Power BI May 2026 Update: Visual Calculations GA, Exploration Perspective, and Copilot Summarize — Enterprise Implementation Guide

TL;DR

• Visual Calculations moved to General Availability in the May 2026 Power BI release. Running sums, moving averages, percent-of-parent, and other visual-scoped patterns no longer require new DAX measures in the semantic model. For enterprise tenants with strict semantic-model governance, this changes the calculation-authorship lifecycle.
• Exploration Perspective lets dataset owners scope the table and field surface presented to report consumers using Explore. For Fortune 500 semantic models with 200–800 tables, this is the difference between a usable self-service experience and a confused one.
• Copilot Summarize is now available at both the report level (ribbon button in Power BI Service) and the visual level (visual header click). Governance and Microsoft Purview labelling apply, but the surface area for inadvertent data exposure expands, and enterprise rollout requires careful Copilot capacity and audit configuration.
• Custom Totals generally available — author totals that do not match the underlying aggregation (median total, weighted average total, business-rule total) without measure proliferation.
• Power Query Get Data redesigned (Preview) for unified discovery across Fabric, Power BI Desktop, and Excel with accessibility-first navigation.
• Enterprises running Power BI Premium should sequence this work with a Premium → Fabric F-SKU readiness assessment; the May 2026 features are F-SKU first in some preview gates, GA across Premium and Fabric in others. EPC Group's typical compliance-native deployment pattern adds a Microsoft Purview classification check before Copilot Summarize is enabled tenant-wide.

Executive Summary

The Power BI May 2026 release is a pivot point for enterprise analytics teams. For three years, the dominant authorship pattern has been: data engineering owns the lakehouse, semantic-model architects own the DAX measures, and report authors compose the visuals. Visual Calculations, Exploration Perspective, Custom Totals, and Copilot Summarize all push some of that authorship surface area outward — toward the report author or the report consumer.

For governed enterprises in regulated industries (healthcare under HIPAA, financial services under SOC 2 and SOX, federal under FedRAMP), this is not automatically a win. Without a governance pattern, the new features can fragment metric definitions, expose sensitive measures through Copilot summaries, and bypass the central semantic model that compliance audits depend on. With a governance pattern, they reduce the model-bloat that has been the dominant Power BI scale problem for the last five years.

This guide walks through each feature, its enterprise implications, and the implementation framework EPC Group uses with Fortune 500 clients across healthcare, finance, and government. We cover the DAX/visual-syntax patterns, Microsoft Purview integration points, audit trail configuration, semantic-model migration considerations, and the 12-week rollout structure that keeps the semantic model authoritative while letting the new features deliver their productivity gains.

Why This Release Matters Now

Three structural shifts are converging in the Power BI ecosystem and Microsoft Fabric platform during 2026:

1. Semantic-model governance has become a compliance bottleneck. A typical enterprise Power BI tenant we audit has 40–200 published semantic models, 600–2,500 measures, and 1,800–9,000 active reports. New regulator expectations under SOC 2, HIPAA's 2026 access-control updates, and Basel III operational-risk reporting all push for a documented, audited single source of truth for analytical metrics. Adding a new DAX measure to support a single report's running-total chart used to be the path of least resistance. That path has now become the path of compliance friction.

2. The Premium → Fabric F-SKU migration window has narrowed. Microsoft's roadmap places Fabric F-SKUs as the long-term home for enterprise Power BI workloads. The May 2026 features arrive with different availability gates across Premium Per Capacity, Premium Per User, and Fabric F-SKU. Enterprises holding Premium contracts need a migration sequencing plan that aligns feature adoption with the capacity model that will host them at renewal.

3. Copilot has moved from preview to enterprise rollout. The Prep-Data-For-AI tooling format (also new in May 2026) standardizes how semantic models declare their Copilot readiness. The Summarize feature is the first Copilot capability that puts AI-generated descriptions of report data directly in front of every consumer who clicks Summarize. For data that carries Purview sensitivity labels, this surface needs to be locked down before it goes broad, not after.

Visual Calculations: General Availability

What changed from preview to GA

Visual Calculations entered public preview in early 2024 and have been gradually expanding. The May 2026 release moves them to General Availability, which carries three practical consequences for enterprise tenants:

• Tenant admin enablement defaults to On. Tenant administrators who had previously disabled the preview need to reaffirm their position now that GA defaults are in effect. The governance question for the next quarter is not "do we allow Visual Calculations" but "where do we draw the line between Visual Calculations and measures."
• Performance characteristics are now formally documented. Visual Calculations execute over the visual's already-aggregated data, not the underlying model. This is fast for simple running totals on a 12-row date axis but can become expensive on a large matrix with hundreds of rows and several added Visual Calculations. The GA documentation provides definitive guidance on which patterns scale and which do not.
• The expression library is broader. Patterns supported at GA include RUNNINGSUM, MOVINGAVERAGE, PERCENTOFPARENT, PERCENTOFPREVIOUSROW, COLLAPSE, EXPAND, INDEX, OFFSET, ORDERBY, and PARTITIONBY. Many of these would previously have required complex DAX measures with explicit ALL/ALLEXCEPT filter context manipulation.

Enterprise authorship pattern

Without a governance pattern, the natural human behavior with Visual Calculations is to skip the request to data engineering and just author the calculation in the report. That creates a fragmentation problem: the same metric ("rolling 12-month revenue") ends up implemented six times across six different reports, slightly differently each time, with no central definition.

EPC Group's standard pattern for governed tenants is a four-category authorship model:

1. Core metrics (the metrics the business officially measures performance against) live as DAX measures in certified semantic models. These are versioned, peer-reviewed, and audited. A report author cannot redefine them locally.

2. Visual-only patterns (running totals, moving averages, rank, percent-of-parent over visual axis) live as Visual Calculations. They are the right tool when the calculation is intrinsically tied to the visual's grain rather than the model's grain. Authorship is permitted in any workspace the author has Build access to.

3. Disposable exploration (ad-hoc analysis a single user does to answer a single question and never publishes) can use either. The rule is: if it leaves the analyst's session, it must enter category 1 or 2 with appropriate review.

4. Custom Totals (a row total that doesn't follow the column's aggregation rule — e.g., a weighted average row at the bottom of a salary table) are the right place for the new Custom Totals feature. Authoring these as a sum-and-divide DAX measure for every visual was the historical pattern; that pattern can now retire.

Practical Visual Calculations syntax

For report authors transitioning from DAX measures, the cheat sheet below shows the most common conversions. Each entry is the new Visual Calculation expression syntax (entered in the Visual Calculations editor in the Modeling pane on a visual).

Business pattern	Old DAX measure pattern	New Visual Calculation
Year-to-date sales	CALCULATE([Sales], DATESYTD(Date[Date])) (with caveats for fiscal year, ALL, etc.)	RUNNINGSUM([Sales]) partitioned by year
Three-month moving avg	Complex window measure with EARLIER or window functions	MOVINGAVERAGE([Sales], 3)
Percent of parent on a matrix	DIVIDE([Sales], CALCULATE([Sales], ALLSELECTED(Category)))	PERCENTOFPARENT([Sales])
Rank within partition	RANKX(ALLSELECTED(Category), [Sales])	INDEX(ORDERBY([Sales], DESC), PARTITIONBY(Category))
Previous row difference	Window-function DAX measure	[Sales] - OFFSET(-1, [Sales])

Governance and audit considerations

For tenants under SOC 2, HIPAA, or FedRAMP scope, the auditor's question is "where is this metric defined and how do we know it has not been tampered with?" A DAX measure in a deployed semantic model has a clear answer — the model is versioned in source control via the Tabular Model Definition Language (TMDL) format and deployed through a CI/CD pipeline. A Visual Calculation embedded in a report's visual lives inside the report file (.pbix or .pbip).

The governance question is therefore not "are Visual Calculations safe" but "are our report files under version control with the same rigor as our semantic models." For tenants where the answer is no, the May 2026 release is the right moment to fix that. The Power BI Project (.pbip) format, combined with Power BI Desktop Developer Mode and Git integration in Power BI Service, provides the foundation. EPC Group's typical pattern is:

• All certified reports stored as .pbip in an Azure DevOps Git repository with branch policies requiring code review before merge.
• A pre-commit hook scans for Visual Calculations and tags them in the PR description for reviewer attention.
• Reports flagged as containing Visual Calculations that should be model measures (the reviewer's call) are sent back for refactor before merge.
• The deployment pipeline pushes from Git to Power BI workspace via the Power BI REST API or Fabric Git integration.

Exploration Perspective: Curating the Consumer Experience

The scale problem this solves

A typical enterprise semantic model in healthcare or finance has 200–800 tables and 1,500–4,000 columns by the time it reflects the full operational data warehouse. When a report consumer clicks Explore on a visual to ask "what else is in here," they are presented with the full surface. For most consumers, this is overwhelming. The Power BI usability research that drove this feature found that consumers in unrestricted Explore sessions abandon the feature after the first time they cannot find what they were expecting to see in the top of the field list.

Exploration Perspective lets the semantic-model owner declare a scoped perspective — a subset of tables, fields, and measures — that consumers see when they Explore from this report. Different reports can use different perspectives over the same model. A finance-team report can present the GL-detail perspective; an executive report can present the curated KPI perspective.

Implementation steps

For a tenant rolling this out across the enterprise, EPC Group's pattern is:

1. Audit existing reports and identify exploration candidates. Not every report needs an Exploration Perspective. The ones that do are the high-traffic, broadly distributed reports where consumers click "Show more visuals" or Explore at high rates. The Power BI Service usage metrics provide this data.

2. For each candidate, define the consumer's "expected" field set. This is a stakeholder workshop, not a technical decision. The 12–30 fields the audience expects to see should be the perspective. Everything else should be hidden from the perspective and reachable only by switching perspectives.

3. Author perspectives in Tabular Editor. Power BI Desktop's modeling UI does not yet support perspective authorship at full fidelity for Exploration; Tabular Editor (free or paid version, depending on the team's licensing) is the production tool. Perspectives are saved into the TMDL definition and deploy with the rest of the model.

4. Set the report's Exploration Perspective in Power BI Desktop. In Report settings, the new "Default exploration perspective" field accepts the perspective name from the model.

5. Communicate the change to report consumers. Existing users who had memorized field positions will be temporarily disoriented. A short adoption-team announcement reduces support tickets.

Governance benefit

For tenants where the semantic model contains PII, PHI, or financial detail subject to confidentiality controls, Exploration Perspectives provide a defense-in-depth layer on top of Row-Level Security (RLS) and Object-Level Security (OLS). RLS and OLS remain the authoritative controls — they prevent unauthorized data from being returned by the engine. Perspectives are a usability and noise-reduction control on top of that. They keep sensitive fields off the consumer's default exploration surface even when the consumer technically has access to them.

This is particularly valuable in healthcare deployments where a clinical operations dashboard surfaces patient-volume metrics that any nurse manager can see, but the underlying model also contains MRN-keyed detail that only a small set of clinical analysts should pivot on. The perspective hides the MRN-keyed columns; RLS prevents unauthorized rows; both layers work together.

Copilot Summarize: Enterprise Rollout Considerations

The capability

The May 2026 release ships two Copilot Summarize entry points:

• Report-wide summary via a Summarize button on the Power BI Service report ribbon. Copilot generates a concise, paragraph-style summary of trends, performance highlights, and notable changes across all pages and visuals in the report.
• Visual-level summary via a click on the visual header. Copilot generates a summary focused on the selected visual.

Both summaries are generated server-side using the tenant's Fabric Copilot capacity and respect Microsoft Purview sensitivity labels applied to the underlying semantic model and the report.

The enterprise rollout question

Three weeks before a typical enterprise launches Copilot Summarize tenant-wide, the data security team should validate the following:

1. Sensitivity label propagation. Every semantic model serving certified reports should have a sensitivity label. Where models contain mixed-sensitivity data, the label should reflect the highest-sensitivity element. Copilot will refuse to summarize where the requesting user does not have the required clearance.

2. Copilot audit logging. Microsoft Purview Compliance Manager captures Copilot prompts and responses for Microsoft 365 Copilot. The Power BI Copilot path uses Fabric capacity-level audit logging. The audit retention window, export pattern (typically to Microsoft Sentinel for regulated industries), and review workflow should be operating before broad rollout.

3. Capacity sizing. Copilot Summarize consumes capacity units (CU) from the assigned Fabric or Power BI Premium capacity. A report serving 5,000 weekly active users may see 1,000–2,500 Summarize invocations per week once the feature has been promoted. This is non-trivial CU consumption. The capacity-planning model should reflect the new workload.

4. Group Policy and tenant settings. The Fabric admin portal exposes per-group enablement for Copilot features. Most regulated-industry tenants we work with stage Copilot Summarize through a pilot group (typically a single business unit) for 4–6 weeks before enabling tenant-wide.

5. Prompt-injection and data-exfiltration mitigation. Copilot Summarize reads visual content; visuals contain user-defined titles, labels, and tooltips. A report author with malicious intent could craft a tooltip designed to confuse the summary. The mitigation is a content review for any report shared outside its authoring team, performed before the report is certified.

The Copilot Tooling Format ("Prep Data for AI")

The May 2026 release introduces a new format for storing the Copilot metadata that makes Power BI semantic models work well with Copilot. This is the artifact that tells Copilot "this measure is named Net Revenue but should be referred to as Revenue in summaries." Previously, this metadata was stored in a way that made source control and team collaboration awkward. The new format is Git-friendly: it is text-based, diff-able, and merge-friendly.

For enterprise teams running Tabular Editor-based development pipelines and TMDL semantic-model definitions, the new Copilot Tooling Format integrates cleanly into the existing repository. The pattern is:

/semantic-models/sales-finance/
  definition.tmdl
  model.bim (legacy fallback)
  /copilot/
    synonyms.json
    description-overrides.json
    sample-questions.json

EPC Group's pattern is to put the Copilot tooling files under the same code-review process as the rest of the semantic model. A change to a synonym is a model behavior change, and reviewers should approve it the same way they approve a DAX measure change.

Custom Totals: Closing a Long-Standing Gap

Custom Totals, also generally available in the May 2026 release, solve a problem Power BI users have wrestled with since the original release. When a matrix shows several rows of data and a total row, the total row is computed by re-evaluating the measure with the row context removed. For SUM-based measures, this matches what users expect. For AVERAGE-based measures, weighted-average measures, percentile measures, or any measure where the row-level computation does not naturally aggregate, the total row was traditionally either wrong or required a complex DAX pattern.

Custom Totals let the report author specify the total-row computation independently of the row computation. The two common patterns this unlocks:

• Weighted average total. A row of (Salary × Headcount) / SUM(Headcount) per department; the total row is the same expression over all departments instead of an average of averages.
• Business-rule total. A row that sums sales attributed to a region; the total row sums sales attributed to "All Regions" using a different attribution rule (typical in headcount, allocation, and operational reporting).

Migration from existing custom-total DAX patterns is mostly mechanical: the existing IF(ISINSCOPE...) or HASONEVALUE patterns can be removed once Custom Totals is enabled, and the total expression can be authored separately in the visual's calculation editor.

Premium to Fabric F-SKU Migration: Sequencing Decisions

For Power BI Premium customers, every Power BI release in 2026 is, in part, a Fabric migration release. The May 2026 features are available with the following caveats at the time of writing (verify against your tenant's current Fabric admin portal):

Feature	Premium Per Capacity (P-SKU)	Premium Per User (PPU)	Fabric F-SKU
Visual Calculations GA	Yes	Yes	Yes
Exploration Perspective	Yes (with model perspective set in Tabular Editor)	Yes	Yes
Copilot Summarize	Requires F-SKU Copilot capacity attached	Limited	Yes (native)
Custom Totals GA	Yes	Yes	Yes
Power Query Get Data (Preview)	Yes	Yes	Yes
Copilot Tooling Format	Read across all tiers; full edit experience F-SKU	Partial	Yes

For enterprises planning a Premium → F-SKU migration within the next 12 months, the recommended sequencing is:

1. Adopt Visual Calculations, Exploration Perspective, and Custom Totals on the current Premium capacity — these work cleanly across tiers and the feature-adoption work transfers to F-SKU without rework.
2. Pilot Copilot Summarize on a dedicated F-SKU capacity (an F2 SKU is sufficient for a 200-user pilot) before broadening Copilot rollout.
3. Sequence the Premium → F-SKU cutover by workspace, not by capacity. Move governed-finance workspaces first (typically the slowest-changing, lowest-risk content), then operational workspaces, then self-service workspaces last.
4. Plan for the Power BI Report Server (PBIRS) gap. PBIRS does not receive Fabric features. Any reports still hosted on PBIRS need a migration plan as part of the F-SKU transition.

EPC Group's typical Premium → F-SKU migration spans 90 days for a 5,000-user tenant. The May 2026 features sequence into that 90-day window without extending it.

EPC Group's 12-Week Enterprise Rollout Framework

For tenants with 1,000+ Power BI users, established governance, and regulated-industry compliance scope, the rollout pattern that consistently delivers without breaking the existing certified-report estate is:

Weeks 1–2: Inventory and impact assessment.

• Audit existing certified reports for measures that would migrate naturally to Visual Calculations.
• Audit semantic models for tables/fields that should be hidden from default Exploration perspectives.
• Run the Copilot data-readiness check on the top 20 certified semantic models.
• Capacity-planning model update for Copilot Summarize CU consumption.

Weeks 3–4: Governance update.

• Update the published Power BI Governance Policy with Visual Calculations rules (the four-category authorship model above).
• Update the Microsoft Purview sensitivity label catalog with Copilot Summarize implications.
• Update the Power BI deployment pipeline to handle .pbip files with Visual Calculations.

Weeks 5–6: Pilot.

• Enable the new features for a pilot business unit (typically 100–300 users).
• Author 3–5 reports using the new features as exemplars.
• Run a 2-week consumption-monitoring window.

Weeks 7–8: Training and adoption content.

• Update the internal Power BI training curriculum.
• Record short (5-minute) demo videos for each new feature.
• Schedule office hours for report authors during the broader rollout.

Weeks 9–10: Broad rollout.

• Enable features tenant-wide.
• Communicate the change through standard adoption channels.
• Monitor support tickets and adoption metrics.

Weeks 11–12: Audit and review.

• Review the new Visual Calculations against the governance policy.
• Review Copilot audit logs for anomalies.
• Capacity utilization review.
• Lessons-learned briefing to the analytics CoE.

Common Pitfalls We've Seen

Across the rollouts EPC Group has guided in 2026, the recurring patterns that cause problems are:

1. Skipping the semantic-model-versioning step before rollout. Tenants where reports are not under source control discover the hard way that Visual Calculations are easy to lose. A report author deletes a calculation, the file is saved over the old version, and the calculation is gone. Source control is no longer optional.

2. Enabling Copilot Summarize before completing the sensitivity-label coverage check. A model without a sensitivity label inherits the default-organization label. If that default is permissive, the Copilot summary surface broadens unexpectedly. Cover the labels first.

3. Authoring a Visual Calculation in a high-traffic certified report without governance review. This is the path that fragments metric definitions. The governance review can be lightweight (a peer review on the merge request), but it cannot be skipped.

4. Underestimating Copilot capacity consumption. F2 capacities are too small for production Copilot Summarize workloads at scale. Plan for F4 or larger if the report estate has 5,000+ weekly active users and Summarize is broadly enabled.

5. Forgetting that Exploration Perspectives are model-side, not report-side. A report cannot define a perspective that does not exist in the model. The model owner has to author the perspective; the report references it.

Frequently Asked Questions

When will Visual Calculations be General Availability across all tenants?

Visual Calculations entered General Availability with the May 2026 Power BI release. The tenant-level rollout window is typically 2–4 weeks following the release announcement; verify availability in your specific tenant via the Power BI Service "What's new" entry and the Fabric admin portal.

Do Visual Calculations replace DAX measures?

No. Visual Calculations operate on the visual's already-aggregated data and are scoped to the visual. DAX measures operate on the underlying semantic model. The two are complementary. For metrics that should have a single authoritative definition across the enterprise, use DAX measures. For visual-grain calculations like running totals over a date axis, use Visual Calculations.

Can I use Visual Calculations in Power BI Pro (non-Premium) workspaces?

Yes. Visual Calculations are available across Pro, Premium Per User, Premium Per Capacity, and Fabric F-SKU tiers at the May 2026 release.

What is the performance impact of Visual Calculations on a large matrix?

Visual Calculations execute over the visual's result set after the visual query has run. For small visuals (a 12-row date axis with 4 columns), the impact is negligible. For large matrices (thousands of rows × dozens of columns) with multiple stacked Visual Calculations, the impact is measurable and can exceed 1–2 seconds of additional render time. The mitigation is to keep Visual Calculations off the largest matrices unless the calculation cannot be expressed efficiently as a measure.

How does Exploration Perspective interact with Row-Level Security?

Independently. Row-Level Security continues to filter rows based on the consumer's identity. Exploration Perspective filters the fields that appear in the default exploration surface for the report. A consumer who has RLS access to a sensitive field can still surface it by switching perspectives if multiple perspectives are configured. Perspective is a usability layer; RLS is the security layer.

Does Copilot Summarize see the data, or only the visual rendering?

Copilot Summarize reads the data underlying the visual, subject to the consumer's RLS context. It does not read data the consumer does not have access to. However, the summary text is generated by a language model, and the prompt the model sees includes the visual's data. Sensitive data should be governed through sensitivity labels and the tenant-level Copilot rollout policy, not relied upon to be invisible to the summary.

Can I disable Copilot Summarize for specific reports?

Yes, in two ways. At the tenant level, Fabric admin portal settings can scope Copilot features by security group. At the report level, report authors can set the report's "Copilot status" to off; this disables the Summarize button for that report.

What sensitivity labels block Copilot Summarize?

Microsoft Purview sensitivity labels with the "block Copilot processing" flag — typically Highly Confidential and equivalent — prevent Copilot from generating summaries for the labeled content. The exact label names depend on the tenant's Purview configuration. Verify with the information protection administrator.

How does the new Copilot Tooling Format relate to TMDL?

TMDL (Tabular Model Definition Language) is the text representation of the semantic model itself. The Copilot Tooling Format ("Prep Data for AI") is the text representation of the Copilot metadata that supplements the semantic model. The two are separate but live in the same repository in EPC Group's recommended pattern, under different directories.

Are Custom Totals available in Power BI Desktop or only Service?

Power BI Desktop authoring is supported. The visual's calculation editor exposes the Custom Total option. Once authored and published, the Custom Total renders in both Desktop preview and Power BI Service.

What is the migration path from existing custom-total DAX patterns to Custom Totals?

The DAX patterns that handled custom totals (IF + ISINSCOPE, IF + HASONEVALUE, SUMX + value lookup) can be retired in favor of Custom Totals where the pattern is purely about overriding the row-total computation. Where the pattern was doing more than overriding totals — e.g., implementing a business rule that depends on filter context — the DAX measure may still be the right tool. Each existing pattern needs an individual evaluation.

How does Visual Calculations interact with Bookmarks and Drillthrough?

Visual Calculations persist with the visual definition. They are preserved across Bookmarks. Drillthrough targets carry their own Visual Calculations; a drillthrough does not propagate the source visual's Visual Calculations.

What is the Power Query Get Data redesign?

The May 2026 release introduces a redesigned Get Data experience in Power Query (still in Preview at release). It unifies data-source discovery across Microsoft Fabric, Power BI Desktop, and Excel with a single search experience, redesigned connection flow, and accessibility-first navigation including keyboard navigation and dark mode. The experience is opt-in during preview; the legacy Get Data flow remains available.

Should I wait for additional Fabric May 2026 features before starting a migration?

Most of the May 2026 announcements (Power BI features, Fabric Real-Time Intelligence enhancements, OneLake refinements) integrate cleanly with the existing Fabric architecture pattern. There is no architectural reason to delay a Premium → Fabric F-SKU migration that was otherwise ready. Coordinate the feature adoption with the migration sequencing as described above.

How does EPC Group's typical enterprise rollout differ for healthcare vs. financial services?

The core 12-week framework is consistent. The differences are in the governance overlays: healthcare tenants extend the Microsoft Purview labeling work to align with HIPAA Privacy Rule access categories and integrate with the EHR's audit logging; financial services tenants extend the Copilot rollout to align with SR 11-7 model risk management documentation. EPC Group has industry-specific implementation patterns for healthcare, financial services, federal, and manufacturing scenarios — each one consistent with the master pattern but adapted to the regulatory framework in scope.

Next Steps

If you are running an enterprise Power BI tenant and the May 2026 features are now visible in your environment, the practical next steps are:

1. Brief your Power BI Center of Excellence and tenant administrators on the rollout sequence.
2. Run the inventory described in Weeks 1–2 of the framework.
3. Update the governance policy to reflect Visual Calculations rules.
4. Plan the Copilot Summarize pilot.
5. Sequence with any in-flight Premium → Fabric F-SKU migration.

EPC Group works with Fortune 500 enterprises in healthcare, financial services, government, and other regulated industries to deliver governed Power BI and Microsoft Fabric implementations. Our consultants — including Microsoft Press bestselling author Errin O'Connor — bring 29 years of enterprise Microsoft consulting experience and a compliance-native delivery pattern that has been refined across thousands of implementations. We are Microsoft Solutions Partner with the core designations, and were historically the oldest continuous Microsoft Gold Partner in North America from 2016 until the program's retirement.

To discuss your Power BI May 2026 rollout or a broader enterprise analytics modernization, contact EPC Group for a 30-minute discovery call.