The question data leaders are now asking isn't whether to bring CRM and ERP data into Microsoft Fabric - it's which system they're running, what mechanism is available, and how much engineering effort the path actually requires. CRM and ERP Microsoft Fabric migration has moved from a forward-looking ambition to a practical, decision-ready question, because Microsoft and its partner ecosystem have shipped native connectors, mirroring capabilities, and Open Mirroring integrations for the majority of the enterprise application landscape in 2025 and 2026. This guide answers the question directly: system by system, mechanism by mechanism, with the current support status and the trade-offs that determine which path fits your organisation.

This is written for CDOs, Data Directors, Analytics Managers, and IT leaders who are evaluating how to unify operational data - customer interactions, sales pipeline, financial records, inventory - with the rest of their analytics estate in Microsoft Fabric. The answer differs meaningfully depending on whether you run Dynamics 365, Salesforce, SAP, Oracle, or a mid-market system like NetSuite or HubSpot.

Why CRM and ERP Data Belongs in Microsoft Fabric

CRM and ERP systems are the operational core of most enterprises - they hold customer records, sales activity, purchase orders, financial transactions, and inventory movements. They are also, historically, the hardest data to bring into a governed analytics layer. Every organisation that has tried to build a customer 360 view, cross-functional dashboard, or AI-driven forecast has eventually confronted the same obstacle: the operational data lives in a system designed for transaction processing, not analytics, and extracting it cleanly, reliably, and without performance impact on the source system is a non-trivial engineering problem.

Microsoft Fabric changes the calculus in two ways. First, OneLake as a unified data layer means that CRM and ERP data, once landed in Fabric, is immediately accessible to Power BI, notebooks, data engineering pipelines, and AI agents - without a second copy or a separate integration step. Second, the mirroring and connector ecosystem means that for the major enterprise systems, the extraction problem now has a supported, maintained answer that doesn't require a team of engineers to build and operate a custom pipeline.

"The most valuable analytics work - margin by customer, pipeline velocity by campaign, inventory risk by supplier - requires data from at least two systems. The question is no longer whether you can bring CRM and ERP into Fabric. It's how clean and how fast the path is for your specific system."

The Three Migration Mechanisms Explained

Before reviewing individual systems, it helps to understand the three distinct mechanisms Microsoft Fabric uses to ingest data from external CRM and ERP sources. The right mechanism for any given system determines the latency, engineering effort, storage cost, and long-term maintenance burden of the integration.

Microsoft Dynamics 365: The Native Path

Dynamics 365 is the most straightforward CRM and ERP Microsoft Fabric migration because Microsoft owns both platforms and has invested deeply in the integration. The mechanism differs depending on which Dynamics 365 module you're running.

Dynamics 365 Customer Engagement (Sales, Service, Customer Insights)

CE apps store data in Dataverse, which makes Dataverse Link to Fabric the natural path. The integration is configured directly in Power Apps - navigate to Link data → New link → Fabric link - and all Dataverse tables are automatically surfaced in OneLake. One architectural constraint to plan for: CE environments use an all-or-nothing table selection model. You cannot choose specific tables as you can with Finance and Operations apps. All Dataverse tables are included, which means storage consumption in OneLake can be significant for large environments. Table-level management controls were added in 2026 to help manage this.

Dynamics 365 Finance and Operations

F&O stores data in its own database, not directly in Dataverse, but it is linked to a Dataverse environment. F&O tables must be manually added via the Add tables option in the Fabric link wizard - they are not autoselected. Finance and Operations supports selective table inclusion, which gives you more control over OneLake storage consumption than CE environments. The legacy Export to Data Lake path was fully decommissioned in March 2025; any organisation still relying on it must have already migrated to Fabric link or Azure Synapse Link.

Salesforce: Native Connector and Mirroring Options

Salesforce is the most common CRM Microsoft Fabric migration after Dynamics 365, and it has multiple supported paths depending on your volume, latency requirements, and existing tooling.

Native Fabric Data Factory connector

Fabric Data Factory includes a native Salesforce Objects connector (V2, updated 2025) that supports both Copy Jobs and pipelines. It uses the Salesforce REST API for smaller volumes and switches to Bulk API 2.0 automatically for high-volume objects. Incremental loads use the SystemModstamp watermark field. This is the right path for organisations building batch or scheduled analytics pipelines against Salesforce - daily or hourly refreshes, not sub-minute latency. It requires no third-party tooling and is maintained by Microsoft.

Open Mirroring via partners

For near real-time Salesforce replication, Open Mirroring partners provide change-data-capture based sync into Fabric OneLake. CData Sync and Cloud Services both offer GA Salesforce Open Mirroring integrations. These reflect changes from Salesforce - new leads, updated opportunities, closed cases - into OneLake in near real-time without requiring Salesforce API polling on a schedule. This is the right path when Power BI dashboards or AI agents need Salesforce data that's current to within minutes rather than hours.

Third-party managed connectors

Fivetran, Airbyte, Hightouch, and Estuary all offer mature Salesforce-to-Fabric connectors. These are fast to configure and operationally predictable, but add a cost layer and a dependency on a third party for a connection that is core to your data estate. For most mid-market organisations, the native Fabric Data Factory connector is sufficient and removes that dependency.

SAP: Mirroring via SAP Datasphere

SAP is the most complex CRM and ERP Microsoft Fabric migration in the enterprise landscape - and as of FabCon 2026, it is also the most mature non-Microsoft system in terms of Fabric mirroring support. Mirroring for SAP Datasphere reached general availability at FabCon 2026, providing near real-time, no-ETL replication of SAP data into Fabric OneLake.

The architecture routes SAP application data through SAP Datasphere, which acts as the extraction and transformation layer before data is mirrored into Fabric. This covers the full SAP application landscape: SAP S/4HANA (on-premises and cloud), SAP ECC, SAP BW, SAP BW/4HANA, SAP SuccessFactors, SAP Ariba, and SAP Concur. SAP BDC (Business Data Cloud) integration with Fabric is on the 2026 roadmap. For organisations that already use SAP Datasphere, this is a low-effort addition. For those that don't, SAP Datasphere must be provisioned as the intermediary - which adds licensing and architecture complexity.

Fabric Data Factory also provides built-in SAP connectors (SAP HANA, SAP Table, SAP ODP, SAP CDC) for pipeline-based extraction where Datasphere-based mirroring is not appropriate. The dab Nexus Open Mirroring integration supports near real-time extraction from SAP ECC and S/4HANA on-premises and Private Cloud Edition without requiring SAP Datasphere, which is an alternative for organisations not yet on Datasphere. For Microsoft Fabric migration programmes that include SAP, Datasphere-based mirroring is the preferred architecture for organisations running cloud SAP, while dab Nexus suits on-premises SAP estates.

Oracle ERP and Oracle Database

Oracle has two distinct scenarios in the context of CRM and ERP Microsoft Fabric migration: Oracle Database (the database engine, often backing multiple applications) and Oracle ERP Cloud (the application suite including Fusion Cloud ERP, HCM, and SCM).

Mirroring for Oracle Database entered preview in late 2025, supporting Oracle Database instances on-premises, on Oracle Cloud Infrastructure (OCI), and on Exadata. This provides near real-time, no-ETL replication of Oracle data into OneLake in Delta Parquet format - which means any application backed by Oracle Database (including Oracle ERP, Oracle E-Business Suite, and custom applications) can get data into Fabric through this path without a custom extraction pipeline. Preview status means this path carries the usual preview caveats - no SLA, potential for breaking changes - and should be evaluated against production requirements before committing.

For Oracle ERP Cloud specifically, where the application exposes REST APIs rather than a directly accessible database, the path is Fabric Data Factory pipelines using the REST connector or OData connector, or a third-party managed connector. CData Sync's Open Mirroring integration also covers Oracle sources. The Oracle mirroring preview is the more architecturally clean path once it reaches GA, as it eliminates API polling and the associated rate limit and scheduling complexity.

NetSuite, HubSpot, ServiceNow, and Workday

Mid-market ERP and CRM systems have a more fragmented support picture than the tier-one enterprise platforms. The path for each depends on whether Fabric Data Factory has a native connector, whether a mirroring partner covers it, or whether a third-party managed connector is the only viable option.

NetSuite does not have a native Fabric Data Factory connector as a first-party offering, but CData Sync's Open Mirroring integration covers NetSuite and supports near real-time change replication into OneLake. Third-party tools including Hightouch, Estuary, and Peliqan also support NetSuite-to-Fabric pipelines. For most NetSuite customers, CData Open Mirroring is the most architecturally clean path currently available.

HubSpot has a native connector in Azure Data Factory (now surfaced through Fabric Data Factory) - V1 is at removal stage and organisations should upgrade to V2 before the cutoff. This supports pipeline-based Copy Activity for batch analytics use cases. No mirroring or near real-time path is available natively.

ServiceNow has a native Fabric Data Factory connector supported for Dataflow Gen2, pipelines, and Copy Jobs. This covers ITSM, ITOM, and custom ServiceNow table data for batch analytics. ServiceNow is frequently included in enterprise Fabric programmes to unify IT operational data with financial and customer data.

Workday does not have a native Fabric Data Factory connector as of July 2026 - it is on the community roadmap request list. Workday data currently reaches Fabric through third-party managed connectors (CData, Fivetran, Informatica) or custom REST API pipelines. This is the most constrained path of the commonly evaluated systems.

Full CRM and ERP Microsoft Fabric Migration Comparison

◑ = Partially native or partner-supported GA. ✓ = Fully native and GA. ✗ = No native path, third-party required.

How to Choose the Right Migration Path

The right CRM and ERP Microsoft Fabric migration path is determined by three variables: the system you're running, the latency your analytics use cases require, and the engineering capacity your team has to build and maintain custom pipelines.

For any organisation running Dynamics 365 - CE or F&O - Dataverse Link to Fabric is the unambiguous first choice. It requires no ETL, no separate storage, and is on Microsoft's primary investment roadmap. The June 2026 low-latency sync update means the freshness argument for building a custom pipeline essentially no longer exists for most use cases.

For Salesforce, the decision depends on latency. If hourly or daily refresh is sufficient for your dashboards and AI use cases, the native Fabric Data Factory connector is the right choice - it's lower cost, easier to govern, and requires no third-party dependency. If near real-time Salesforce data is a genuine requirement, Open Mirroring via CData or Cloud Services is the current GA path.

For SAP, the Datasphere mirroring path is cleanest for organisations already on SAP Datasphere or willing to adopt it. For on-premises SAP without Datasphere, dab Nexus Open Mirroring or the native ADF SAP connectors are the practical options. For Oracle, native mirroring is in preview and worth evaluating in a non-production environment - it will be the recommended path once it reaches GA.

For NetSuite and Workday, third-party managed connectors are currently the only reliable path. CData Open Mirroring covers NetSuite; Fivetran, Informatica, and CData cover Workday. Factor the ongoing cost and vendor dependency into the architecture decision.

Common Mistakes in CRM ERP Microsoft Fabric Migration

Choosing a custom pipeline when a native connector exists. Data teams familiar with Azure Data Factory sometimes default to building bespoke extraction pipelines even when a native Fabric connector or mirroring path is available. Custom pipelines add maintenance burden without adding capability. Always evaluate native options first.

Underestimating Dataverse storage consumption for CE environments. Fabric link for CE environments includes all Dataverse tables - including system tables and tables from installed Dynamics apps like Customer Insights. In large CE estates, OneLake storage consumption can be substantial. Use the Manage tables controls added in 2026 to unlink tables you don't need for analytics.

Treating mirroring as a data warehouse replacement. Mirroring lands raw operational data into OneLake in near real-time. It does not transform it, model it, or apply business logic. A Fabric data engineering layer - medallion architecture, semantic model, or warehouse SQL endpoint - is still required to turn mirrored CRM and ERP data into analytics-ready assets.

Not testing schema change handling. CRM and ERP schemas change as new fields, modules, and customisations are added. Before going to production, validate how your chosen connector or mirroring path handles new columns, renamed tables, and deleted fields. Some paths require manual re-sync; others handle schema evolution automatically.

Conclusion

The answer to which CRMs and ERPs can be migrated to Microsoft Fabric is now a confident "most of them" - with the qualifier that the mechanism, latency, and engineering cost vary significantly. Dynamics 365 has the most complete and lowest-effort path. Salesforce, SAP, ServiceNow, and HubSpot have native or GA-partner paths that work well in production. Oracle Database mirroring is in preview and worth watching. NetSuite and Workday require third-party managed connectors today.

For data leaders planning a Microsoft Fabric migration that includes CRM or ERP data, the first step is an honest audit of which systems are in scope, which analytics use cases have latency requirements, and which connector paths are mature enough for production commitment. Numlytics has delivered Fabric migration programmes across all of the major systems covered here - speak with a certified Fabric consultant to scope the right architecture for your estate.