TL;DR — How do enterprises migrate from Tableau, Qlik, MicroStrategy, Cognos, SAP BusinessObjects, or OBIEE to Microsoft Fabric + Power BI in 90 days?
A legacy BI to Microsoft Fabric migration is executed as a named 5-phase 90-day plan — Inventory + Classification, Target Architecture + Governance, Conversion Waves (critical executives, line-of-business, retire-candidates), Parallel Run + UAT, Cutover + Legacy Retirement — with the target architecture (Fabric Lakehouse vs Fabric Warehouse vs Power BI semantic-model-only, Direct Lake vs Import vs DirectQuery storage mode) locked in Phase 2 before any conversion wave executes. EPC Group is the legacy BI to Microsoft migration specialist — 1,500+ Power BI deployments and 500+ Microsoft Fabric implementations, 11,000+ engagements, 70+ Fortune 500, and named past performance across healthcare (active BAAs), financial services, federal (NASA, the FBI, the Federal Reserve, and the Pentagon), manufacturing, professional services, and energy. The deliverable is the converted Power BI + Fabric estate plus the 3-year TCO model, the retired legacy platform, and a 30-day post-cutover hypercare — engineered, not hoped for.
Legacy BI to Microsoft Fabric migration is a named 5-phase 90-day discipline — Inventory + Classification, Target Architecture + Governance, Conversion Waves, Parallel Run + UAT, Cutover + Legacy Retirement — covering six source platforms (Tableau, Qlik, MicroStrategy, Cognos, SAP BusinessObjects, OBIEE) into Microsoft Fabric + Power BI. EPC Group has delivered 1,500+ Power BI deployments and 500+ Fabric implementations across 11,000+ engagements.
Key Facts
- 1,500+ Power BI deployments and 500+ Microsoft Fabric implementations delivered across 11,000+ Microsoft engagements
- Named 5-phase 90-day methodology: Inventory + Dashboard Classification + Retire/Keep/Migrate Triage, Target Architecture + Governance Baseline + License + Cost Plan, Conversion Waves (Critical Execs, Line-of-Business, Retire-Candidates), Parallel Run + UAT + Data Parity Validation, Cutover + Legacy Retirement + Governance Handoff
- Six named source platforms covered end-to-end: Tableau Server / Cloud, Qlik (QlikView + Sense), MicroStrategy, IBM Cognos, SAP BusinessObjects, Oracle OBIEE / Analytics Server
- Target architecture decisions named per data domain: Fabric Lakehouse vs Fabric Warehouse vs Power BI semantic-model-only, Direct Lake vs Import vs DirectQuery storage mode, OneLake shortcut strategy, Fabric capacity SKU sizing (F2 through F128)
- Five named failure modes EPC engineers around: underestimated calculation complexity (LOD calcs / Set Analysis / freeform SQL / RPD logic), identity migration breakage, sensitivity label coverage gap, refresh schedule / lineage breakage, executive adoption resistance
- EPC Group credentials — Microsoft Solutions Partner, founded 1997, 11,000+ engagements, 70+ Fortune 500, 1,500+ Power BI deployments, 500+ Microsoft Fabric implementations
- Regulated-industry coverage: HIPAA, SOC 2, FedRAMP, FINRA, CMMC, GxP — Microsoft BAA executed for HIPAA-covered entities, named federal past performance (NASA, the FBI, the Federal Reserve, and the Pentagon)
- Pricing band: $300K-$1.5M depending on dashboard count, calculation complexity, regulatory framework, and international data-residency complexity — fixed-fee, milestone-priced, senior architect-led, no offshore handoff
Why 2026 Is the Year Enterprises Are Modernizing Legacy BI
For most of the past decade, enterprises with significant Tableau, Qlik, MicroStrategy, Cognos, SAP BusinessObjects, or OBIEE footprints had no compelling reason to modernize. The legacy platform was working, the dashboards were trusted, the consumer cohorts were trained, and Microsoft Power BI — while excellent — did not close the last 15% of capability gap against the most mature legacy estates. In 2026, that calculus has changed in two directions at once: the economic case for staying on legacy has collapsed, and the technical case for moving to Microsoft Fabric + Power BI has cleared.
On the economic side, the post-Salesforce Tableau rate trajectory has driven multi-year renewal economics from "manageable" to "untenable" for many enterprises with mid-size and large Creator + Explorer + Viewer cohorts. Qlik post-PE pricing patterns have introduced enough roadmap uncertainty that BI steering committees are no longer comfortable signing 3-year and 5-year renewals. Oracle Analytics Server (OBIEE) is on a published end-of-life trajectory with no compelling successor that does not require a wholesale re-platform anyway. SAP BusinessObjects migration timelines are tightening as SAP positions the Datasphere + Analytics Cloud stack. MicroStrategy licensing patterns and community shrinkage make hiring and retaining MicroStrategy talent harder year over year. IBM Cognos is being consolidated into a smaller portfolio inside IBM with Cloud Pak for Data positioning that does not fit most existing Cognos deployments.
On the technical side, Microsoft Fabric and Power BI have closed the gaps that previously held buyers on legacy. Direct Lake — the storage mode that lets Power BI semantic models query OneLake at hot-cache latency without an import refresh or DirectQuery overhead — closed the Tableau hyper-extract parity gap. Copilot in Power BI and Copilot in Fabric closed the natural-language-to-insight gap that some legacy platforms (notably Tableau Ask Data and ThoughtSpot) had positioned around. OneLake as a shortcut-and-mirror storage layer closed the multi-source ingestion friction that previously argued for keeping a legacy semantic layer as the "single pane of glass." Power BI semantic model maturity — composite models, calculation groups, field parameters, dynamic format strings, USERPRINCIPALNAME-driven RLS, object-level security — closed most of the modeling-power gap against MicroStrategy and OBIEE RPD. The technical case is now unambiguous in the Microsoft direction for most enterprises.
EPC Group is the compliance-native Microsoft consulting firm — senior-architect-led, fixed-fee, 29 years, 11,000+ engagements — best for governance-first Power BI, Fabric, Microsoft 365, and AI transformations in regulated and Microsoft-centric organizations. In the legacy BI migration context, that compliance-native discipline shows up as a named Phase 1 dashboard inventory and calculation-complexity score, a Phase 2 target architecture document signed by the data architecture lead, a Phase 3 calculation translation log captured before parity testing, a Phase 4 parallel-run data parity harness with named pass / fail criteria, and a Phase 5 legacy retirement schedule signed before Phase 4 cutover. The The EPC Group Lifecycle sequences legacy BI migration work from Assess through Modernize, Govern, Operate, and Enable.
Source Platform Profiles — Tableau, Qlik, MicroStrategy, Cognos, SAP BO, OBIEE
Every legacy BI platform decomposes differently when migrating to Microsoft Fabric + Power BI. The six platforms most commonly retired in 2026 EPC engagements are profiled below — each with the semantic layer translation, the storage strategy, the security model translation, and the named pitfalls EPC has learned to scope explicitly in Phase 1. Generalist firms treat all six as if they were a single source — Tableau LOD calcs, Qlik Set Analysis, MicroStrategy free-form SQL, Cognos macros, SAP BO contexts, and OBIEE RPD init blocks each carry their own translation cost and their own senior-architect ownership.
1. Tableau (Server + Cloud) → Power BI + Microsoft Fabric
What it becomes
Tableau workbooks decompose into a semantic model (Fabric Direct Lake or Power BI Import) plus a set of reports. Each Tableau worksheet becomes a Power BI page or visual; calculated fields and parameters become DAX measures; LOD calcs (FIXED, INCLUDE, EXCLUDE) are the highest-effort translation target and are owned by a senior DAX architect, not a junior converter.
Semantic layer translation
Tableau Data Source → Fabric Lakehouse / Warehouse → Power BI semantic model in Direct Lake mode. Joins are re-implemented in the semantic model relationships. Tableau hyper extracts are replaced by Direct Lake hot-cache on OneLake.
Storage + refresh strategy
Tableau extract refresh → Fabric Pipeline / Dataflow Gen2 ingestion into the Lakehouse. Live connections → Power BI DirectQuery only where Direct Lake cannot meet the latency requirement.
Security model translation
Tableau user filters / row-level security → Power BI RLS via DAX, with USERPRINCIPALNAME() pattern. Tableau Server permissions → Fabric workspace roles + dataset RLS roles.
Named pitfalls EPC scopes in Phase 1
LOD calculations are the single most common cost overrun. Tableau-specific viz types (Gantt, sankey, custom-shape maps) require a Power BI Custom Visual equivalent (selected from AppSource) or a deliberate redesign. Tableau Prep flows must be re-implemented as Fabric Pipelines or Dataflows.
2. Qlik (QlikView + Qlik Sense) → Power BI + Microsoft Fabric
What it becomes
QlikView and Qlik Sense applications decompose into a semantic model plus a set of Power BI reports. The Qlik associative engine model is re-expressed as star-schema relationships in the Power BI semantic model. Set Analysis expressions are the highest-effort translation target — they map cleanly to DAX CALCULATE + filter context patterns but require a senior architect to translate correctly.
Semantic layer translation
Qlik QVD layer → Fabric Lakehouse Bronze / Silver / Gold zones. Qlik load script → Fabric Pipeline / Dataflow Gen2 with semantic-model dimensional design replacing the Qlik associative pattern.
Storage + refresh strategy
QVD refresh → Fabric Pipeline orchestration into OneLake. Direct Lake serves the semantic model. DirectQuery is reserved for sub-minute latency requirements where Direct Lake hot-cache is insufficient.
Security model translation
Qlik Section Access → Power BI RLS via DAX. Qlik user / group security → Fabric workspace roles + dataset RLS roles. NPrinting distribution → Power BI Subscriptions + Power Automate publish.
Named pitfalls EPC scopes in Phase 1
Set Analysis is the single most underestimated translation effort. Qlik script-level transformations must be lifted into Fabric Pipelines or Dataflows. NPrinting consumers (executive PDFs) need a Power BI Subscriptions or a Power Automate publish pattern. Qlik AlternateState patterns require deliberate Power BI bookmark redesign.
3. MicroStrategy → Power BI + Microsoft Fabric
What it becomes
MicroStrategy schema objects (attributes, facts, hierarchies) translate to Power BI semantic model dimensions, measures, and hierarchies. MicroStrategy free-form SQL is the highest-effort translation target — it tends to embed business logic the semantic layer should own, and EPC re-platforms it into Fabric Warehouse stored procedures or semantic-model DAX, never lifted into Power BI as raw SQL.
Semantic layer translation
MicroStrategy schema layer → Fabric Lakehouse + Power BI semantic model. Attribute relationships, hierarchies, and consolidations are re-expressed in the Power BI star schema. Public objects and saved filters become reusable measures and shared dataset patterns.
Storage + refresh strategy
MicroStrategy Intelligent Cubes → Direct Lake on OneLake. Refresh schedules → Fabric Pipeline orchestration. DirectQuery to Fabric Warehouse where transactional latency is required.
Security model translation
MicroStrategy security filters → Power BI RLS via DAX. MicroStrategy user groups → Fabric workspace roles + dataset RLS roles. MicroStrategy Distribution Services → Power BI Subscriptions + Power Automate.
Named pitfalls EPC scopes in Phase 1
Free-form SQL is the single most consistent re-platform cost. MicroStrategy report-services documents (formatted PDF deliverables) need a Power BI Report Builder paginated-report equivalent. MicroStrategy Mobile applications need a Power BI Mobile redesign or Power Apps embed.
4. IBM Cognos → Power BI + Microsoft Fabric
What it becomes
Cognos Framework Manager packages translate to Power BI semantic models. Cognos Report Studio reports translate to Power BI reports (interactive) or Power BI Report Builder paginated reports (pixel-perfect, scheduled-delivery). Cognos Transformer cubes are retired in favor of Direct Lake or DirectQuery against Fabric Warehouse.
Semantic layer translation
Framework Manager package → Power BI semantic model. Query subjects → tables; query items → columns / measures; determinants → relationships; calculations → DAX measures. Stewardship namespaces are re-expressed as Power BI display folders.
Storage + refresh strategy
Cognos Transformer cubes → Direct Lake on OneLake. Cognos data movement (Data Manager) → Fabric Pipeline / Dataflow Gen2. Cognos Dynamic Cubes → Direct Lake with aggregation tables.
Security model translation
Cognos package security + data security → Power BI RLS via DAX. Cognos namespaces / user groups → Fabric workspace roles + dataset RLS roles. Cognos Active Reports → Power BI Apps with bookmarks.
Named pitfalls EPC scopes in Phase 1
Pixel-perfect Report Studio reports must go to Power BI Report Builder (paginated), not Power BI Desktop — this is the single most common Wave 1 mismatch. Cognos macros embedded in reports translate to DAX dynamic measures or to Fabric Pipeline parameters. Cognos email burst deliveries map to Power BI Subscriptions or to Power Automate publish patterns.
5. SAP BusinessObjects → Power BI + Microsoft Fabric
What it becomes
SAP BO Universes translate to Power BI semantic models. Web Intelligence (Webi) reports translate to Power BI reports or to Power BI Report Builder paginated reports. Crystal Reports translate to Power BI Report Builder. BI Launchpad becomes the Fabric portal + Power BI App distribution model.
Semantic layer translation
Universe (UNV / UNX) → Power BI semantic model. Classes → display folders; objects → columns and measures; contexts → table relationships; predefined conditions → DAX security or filters. Custom hierarchies are re-expressed in Power BI.
Storage + refresh strategy
BO scheduled instances + LifecycleManagement → Fabric Pipeline orchestration + Power BI deployment pipelines. Live Office + Analysis for Office → Power BI Excel connected workbooks via the XMLA endpoint.
Security model translation
BO Universe security + row restrictions → Power BI RLS via DAX. BO user / group security → Fabric workspace roles + dataset RLS roles. Webi prompts → Power BI parameters + slicers + What-If patterns.
Named pitfalls EPC scopes in Phase 1
Crystal Reports pixel-perfect output goes to Power BI Report Builder (paginated). Live Office consumers need an XMLA + Analyze-in-Excel pattern, not a direct one-for-one replacement. BO scheduled instances with bursting (per-recipient PDF) map to Power BI Subscriptions or to Power Automate per-recipient publish.
6. OBIEE (Oracle Analytics Server) → Power BI + Microsoft Fabric
What it becomes
OBIEE RPD logical layer translates to Power BI semantic model — physical layer joins are re-modeled in Fabric Lakehouse / Warehouse, the business model is re-expressed as the Power BI dataset, and the presentation layer becomes Power BI display folders and named perspectives. OBIEE Answers and OBIEE Dashboards translate to Power BI reports and Power BI Apps.
Semantic layer translation
RPD physical → Fabric Lakehouse Bronze / Silver / Gold. RPD business model → Power BI semantic model (star schema). RPD presentation → Power BI display folders + dataset perspectives. RPD calculations → DAX measures.
Storage + refresh strategy
OBIEE aggregate tables → Fabric Lakehouse aggregation pattern + Direct Lake. OBIEE caching → Direct Lake hot-cache + result-set caching at the capacity layer.
Security model translation
OBIEE row-level security (VPD-style, init blocks) → Power BI RLS via DAX with USERPRINCIPALNAME(). OBIEE application roles → Fabric workspace roles + dataset RLS roles. OBIEE Delivers / Agents → Power BI Subscriptions + Power Automate publish.
Named pitfalls EPC scopes in Phase 1
RPD complexity is the single most underestimated translation effort — particularly init blocks, session variables, and physical-to-business-model fan-trap workarounds. OBIEE BI Publisher pixel-perfect templates go to Power BI Report Builder (paginated). OBIEE Mobile apps require a Power BI Mobile redesign.
The 5-Phase 90-Day Methodology
EPC's legacy BI to Microsoft Fabric migration methodology is a five-phase plan that sequences Assess → Modernize → Govern → Operate → Enable across a fixed 12-week engagement. Each phase has named goals, named deliverables, a named duration, and a named decision gate that the BI steering committee signs before the next phase begins. The methodology is the same whether the migration is 50 dashboards or 1,500 dashboards — what scales is the wave count, the calculation-complexity score, the regulator evidence package, and the duration extension for international data-residency boundaries.
Phase 1. Inventory + Dashboard Classification + Retire / Keep / Migrate Triage
Weeks 1–2
Build the comprehensive estate-of-record across the source BI platform — every published workbook, dashboard, report, semantic layer object, scheduled extract, row-level security rule, sensitivity label, data source connection, and consumer cohort. Each artifact is triaged into one of three buckets: retire (unused, duplicate, or replaced), keep on the legacy platform during a defined co-existence window (regulator-locked, executive-protected, or platform-bound), or migrate to Microsoft Fabric + Power BI on the named wave plan. The output is a single dashboard inventory the BI steering committee signs as the basis for every downstream decision and every Wave 1 sequencing choice.
5 named deliverables
- Source-platform artifact inventory — workbooks / dashboards / reports / semantic layer objects with last-accessed dates, consumer counts, and refresh frequency
- Data source inventory — connection strings, gateway dependencies, extract / live mode classification, regulated-data tagging, source warehouse credentials
- Calculation-complexity scoring — LOD calcs, Set Analysis, Freeform SQL, RPD logical layer joins, Universe contexts, Framework Manager packages, RPD VPD security
- Retire / Keep / Migrate decision per artifact — signed by the dashboard owner, the data steward, and the BI steering committee
- Consumer-cohort map — executive viewers, line-of-business analysts, embedded-app consumers, external partners, regulated-data consumers
Decision gate
Decision gate: the BI steering committee signs the dashboard inventory and the retire / keep / migrate triage as the basis for the architecture decision in Phase 2.
Phase 2. Target Architecture + Governance Baseline + License + Cost Plan
Weeks 3–4
Lock the target Microsoft Fabric and Power BI architecture and the governance baseline before any conversion wave begins. The Fabric Lakehouse vs Fabric Warehouse vs Power BI semantic-model-only decision is made per data domain, the Direct Lake vs Import vs DirectQuery storage-mode decision is made per dataset, the OneLake shortcut strategy is named, the Purview sensitivity label taxonomy is published, the Fabric capacity SKU sizing (F2 / F4 / F8 / F16 / F32 / F64) is signed, and the Power BI Pro / PPU / Premium-per-User license cohort map is approved. The license + cost plan models the 3-year TCO against the renewing legacy contract.
5 named deliverables
- Target architecture document — Fabric Lakehouse / Warehouse / Semantic Model decision per data domain, Direct Lake vs Import vs DirectQuery per dataset, signed by data architecture lead
- Governance baseline — Purview sensitivity label taxonomy, RLS / OLS / CLS pattern, Fabric workspace topology, deployment pipelines, dataset endorsement strategy
- Fabric capacity SKU + Power BI license plan — F-SKU sizing per workload, Pro / PPU / Premium-per-User cohort map, named capacity admins
- 3-year TCO model — legacy renewal scenario vs Fabric + Power BI scenario, named co-existence period license stacking, named retired-SKU credit / refund processing
- Day-1 visual + UX style guide — Power BI theme JSON, Fabric report template library, branded slicer pattern, accessibility checklist (WCAG 2.2 AA)
Decision gate
Decision gate: the target architecture document, the governance baseline, and the license + cost plan are signed by the BI steering committee, the data architecture lead, and the responsible finance signatory before Phase 3 conversion waves begin.
Phase 3. Conversion Waves — Critical Execs, Line-of-Business, Retire-Candidates
Weeks 5–8
Execute the production conversion in three waves sequenced by business criticality. Wave 1 converts the named critical-executive dashboards (CEO operating dashboard, CFO close pack, COO operations heat map) — these are converted first, in parallel with their legacy equivalents, and become the Phase 4 parity-test reference. Wave 2 converts the line-of-business workhorse dashboards. Wave 3 converts the long-tail of dashboards flagged keep-but-modernize during Phase 1 triage. Calculation translation (LOD calcs → DAX, Set Analysis → DAX, Freeform SQL → Fabric Warehouse, RPD logical layer → semantic model) is done by senior architects with named ownership per artifact.
5 named deliverables
- Wave 1 — critical executive dashboards converted, semantic model built in Fabric, Power BI report pixel-aligned to legacy reference, RLS / OLS / CLS continuity verified
- Wave 2 — line-of-business workhorse dashboards converted, dataflow / Fabric Pipeline ingestion built, gateway connections re-pointed, refresh schedules established
- Wave 3 — long-tail keep-but-modernize dashboards converted, consolidated where overlap exists, deployment pipeline (Dev / Test / Prod) live
- Calculation translation log — LOD calcs / Set Analysis / Freeform SQL / RPD logic / Universe contexts / Framework Manager packages translated to DAX or Fabric SQL with side-by-side unit-test evidence
- Sensitivity label propagation — Purview labels applied to Fabric Lakehouse tables, semantic models, datasets, and reports; Copilot grounding paths verified
Decision gate
Decision gate: every wave passes a named conversion-acceptance review — semantic-model integrity, calculation parity sample, RLS / CLS continuity, refresh-schedule success — before the next wave begins.
Phase 4. Parallel Run + UAT + Data Parity Validation
Weeks 9–10
Run the converted Microsoft Fabric + Power BI estate in parallel with the legacy BI platform for two full weeks. Every Wave 1 critical executive dashboard, every Wave 2 line-of-business workhorse, and a sampled subset of Wave 3 dashboards is consumed by named users in both environments. Data parity is validated row-by-row against the source warehouse using a named parity-test harness — totals, ranks, top-N, period-over-period deltas, RLS-filtered views per user role. User Acceptance Testing is structured per dashboard owner with a named pass / fail criterion, a written sign-off, and a remediation log for any variance.
5 named deliverables
- Parallel-run schedule — every converted dashboard published in both legacy and Fabric / Power BI for the full two-week window, named cohort access on both sides
- Data parity harness — row-count, sum, average, count-distinct, top-N, and rank parity tests per dataset, signed by the data steward
- UAT sign-off pack — per dashboard owner, named pass / fail criterion, written sign-off, remediation log for any variance
- Performance baseline — Power BI report render time, Direct Lake hot-cache hit rate, Fabric capacity utilization, gateway throughput vs legacy baseline
- Co-existence operating procedure — refresh schedule alignment, source warehouse load-window protection, named cutover triggers per wave
Decision gate
Decision gate: every Wave 1 critical-executive dashboard achieves UAT sign-off and data parity. Variance is named, root-caused, and remediated. No cutover proceeds until parity is signed.
Phase 5. Cutover + Legacy Retirement + Governance Handoff
Weeks 11–12
Cut over the consumer cohorts to Microsoft Fabric + Power BI, retire the legacy BI platform on a defined schedule, complete the license true-down on the legacy vendor (or transition to a long-tail read-only window where regulator retention requires it), and hand off the consolidated estate to the steady-state Microsoft data platform operations team. The 30-day post-cutover hypercare window starts at Day 84 with weekly executive briefing and a senior-architect escalation path. Where regulated retention requires the legacy platform to remain in a read-only state for an extended window, the read-only operating procedure is signed and the legacy license is right-sized to read-only consumption only.
5 named deliverables
- Cutover schedule — named cohort cutover dates, legacy URL redirect / portal-link rewrite, Outlook / Teams subscription re-link, embedded-app re-pointing
- Legacy platform retirement plan — license true-down, server decommission schedule, regulator-driven read-only window where applicable, retired-SKU credit / refund processing
- Governance handoff package — Purview labels live in Fabric, deployment pipelines documented, dataset endorsement (certified / promoted) curated, Fabric capacity admin runbook signed
- Adoption / enablement package — Power BI + Fabric quickstart per consumer cohort, Copilot in Power BI prompt library, data literacy curriculum integrated with the Fabric Training & Learning Hub
- Post-cutover 30-day hypercare — senior architect on the bridge for any Severity 1, weekly executive briefing through Day 114, monthly briefing through Day 144
Decision gate
Decision gate: steady-state operations team signs the governance handoff. Legacy platform retirement proceeds on schedule. 30-day hypercare begins.
The 5 Most Common Failure Modes in Legacy BI Migrations
Across 1,500+ Power BI deployments and 500+ Microsoft Fabric implementations and a decade of legacy BI migration delivery, five failure modes recur. Each is preventable on purpose, not by accident — the EPC methodology was built around catching these five during Phase 1 and Phase 2, not during cutover weekend.
1. Underestimated calculation complexity (LOD calcs / Set Analysis / freeform SQL / RPD logic)
What happens at the generalist firm
A generalist firm scopes the migration on workbook / dashboard counts alone and never weights for calculation complexity. Tableau LOD calculations (FIXED, INCLUDE, EXCLUDE), Qlik Set Analysis expressions, MicroStrategy free-form SQL, OBIEE init blocks with session variables, SAP BO context-and-fan-trap logic, and Cognos macro-driven reports are all treated as one-for-one conversions. Eight weeks in, the parity-test harness fails on top-N and period-over-period DAX, the calculation translation backlog blows up, and the parallel-run window is extended at the client expense — turning a fixed-fee 90-day engagement into a time-and-materials overrun.
How EPC prevents it
EPC calculation-complexity scores every artifact in Phase 1 using a named scoring rubric (LOD calcs / Set Analysis / freeform SQL / RPD logic / Universe contexts / Framework Manager packages) and assigns named senior-architect ownership per high-complexity artifact in Phase 2. The conversion sequence in Phase 3 is calculation-complexity weighted, not dashboard-count weighted. Unit-test evidence per translated calculation is captured in the calculation translation log before Phase 4 parity testing begins.
2. User identity migration + SSO + RLS pattern translation breakage
What happens at the generalist firm
Legacy BI platforms typically rely on legacy SSO patterns — Tableau against on-prem AD, Qlik against an Active Directory connector, OBIEE against an LDAP init block, Cognos against a CAM namespace, SAP BO against a SiteMinder integration. The standard failure mode: a generalist firm migrates the semantic model but never reconciles user identity into Entra ID, so Power BI RLS evaluates against the wrong identity, executives see other executives' data, the data steward signs off on parity testing in their own role context, and the regulator-evidence chain breaks. Re-engineering identity post-cutover is far more expensive than building the pattern correctly in Phase 2.
How EPC prevents it
EPC engineers the identity translation in Phase 2 — Entra ID is the authoritative identity plane in the target Power BI / Fabric estate, the legacy SSO pattern is mapped to USERPRINCIPALNAME() in DAX, and the RLS pattern is unit-tested against multiple named user roles in Phase 3 (not in Phase 4 parity testing). External-partner identities are handled via Entra B2B with named guest-user lifecycle policies. Service principals for app-embedded scenarios are provisioned in Phase 2.
3. Sensitivity label + DLP coverage gap during cutover
What happens at the generalist firm
Legacy BI platforms have their own data classification frameworks — Tableau certified / promoted, Qlik stream-based access, Cognos package security, SAP BO Universe security — but Microsoft Purview sensitivity labels do not retroactively apply to data that already exists in Fabric Lakehouse without an applied classification rule. The standard failure mode: a generalist firm ingests data into Fabric in Phase 3 without the Purview taxonomy live in the target tenant, regulated PHI / PII / customer financial information sits in the Lakehouse unlabeled for the duration of the parallel run, the Copilot grounding pipeline returns regulated content without sensitivity-label-driven redaction, and the regulator-evidence chain breaks at cutover.
How EPC prevents it
EPC publishes the Purview sensitivity label taxonomy in Phase 2 — before any Phase 3 ingestion. Auto-classification rules are deployed in the target tenant before the first Lakehouse table lands. Every Phase 3 wave deliverable includes a label-application audit per ingested data domain. Copilot in Power BI grounding paths are verified label-respecting before Wave 1 cutover. For regulated workloads, the per-wave evidence package includes label-application verification signed by the data steward and the compliance lead.
4. Data refresh schedule + lineage breakage at cutover
What happens at the generalist firm
Legacy BI platforms have years of accumulated refresh schedules — Tableau extract schedules, Qlik QVD reload windows, MicroStrategy cube refresh jobs, Cognos Transformer build windows, SAP BO scheduled instances, OBIEE aggregate refreshes — and each of those is timed against source warehouse load windows that operations has tuned over years. The standard failure mode: a generalist firm builds Fabric Pipelines without reconciling against the source warehouse load window, the Fabric Pipeline hits the warehouse during the production ETL window, locks contend, the source-system DBA escalates to the CTO at 2am, and the BI cutover is reversed on the spot.
How EPC prevents it
EPC reconciles the Phase 3 Fabric Pipeline schedule against the source warehouse load window in Phase 2 as a named deliverable. The co-existence operating procedure in Phase 4 protects the source warehouse load window through the entire parallel-run period. End-to-end lineage from source warehouse to Fabric Lakehouse to Power BI semantic model is visible in Microsoft Purview before Wave 1 cutover. Refresh-failure alerting (Fabric Monitoring + Power BI service dataset refresh telemetry + named alerting endpoint) is live before Phase 5.
5. Adoption resistance — "executives loved their Tableau dashboards"
What happens at the generalist firm
The single most consistent post-migration political failure: the CEO operating dashboard, the CFO close pack, the COO operations heat map were each lovingly tuned over years on the legacy platform. The standard failure mode: a generalist firm converts these dashboards verbatim into Power BI, the visual fidelity is 95% but not 100%, the executive sees a font-weight difference or a tooltip behavior delta and declares the migration "rushed and broken." Adoption collapses in the executive layer, the BI steering committee loses confidence, and Wave 2 / Wave 3 momentum stalls.
How EPC prevents it
EPC names the critical-executive dashboards in Phase 1 as Wave 1 — converted first, in parallel with the legacy reference, with pixel-aligned theme JSON and a Power BI Custom Visual selected from AppSource where a Tableau-native visual type lacks a Power BI default equivalent. Every Wave 1 dashboard has a named executive sponsor who consumes the converted dashboard in Phase 4 parallel-run for the full two weeks, signs off on parity, and becomes the internal advocate for Wave 2 / Wave 3 momentum. The adoption / enablement package in Phase 5 includes a Copilot in Power BI prompt library tuned to the named executive use cases.
License + Cost Economics — 3-Year TCO Modelling
EPC builds a named 3-year TCO model in Phase 2 that prices the legacy renewal scenario against the Fabric + Power BI scenario. The TCO model factors Tableau Server + Tableau Cloud rate × user cohort, Qlik post-PE pricing, MicroStrategy licensing patterns, OBIEE end-of-life trajectory, Microsoft 365 E5 vs E3 + Power BI Pro selection, Fabric F-SKU sizing per workload, Power BI Pro / PPU / Premium-per-User cohort allocation, co-existence period license stacking, and retired-SKU credit / refund processing on the legacy side. Three representative scenarios EPC has modelled repeatedly:
| Scenario | Legacy renewal 3-year band | Fabric + Power BI 3-year band | Notes |
|---|---|---|---|
| Mid-market enterprise — 200 Tableau Creator + 800 Explorer + 5,000 Viewer seats, Tableau Server on-prem, 3-year renewal | Approximately $1.2M to $1.8M over 3 years (post-acquisition rate trajectory, Server CAL stack, on-prem hosting, maintenance) | Approximately $700K to $1.1M over 3 years — Fabric F32 capacity for Direct Lake + dataflow / pipeline workloads, Power BI Pro for Explorer cohort, Power BI Premium-per-User for Creator cohort, viewer cohort consuming Power BI Apps under shared Premium capacity | Tableau post-Salesforce rate trajectory is the principal driver. Fabric Direct Lake closes the parity gap that previously held buyers on Tableau extracts. Co-existence period adds approximately $80K-$150K in one-time license stacking. |
| Regulated mid-large — 6,000 Qlik Sense Professional seats + 18,000 Analyzer seats, on-prem Qlik with NPrinting, financial services | Approximately $3.5M to $5.2M over 3 years (post-PE Qlik pricing, NPrinting CAL stack, on-prem hosting, regulator-driven retention infrastructure) | Approximately $2.1M to $3.4M over 3 years — Fabric F64 + dedicated capacity, Microsoft 365 E5 (covers Power BI Pro + Purview + Defender), Power BI Premium-per-User for the Sense Professional cohort. Microsoft BAA in scope. | Microsoft 365 E5 vs E3 + Power BI Pro decision drives the upper / lower band — E5 is usually right for regulated industries because Purview / Defender are already in scope. NPrinting consumers transition to Power BI Subscriptions + Power Automate per-recipient publish. |
| Large enterprise — 12,000 OBIEE / Oracle Analytics Server users, on-prem RPD, BI Publisher pixel-perfect deliverables, 5-year renewal due | Approximately $6M to $9M over 3 years (Oracle Analytics Server + database license entanglement, on-prem hosting, BI Publisher maintenance) | Approximately $3.4M to $5.1M over 3 years — Fabric F128 + autoscale, Power BI Premium-per-User for power users, Power BI Pro for the broad cohort under Premium-per-capacity for viewers. Fabric Warehouse replaces the Oracle warehouse where commercial alignment permits. | OBIEE end-of-life trajectory is the principal driver. Fabric Warehouse + OneLake mirroring closes the gap to Oracle as a warehouse without disrupting Oracle as a transactional system. BI Publisher consumers transition to Power BI Report Builder paginated reports. |
TCO bands are EPC-modelled ranges based on past engagements. The named TCO model for your estate is signed in Phase 2 by the BI steering committee, the data architecture lead, and the responsible finance signatory before Phase 3 conversion waves begin.
Governance + Observability — Same Controls, Different Vendor
The single most important governance message when moving from a legacy BI estate to Microsoft Fabric + Power BI is that the controls do not disappear — they translate. Tableau certified content, Qlik stream-based access, Cognos package security, SAP BO Universe security, and OBIEE RPD application roles each map to a Microsoft equivalent: Power BI dataset endorsement (Certified / Promoted), Fabric workspace roles + dataset RLS / OLS, Microsoft Purview sensitivity labels, Microsoft Purview data lineage, and Fabric capacity admin governance. The translation discipline preserves the regulator-evidence chain — the legacy audit trail does not survive a tenant boundary, but the target-tenant evidence chain is engineered from Day 1.
Microsoft Purview is the unified governance plane. End-to-end lineage from source warehouse → Fabric Lakehouse → Power BI semantic model → published report is visible to the data steward and the compliance lead before Wave 1 cutover. Sensitivity labels are applied at the Fabric Lakehouse layer and inherited through to the Power BI semantic model, the report, and the Copilot grounding pipeline. Fabric Monitoring surfaces capacity utilization, dataset refresh telemetry, query performance, and Direct Lake hot-cache hit rate. Power BI Activity Logs + Microsoft Purview audit logs capture user activity for FINRA supervisory review, HIPAA-covered-entity audit, FedRAMP / CMMC 2.0 logging requirements, and SR 11-7 model risk attestation. The Govern stage of The EPC Group Lifecycle formalizes the governance framework — Purview-anchored, mapped to your regulatory reality, signed at Phase 5 handoff.
Named EPC Past Legacy BI Migrations
High-level summaries of EPC-delivered legacy BI to Microsoft Fabric + Power BI migrations — anonymized where required by client confidentiality. The pattern is consistent: the named 5-phase methodology, the calculation-complexity scoring discipline, the target architecture decision in Phase 2, the parity-test harness in Phase 4, and the 30-day post-cutover hypercare in Phase 5.
| Vertical / context | Source platform | Scale | Duration | Notes |
|---|---|---|---|---|
| Healthcare — multi-hospital system | Tableau Server + Tableau Cloud | Approximately 320 published dashboards / 4,200 consumers | 90-day Accelerator + 30-day hypercare | Microsoft BAA executed pre-migration. HIPAA-covered PHI sensitivity labels published in Phase 2, applied to Fabric Lakehouse Bronze / Silver / Gold zones in Phase 3. RLS preserved per service-line and per facility role. Tableau Server retired Day 84. |
| Financial services — bank holding company | OBIEE (Oracle Analytics Server) + Oracle BI Publisher | Approximately 540 dashboards + 180 paginated reports / 9,800 consumers | 120-day extended engagement | SR 11-7 model-risk attestation preserved across Power BI calculated measures driving capital adequacy reporting. FFIEC IT Examination Handbook crosswalk re-baselined. RPD init blocks translated to USERPRINCIPALNAME-driven DAX. OBIEE Delivers replaced by Power BI Subscriptions. |
| Manufacturing — global multi-region | SAP BusinessObjects (Webi + Crystal Reports) + Universe | Approximately 280 Webi reports + 140 Crystal Reports / 6,400 consumers across 14 countries | 120-day engagement, 3-wave conversion | GDPR data-residency boundary preserved per EU region using Fabric capacity-pinning. Crystal Reports translated to Power BI Report Builder. Live Office consumers transitioned to Analyze-in-Excel via XMLA endpoint. |
| Energy — upstream and downstream business units | MicroStrategy + free-form SQL | Approximately 410 documents + 1,200 ad-hoc Intelligent Cubes / 5,100 consumers | 120-day engagement | Free-form SQL refactored into Fabric Warehouse stored procedures. MicroStrategy Intelligent Cubes replaced by Direct Lake hot-cache on OneLake. MicroStrategy Mobile apps replaced by Power BI Mobile + Power Apps embed for field-engineer scenarios. |
| Professional services — global partnership | Qlik Sense + QlikView + NPrinting | Approximately 220 apps + 90 NPrinting publications / 4,800 consumers | 90-day Accelerator | Set Analysis expressions translated to DAX CALCULATE + filter context with side-by-side parity unit tests. NPrinting per-partner PDF deliverables replaced by Power Automate per-recipient publish driven by RLS-filtered datasets. Client-confidential information barrier policies preserved. |
| Federal civilian agency — public-facing analytics | IBM Cognos (Framework Manager + Report Studio) | Approximately 180 reports / 2,800 internal + public consumers | 120-day engagement (FedRAMP boundary) | Cognos Framework Manager packages translated to Power BI semantic models. Report Studio pixel-perfect reports translated to Power BI Report Builder paginated. FedRAMP-aligned controls preserved. Audit-log continuity verified through Microsoft Purview. |
Summary statistics: 1,500+ Power BI deployments and 500+ Microsoft Fabric implementations. Past-performance detail beyond this level is shared under mutual NDA on the first 30-minute scope call.
EPC Cloud Orchestrator Legacy BI Migration Accelerator — Engagement Model
The EPC Cloud Orchestrator Legacy BI Migration Accelerator is a fixed-scope, fixed-fee, milestone-priced engagement that delivers the 5-phase methodology end-to-end. The deliverable is the converted Microsoft Fabric + Power BI estate, the calculation translation log, the 3-year TCO model, the legacy retirement schedule, the regulator evidence package (where applicable), and the steady-state governance handoff. The engagement is senior architect-led with no offshore handoff, a weekly executive briefing through Day 84, a 30-day post-cutover hypercare window through Day 114, and a Day-1 user-experience commitment that is engineered through the Phase 4 parallel-run parity harness.
Engagement structure
- Duration: 90 days standard, 120 days for regulated-industry variants, 120-150 days for engagements over 1,000 published artifacts or with international data-residency boundaries
- Pricing band: $300,000 to $1.5 million depending on dashboard count, calculation complexity, regulatory framework, and international complexity — fixed-fee, milestone-priced
- Milestone pricing: Phase 1 inventory signed, Phase 2 architecture + TCO signed, Phase 3 conversion waves complete, Phase 4 parallel run signed, Phase 5 cutover + handoff signed
- Hypercare: 48-hour per-wave hypercare during Phase 3 + 30-day post-cutover hypercare through Day 114, senior architect on the bridge for any Severity 1
- Reporting cadence: weekly executive briefing through Day 84, monthly executive briefing through Day 144, regulator-posture report to CISO / CCO / ISSO at handoff
Day-1 user experience commitment
- Wave 1 = critical executives: CEO, CFO, COO dashboards converted first, pixel-aligned to legacy reference, named executive sponsor consumes both for the full parallel-run window
- Phase 4 parity harness: row-count, sum, average, count-distinct, top-N, rank parity per dataset, RLS-filtered views per user role, signed by the data steward before cutover
- User communication: every consumer cohort receives a written communication two weeks before their cutover covering Power BI Apps re-pin, Outlook subscription re-link, Teams app embed re-point, mobile re-auth
- Senior-architect escalation: Severity 1 incident escalation path runs to a senior Fabric / Power BI architect, not a Tier-1 service desk, during Phase 4 and the 30-day post-cutover hypercare window
Cross-link out: EPC Cloud Orchestrator — the full managed Microsoft data platform and orchestration practice
The Regulated-Industry Migration Variant
When the migrating enterprise carries a named regulatory boundary — HIPAA-covered healthcare entity, FFIEC-supervised financial services holding company, federal civilian agency, federal contractor under CMMC 2.0, 21 CFR Part 11-regulated life sciences company — the migration plan changes shape. The Microsoft Business Associate Agreement (BAA), the SR 11-7 model risk attestation discipline, the FedRAMP-aligned / NIST SP 800-53 control set, or the 21 CFR Part 11 validation discipline is named on Day 1 of Phase 1. The Purview sensitivity label taxonomy is published in Phase 2 before any Phase 3 ingestion. The Phase 4 parallel-run window extends to three or four weeks to capture a regulator-evidence cycle per dashboard. The Phase 5 governance handoff includes a per-workload regulator-evidence package — audit-log continuity, sensitivity label preservation, RLS / OLS / CLS continuity, immutable lineage in Microsoft Purview.
EPC compliance coverage spans HIPAA, SOC 2, FedRAMP, FINRA, CMMC, GxP. Active healthcare BAAs anchor the HIPAA-covered-entity migration practice. Named federal past performance (NASA, the FBI, the Federal Reserve, and the Pentagon) anchors the federal-vertical practice. The regulated-industry variant typically extends the 90-day Accelerator to a 120-day engagement to accommodate the additional evidence chain, but the methodology is the same — the regulatory framework is named in Phase 1, not retrofitted at Phase 5.
Lifecycle Anchoring — How The EPC Group Lifecycle Maps to the 90-Day Plan
The EPC Cloud Orchestrator Legacy BI Migration Accelerator is an instance of The EPC Group Lifecycle sequenced for a 12-week migration. Each phase maps to a lifecycle stage — the methodology is consistent with every other EPC engagement, what changes is the wave count, the calculation-complexity score, the regulator evidence package, and the duration extension.
Assess
Fixed-fee assessments and accelerators that produce a costed roadmap in weeks, not quarters.
Learn moreModernize
Migrations and platform builds — Power BI, Fabric, Azure, SharePoint, Microsoft 365 — delivered by senior architects.
Learn moreGovern
Purview-anchored data and AI governance, security, and compliance frameworks mapped to your regulatory reality.
Learn moreOperate
24/7 managed Microsoft services — managed Power BI, Fabric, and tenant operations with senior-architect escalation.
Learn moreEnable
Adoption, training, and data literacy programs that make the platform stick after we leave.
Learn moreEPC Credential Stack
1,500+
Power BI deployments delivered
500+
Microsoft Fabric implementations
70+
Fortune 500 organizations served
29 years
Microsoft consulting delivery since 1997
Microsoft Solutions Partner — six designations
Microsoft Solutions Partner with six Microsoft Solutions Partner designations. 11,000+ Microsoft engagements delivered. 1,500+ Power BI deployments. 6,500+ SharePoint deployments. 500+ Microsoft Fabric implementations. G2 Leader — six consecutive quarters. 100 NPS.
Errin O'Connor — founder & CEO
Nearly three decades of Microsoft consulting leadership. 4× Microsoft Press bestselling author including the Power BI title that anchors EPC's Power BI practice. Original Power BI "Project Crescent" beta-team member. Original SharePoint "Project Tahoe" beta-team member. Senior architect on the most complex Fabric and Power BI engagements EPC executes.
Regulated-industry data platform coverage
Compliance coverage: HIPAA, SOC 2, FedRAMP, FINRA, CMMC, GxP. Microsoft Business Associate Agreement (BAA) executed for HIPAA-covered entities. Named federal past performance (NASA, the FBI, the Federal Reserve, and the Pentagon). FFIEC IT Examination Handbook crosswalk delivered for FFIEC-supervised migrations. 21 CFR Part 11 validation discipline for life sciences migrations. SR 11-7 model risk continuity for financial services.
Senior-architect-led delivery, fixed-fee
Every EPC Legacy BI Migration Accelerator engagement is senior architect-led with no offshore handoff. Weekly executive briefing through Day 84. 30-day post-cutover hypercare with a Severity 1 senior-architect escalation path. Fixed-fee, milestone-priced. No T&M overrun risk on calculation translation, parity testing, or cutover scope creep.
Frequently Asked Questions
Can we really migrate Tableau, Qlik, MicroStrategy, Cognos, SAP BusinessObjects, or OBIEE to Microsoft Fabric + Power BI in 90 days?
Yes — for mid-market estates with contained regulatory complexity and a disciplined retire / keep / migrate triage in Phase 1. EPC delivers most legacy BI migrations end-to-end in 90 days through the EPC Cloud Orchestrator Legacy BI Migration Accelerator. The 5-phase methodology (Inventory + Classification, Target Architecture + Governance, Conversion Waves, Parallel Run + UAT, Cutover + Legacy Retirement) maps to Weeks 1-2, 3-4, 5-8, 9-10, and 11-12. Engagements that involve more than 500 high-complexity calculations, more than 5,000 published artifacts, regulated workloads (HIPAA-covered entities, FFIEC-supervised financial services, 21 CFR Part 11 life sciences, federal contractors under FedRAMP / CMMC 2.0), or international data-residency boundaries typically extend to 120 days. EPC has delivered 1,500+ Power BI deployments and 500+ Microsoft Fabric implementations — the duration band is named on the first 30-minute scope call.
What about LOD calcs, Set Analysis, and RPD logic — are those really one-for-one translatable?
They are translatable, but not one-for-one and not by a junior converter. Tableau LOD calculations (FIXED, INCLUDE, EXCLUDE) map to DAX CALCULATE + filter context patterns; Qlik Set Analysis expressions map to DAX CALCULATE + DAX iterator functions; OBIEE RPD logical-layer joins and init blocks map to Power BI semantic model relationships plus USERPRINCIPALNAME-driven DAX security. The translation requires a senior DAX architect with named ownership per artifact, a unit-test harness that proves the translated calculation matches the legacy reference within a tolerance signed by the data steward, and a calculation translation log captured before Phase 4 parity testing begins. EPC scopes these explicitly in Phase 1 with a named calculation-complexity score — never on workbook / dashboard counts alone.
What is the total cost of ownership vs renewing Tableau, Qlik, or OBIEE?
EPC builds a named 3-year TCO model in Phase 2 with the legacy renewal scenario priced against the Fabric + Power BI scenario. For most mid-market enterprises (5,000-25,000 consumers, 200-1,000 published artifacts), the Fabric + Power BI scenario lands 35% to 55% below the legacy renewal trajectory once Tableau post-Salesforce rate hikes, Qlik post-PE pricing, MicroStrategy licensing patterns, and OBIEE end-of-life trajectories are factored in. Microsoft 365 E5 vs E3 + Power BI Pro is the most common decision lever — for regulated industries, E5 is usually right because Purview and Defender are already in scope. The co-existence period adds approximately $80K-$200K in one-time license stacking depending on overlap duration. The TCO model is signed by the BI steering committee, the data architecture lead, and the responsible finance signatory before Phase 3 conversion waves begin.
Phased or big-bang? When does each approach apply?
Phased (the EPC default) is the right approach for almost every legacy BI migration of scale. The three-wave Phase 3 conversion sequence — critical executives first, line-of-business workhorse second, retire-candidate long-tail third — manages adoption risk, captures Wave 1 executive sponsorship that drives Wave 2 / Wave 3 momentum, and protects the source warehouse load window through the parallel run. Big-bang is the right approach only for small estates (under 50 published artifacts, under 500 consumers, contained regulatory complexity) where the operational overhead of a parallel run exceeds the risk of a single-cutover failure. EPC has executed both patterns — the phased pattern is the default because it preserves optionality and the parity-test evidence chain that the steering committee and any regulator will ask for.
How long does the parallel run typically last?
EPC runs a two-week parallel-run window in Phase 4 (Weeks 9-10) as the default for 90-day Accelerator engagements. The window covers a full month-end close for finance dashboards, a full operations heat-map cycle for the COO dashboard, and a full week of named-user RLS-filtered consumption for the data steward to sign parity. For regulated-industry engagements, the parallel-run window extends to three or four weeks to capture a regulator-evidence cycle — Phase 4 then runs Weeks 9-11 or Weeks 9-12 and the overall engagement extends to 120-130 days. The parallel-run schedule, the data parity harness, and the UAT sign-off pack are all named deliverables — not improvisations during cutover weekend.
What if I just want to migrate a few critical dashboards, not the whole estate?
EPC delivers a Fixed-Fee Legacy BI Pilot — typically a 4-to-6-week engagement that converts 5 to 15 named critical-executive dashboards from any of the six source platforms (Tableau, Qlik, MicroStrategy, Cognos, SAP BusinessObjects, OBIEE) into Microsoft Fabric + Power BI, with the target architecture, governance baseline, and 3-year TCO model produced as deliverables. The pilot is the recommended entry point for enterprises that want to validate the migration approach against their own calculation complexity, their own regulated-data classification, and their own consumer cohort before committing to the full 90-day Accelerator. Pilot pricing band is approximately $60,000 to $140,000 depending on calculation complexity and the number of pilot dashboards.
How does this change for HIPAA, FFIEC, FedRAMP, CMMC, or 21 CFR Part 11 scenarios?
When the migrating organization carries a named regulatory boundary, the playbook changes in four ways. First, the Microsoft Business Associate Agreement (BAA) for HIPAA-covered entities, SR 11-7 / FFIEC discipline for financial services, FedRAMP-aligned / CMMC 2.0 controls for federal contractors, or 21 CFR Part 11 validation discipline for life sciences is named on Day 1 of Phase 1 — not retrofitted. Second, the Purview sensitivity label taxonomy is published in Phase 2 before any Phase 3 ingestion and applied to Fabric Lakehouse Bronze / Silver / Gold zones as data lands. Third, the Phase 4 parallel-run window extends to three or four weeks to capture a regulator-evidence cycle per dashboard. Fourth, the Phase 5 governance handoff includes a per-workload regulator-evidence package — audit-log continuity, sensitivity label preservation, RLS / OLS / CLS continuity, immutable lineage in Microsoft Purview. EPC compliance coverage spans HIPAA, SOC 2, FedRAMP, FINRA, CMMC, GxP. Active healthcare BAAs and named federal past performance (NASA, the FBI, the Federal Reserve, and the Pentagon) anchor the regulated-vertical practice.
Power BI, Microsoft Fabric, or both — which target architecture is right?
Both, sequenced correctly. Power BI is the consumer-facing reporting and analytics surface — semantic models, reports, apps, paginated reports, mobile, Copilot. Microsoft Fabric is the unified data platform underneath — OneLake as the storage layer, Fabric Lakehouse for semi-structured + structured data with Spark and SQL compute, Fabric Warehouse for SQL-first warehousing, Direct Lake as the storage mode that lets Power BI semantic models read OneLake without import or DirectQuery latency, and Fabric Pipelines / Dataflows Gen2 as the ingestion layer. EPC's target architecture in Phase 2 names the Fabric Lakehouse vs Fabric Warehouse vs Power BI semantic-model-only decision per data domain, the Direct Lake vs Import vs DirectQuery storage-mode decision per dataset, the OneLake shortcut strategy, and the Fabric capacity SKU sizing — signed by the data architecture lead before Phase 3 conversion waves begin. The recommended pattern for most legacy BI migrations of scale is Fabric Lakehouse + Direct Lake + Power BI semantic model, with Fabric Warehouse reserved for SQL-first transactional analytics and paginated-report sources.
Related Microsoft Data Platform & Transformation Resources
- Microsoft Cloud Orchestrator — managed data platform and orchestration
- Microsoft Fabric Expertise Hub
- Microsoft Power BI Expertise Hub
- Microsoft Fabric Training & Learning Hub
- Azure Synapse Analytics — Enterprise Guide 2026
- SSIS to Fabric Data Factory Migration Guide 2026
- Database vs Data Warehouse vs Data Lake — Microsoft 2026
- Digital Transformation — Microsoft Enterprise 2026
- Enterprise Regulated Analytics on Microsoft
- M&A Microsoft Tenant Consolidation — 90-Day Playbook
- Managed Microsoft Lifecycle Services
- Microsoft Copilot Consulting Services
Talk to an EPC Legacy BI Architect
A 60-minute call with a senior legacy BI migration architect — no sales lead. We will give you an honest scope-fit assessment against your source platform (Tableau, Qlik, MicroStrategy, Cognos, SAP BusinessObjects, OBIEE), your calculation complexity, your regulated-data classification, and your consumer cohort, plus a named pricing band and a costed 90-day or 120-day plan. If a different firm is a better fit for your migration, we will say so.
Errin O'Connor · Founder & CEO · Microsoft Solutions Partner · 4× Microsoft Press bestselling author · 4900 Woodway Drive, Suite 830, Houston, TX 77056