Chief AI Architect, EPC Group
- Duration
- 4–6 weeks
- Team
- Senior architect named on the Statement of Work; engagement lead where the program size warrants it.
- How it is priced
- Fixed fee, quoted after a 30-minute scoping call. EPC Group does not publish price ranges.
What it is
Each model in scope is baselined — query duration, memory, refresh duration, capacity consumption — and then tuned: DAX refactoring and calculation-group consolidation, star-schema refactoring where a snowflake crept in, DirectQuery and Direct Lake tuning with partitioning and aggregations, and composite-model design where a hybrid of Direct Lake facts, imported dimensions and DirectQuery real-time data is the right shape.
A capacity sizing recommendation aligned to the tuned workload and a documented performance test suite close the engagement, so regressions are caught by the organization's team and not by its executives.
Who it is for
Enterprises with slow dashboards, DirectQuery timeouts, capacity throttling, refresh failures on large datasets or Fabric capacity exhaustion — typically five to thirty critical semantic models in scope.
Start here by size — Stand up the controls. Governance frameworks, security hardening and quick-starts for organizations that already run the platform and need it governed. The organizations EPC Group serves, and the ones it does not, are set out on the fit page.
Deliverables
Per-model performance baseline
Query duration, memory utilization, refresh duration and capacity consumption before any change.
DAX optimization
Measure refactoring, calculation-group consolidation, expensive-iterator replacement and query-folding analysis.
Star-schema refactoring
Snowflake to star, role-playing dimensions and slowly-changing-dimension handling.
DirectQuery and Direct Lake tuning
Query folding, partition strategy, aggregations and automatic aggregations.
Composite model design
Direct Lake facts, imported dimensions and DirectQuery real-time data where the workload calls for it.
Capacity sizing recommendation
Aligned to the tuned workload and concurrency, not the pre-tuning peak.
Documented performance test suite
Repeatable tests for ongoing regression detection.
Knowledge transfer and post-tuning playbook
How the client team keeps the models fast.
What is not included
- Source-system or warehouse tuning upstream of the model.
- Models beyond the agreed count without a scoping adjustment.
- Report redesign beyond the changes a model refactor requires.
Anything outside the deliverables above is a change order under the Engagement Charter — written, priced and approved before the work starts.
When to start
- Executives report slow dashboards or DirectQuery times out on real-time use cases.
- Fabric capacity is exhausted and the next step on the menu is a larger SKU.
- Copilot for Power BI is planned and the answers depend on model quality.
How it is priced
Fixed fee, quoted after a 30-minute scoping call. EPC Group does not publish price ranges. The scoping call is run by the architect who would lead the work; the fixed fee, the deliverables and the acceptance criteria are written into the Statement of Work, and the change-control, escalation and acceptance rules are the ones in the Engagement Charter.
Frequently Asked Questions
From the capacity and usage data: the models that consume the most capacity or serve the most critical audiences; the count is fixed in the Statement of Work.
Book the scoping call
A 30-minute call with the architect who would lead the Power BI Performance Tuning engagement; the fixed fee follows in writing.
Contact EPC Group