About this series
This is the first in a six-part Power Query Corner series exploring the new Power Query experience. Each post takes a closer look at a different part of the experience, including what’s changing, why it matters, and how you can put it to work. Getting data doesn’t always begin with a connector name. You might know which database to use, recognize a team-maintained Fabric item, want to return to a source from yesterday, or already have the query logic you need.
A connector list is useful once you’ve decided how to connect. But choosing the right data often comes before that decision. We moved to a module-based approach so Get data can support those different starting points, rather than making every task begin with the same connector picker.
The idea is simple: start with what you know, then follow the route that fits the task. Modules organize discovery, reconnection, and authoring around that idea. They don’t remove the need to authenticate, select the right objects, inspect a preview, or maintain the query afterward.
This first Power Query Corner issue follows a sales-report scenario in the new Power BI Desktop experience. It also uses clearly labeled Power Query Online examples to explain the wider design, then looks at the five most frequently reported Get Data problems and the work addressing them. The customer-feedback analysis draws on publicly available reports in our community forum thread, where we track feedback on the new Power Query experience.
Choose a Get data route that matches your task
Priya is preparing a regional sales review. A colleague offers a spreadsheet export, but the team also maintains sales data in Fabric. Before choosing Excel or another connector, she needs to decide whether to reuse that maintained item, return to a known source, or establish a new connection.
A module is a task-focused view inside Get data. Home brings the available routes together; the other modules give each task room for its own search, context, and actions. In Desktop, that includes New, Recent data, OneLake catalog, Blank table, and Blank query. The names and choices offered depend on the product hosting Power Query.
Figure: Home brings different starting points into one Get data experience. View more opens the corresponding full module; a source shortcut starts that source’s connection flow. Official Desktop documentation screenshot. Source: Microsoft Learn
Know the source? Keep the direct route.
If Priya already has the required server, file location, or service URL, New remains the straightforward route: search for the connector or browse its category, then supply the required settings. Home can provide a shortcut without first opening the full list. Selecting a connector starts the connection workflow; it doesn’t immediately load data.
This matters for familiar work. The module-based approach isn’t an instruction to browse every module before choosing a CSV file. It should preserve a direct path for someone who already knows the source.
Know the business data? Discover the item first.
If Priya knows the team maintains sales data but doesn’t know its endpoint, OneLake catalog gives her a different starting point. She can find an item by its business context, then review its type, owner, location, and available endorsement or sensitivity information before deciding to connect.
Imagine two items, both named Sales: one is in the team’s maintained reporting location, while the other belongs to a personal test project. The name alone isn’t enough. Catalog context helps Priya ask the right questions before she commits the report to a source. Endorsement is a useful signal, not a guarantee of suitability, and discovering an item does not grant permission to read it.
The benefit is not merely another way to open the same dialog: it lets Priya choose an appropriate existing data item before looking up connection details or creating another export.
To learn more, refer to the OneLake catalog overview for the catalog’s broader discovery, governance, and security roles.
Used it before, or need a small starting table?
Recent data is the route back to a source Priya already recognizes. It avoids starting the decision again at the connector list, but a recent entry isn’t a snapshot of the data or a guarantee that its location and permissions are unchanged.
Blank table and Blank query answer a different question: what if the starting point isn’t an external source? Priya might enter a small RegionCode-to-RegionName mapping in Blank table or paste a self-contained M expression into Blank query. M is Power Query’s formula language. Entered values become part of the query; pasting a spreadsheet’s values does not create a refreshable connection to that spreadsheet.
Power Query Online examples, not Desktop capabilities
The broader module design also accommodates finding a source by browsing the place where it lives. Recognizing an Azure resource or a SharePoint site is different from knowing its server address or file URL. Where those source-browsing experiences are offered, they can carry the user from a recognizable location to the contents they want. Browsing into a location and selecting it for a connection are different actions, and a file’s contents are different from a folder’s file list.
Browse Azure, SharePoint browsing, and Upload is not currently available as modules in Power BI Desktop. They are part of Dataflow Gen2, and some are also available in Power BI web modeling; the available modules vary by host. There is no committed release date for these modules in Power BI Desktop. This module distinction should not be confused with the availability of individual Azure or SharePoint connectors through New. Search scope, filters, previews, and available actions depend on the browser and host. The useful principle is to support recognition and navigation where appropriate while retaining direct connector configuration.
Dataflow Gen2 illustrates another distinction: reusing data is not the same as reusing query logic. OneLake catalog helps find a data item; My Queries holds personal M scripts; Shared queries (Preview) let authors import logic shared by others. For example, Priya could reuse an existing region-cleanup script in a dataflow instead of recreating those transformations. Importing shared logic creates an independent copy, not a live link to its original, and does not transfer credentials or destinations.
To learn more, refer to the My Queries or Shared Queries documentation.
Figure: Dataflow Gen2 example, not Power BI Desktop: My Queries is a personal library of query logic, distinct from recent sources and from the separate Shared queries module. This published Dataflow Gen2 screenshot illustrates why different tasks need different entry points. Source: Microsoft Learn
Modules can also support tasks such as describing transformations with Copilot in Dataflow Gen2, subject to that feature’s requirements. That does not make every Power Query Online module a Desktop capability. Use Get Data in Dataflow Gen2 for its supported entry points and Data sources in Power BI Desktop for Desktop.
Try it: Choose and preview a sales source in Power BI Desktop or web modeling
You can use the October 2026 release of Power BI Desktop or try Power BI web modeling today with an authorized source in a nonproduction report. In Desktop, enable the preview: navigate to File > Options and settings > Options > Preview features, select New Power Query experience, and select OK. The exercise below is a workflow, not a supplied sample dataset.
For the documented preview prerequisite, refer to the Data sources in Power BI Desktopdocumentation. Power BI web modeling uses Power Query Online, not this Desktop toggle.
Define the decision: Priya needs sales by region for the intended reporting period. In Desktop or Power BI web modeling, choose a maintained item in OneLake catalog, a known source through New, or an appropriate Recent data connection; the available routes vary by host.
Connect with context: compare the item’s owner, type, and location, or provide the connector’s settings, then authenticate with your authorized account. Catalog visibility does not grant data access.
Preview before committing where Navigator is offered, select the intended object and check its columns and level of detail. In the new Desktop Navigator, use Transform data in the Load dropdown when shaping is needed; follow the web modeling controls for that host.
Check the outcome in the editor: confirm the selected query and its Source step and explain which source the report now depends on. Use the linked Microsoft Learn guidance for detailed connection steps and host-specific options.
Figure: The selected item’s type determines its available actions and subsequent connection path. This Desktop documentation screenshot shows Connect and Edit, not an Import/DirectQuery selector. It does not imply the same actions for every catalog item. Source: Microsoft Learn
The route changes how you find the source; it does not skip the source-specific connection and preview stages. Some connectors provide a table preview rather than Navigator.
For those patterns and host distinctions, refer to the Getting Data Overview. documentation.
In the Power BI service, use Create > Get Data for a new semantic model, or Get data in an existing model’s ribbon, with the required editing permissions. A semantic model contains the tables, relationships, and calculations used for analysis. This route is not a Dataflow Gen2 destination workflow.
Refer to the Edit semantic models in the Power BI service. documentation to learn more.
Compare the decision, not just the dialog layout
The legacy Desktop Get Data dialog is organized around connector categories: choose the source type, then connect. That remains an effective mental model when you already know the source. Legacy Desktop also has recent-source and Fabric-data entry points; the point is not that these ideas never existed before.
The new experience brings different starting points into an organized Get data surface. New starts with the connector; OneLake catalog starts with an existing data item and its context; Recent data starts with previous work; Blank table and Blank query start with values or code. Home helps choose among the available routes instead of acting as the only place where everything must fit.
That changes Priya’s first decision from “Which connector should I pick?” to “What am I trying to reuse or create?” For a maintained Fabric source, she can investigate the item before configuring a new source connection. For a known CSV file, she should still be able to take the direct route. In supported Power Query Online hosts, query-library and source-browsing modules extend the same principle to other tasks.
A common organizing pattern does not mean identical behavior. Power BI Desktop and Power Query Online differ in how connection settings, authentication, and load or save actions work. In the new Desktop Navigator, Transform data is inside the Load dropdown; the documented legacy flow has a separate Transform Data button. Choose the transformation route when the data needs inspection or shaping before loading.
For host-specific gaps and workarounds, refer to the New Power Query experience known limitations and workarounds documentatin. Some Desktop workflows still use a legacy interface while preview is enabled.
The design is useful only if those routes remain practical. Discovering the right data should not come at the expense of an established SQL-editing workflow, required Web options, a reliable Combine action, or prompt access to a familiar connector. That is why the feedback below matters: it tests the intended benefits against the work people need to finish.
State of the new UI
Status summary
The new Get data experience remains in preview. Alongside making sources easier to discover, an important part of this work is preserving the connection workflows people already depend on in the legacy experience.
We’ve addressed the missing native-SQL input when creating connections with supported connectors. A native SQL statement is a database query supplied directly to the source, rather than a sequence authored entirely through Power Query’s transformation controls.
Entering that statement when first connecting and reopening it later are different workflows. Coverage for reopening connection settings through the Source step’s gear icon is still partial for native SQL and other extension-based connectors. We’re continuing work on the underlying connection dialog to support these editing scenarios more consistently.
For the maintained reference to current gaps and workarounds, refer to the New Power Query experience known limitations and workarounds documentation.
What we’re hearing
For this first issue, we reviewed the discussion from its beginning on May 11 through October 26, 2026: the opening post and all 386 replies available at the review cutoff. The five themes below were selected by the number of distinct contributors explicitly reporting each problem, not the number of posts or likes. A person is counted once per issue and can appear in more than one theme. These counts describe the discussion, not the proportion of all customers affected.
Because this is a full-history review, the ranking includes the initial native-SQL gap that has since been addressed. We keep it separate from later editing problems so progress on one is not mistaken for completion of the other.
Entering native SQL is an essential part of connecting
Twenty-four distinct contributors in the first-run feedback review.
For some teams, connecting to a database starts with an existing SQL statement, not simply selecting a table. That statement defines the dataset they want to bring into the report and lets them reuse a familiar database-querying workflow.
One Amazon Redshift user reported that the custom SQL input was missing from Advanced options. Another explained that their BI group primarily used Redshift and disabled the preview after discovering the missing native-query field.
In their established connection workflow, entering SQL was part of getting started. Removing that input meant the new dialog could not accommodate how their team already prepared report data. This was a workflow blocker, not simply a preference about the dialog’s layout.
We’ve addressed the missing input for supported connectors. That change concerns creating the connection; maintaining its SQL afterward requires separate attention.
Figure: Legacy Desktop Amazon Redshift connection dialog from Microsoft Learn. Advanced options is shown collapsed; the connector documentation lists SQL statement among its advanced capabilities. This is a legacy workflow reference, not a screenshot of the preview fix. Source: Microsoft Learn
Creating a query is only the beginning: people need to edit it later
Sixteen distinct contributors in the first-run feedback review.
In the legacy experience, users could return to a query’s Source step, select its gear icon, and reopen the connection dialog to change the SQL statement. The gear icon is the shortcut for editing settings associated with that step.
An Azure Databricks contributor described the difference between creating and maintaining a query: the SQL input was available when creating a connection but disappeared when reopening its settings. They relied on that dialog for quick updates to existing reports, saving time and transformation steps, and reverted to the legacy experience when it was unavailable.
There is also a broader productivity concern. One detailed example explains how a team moves parameterized SQL between Power Query and DBeaver. They use reporting-period parameters so business users can change dates without rewriting SQL. Their editing workflow depends on extracting readable SQL; copying it from the Advanced Editor instead introduces M-specific quotation marks and escape sequences that must be cleaned up before it can run in the external tool.
These examples show why the gear-icon workflow matters. It provides access to an editable connection form instead of requiring users to reconstruct or modify the surrounding M expression. Coverage remains partial, and we’re continuing work on reopening settings for native SQL and other extension-based connectors. The parameterized-query example is a distinct scenario, not a claim that every variation is already covered.
Get data needs to become usable quickly
Eight distinct contributors in the first-run feedback review.
For a familiar source, users often already know exactly what they want to connect to. They expect to select the connector and proceed without waiting for the rest of the discovery experience.
One contributor compared their CSV workflow directly: previously, they selected CSV and chose a file; in the new flow, they waited for the interface and then selected Browse before choosing the file. Another reported delay before the connector options were even visible.
The concern is therefore not only total loading time. It is how quickly someone can act on a choice they have already made. A discovery page should help someone exploring an unfamiliar source without slowing down someone returning to a familiar one. Improvements to Get Data startup and loading are expected in the December 2026 Power BI Desktop release.
Advanced Web settings can determine whether a connection is possible
Five distinct contributors in the first-run feedback review.
Advanced connection settings are not necessarily optional refinements. A web service may require a particular request header before it will return data, or a connection may need a URL assembled from several parts. A request header is additional information sent with the request, such as a service’s required authentication header.
One contributor described using an X-API-Key header for a read-only service account. That configuration worked in the classic dialog, but the new dialog provided no way to enter the header. Other feedback distinguished missing custom-header configuration from the missing API-key authentication option. These are related concerns, but restoring a header input is not the same as implementing every authentication method or secure credential-storage scenario.
In the legacy From Web experience, Advanced options exposed URL parts, timeout or page-wait controls, and request-header inputs where applicable. For people relying on these settings, a simpler dialog that omits a required input cannot replace the existing connection workflow.
We’re working on the Web API and Web Page connectors’ advanced configuration to restore the missing settings. The two connectors have different purposes and applicable options, so this is not a promise that every setting belongs in both dialogs.
Figure: Legacy Desktop from Web Advanced settings from Microsoft Learn. This reference shows the kinds of controls discussed in the feedback; applicability differs between Web API and Web Page. It does not show the current preview or a completed fix. Source: Microsoft Learn
Combine Files needs to produce the combined contents people expect
Five distinct contributors in the first-run feedback review.
Folder and SharePoint Folder workflows commonly bring similarly structured files together into one table, such as monthly exports for a reporting period. The file list and the combined data are different results: the list describes the files, while the combined table contains the rows read from inside them.
In the documented legacy Desktop Folder workflow, Combine & Transform Data combines file contents and opens the result in the editor. Transform Data alone opens the file list as-is. That distinction lets users choose whether to combine immediately or filter the list before combining.
One report described receiving file metadata even though the preview showed file contents. Another required a second Combine Files action from the ribbon, while a SharePoint Folder report described output containing the processed files’ headers rather than the expected combined data.
These are different symptoms, but they interrupt the same task: turning a collection of files into a usable dataset. The expected output is the files’ combined contents, not a list that still requires the user to repeat the operation or repair the result manually. We’re continuing to address the reported Combine Files issues.
Figure: Legacy Desktop Folder preview from Microsoft Learn. The rows shown are file metadata, not the contents of the CSV files. The Combine action begins the workflow that reads and combines those contents. This is a reference image, not a screenshot of a reported failure. Source: Microsoft Learn
Beyond these five themes, contributors also raised concerns about authentication choices, catalog and Navigator completeness, recent-source behavior, custom-connector compatibility, localization, and the visibility of Load and Transform actions. We acknowledge those reports as well; appearing less frequently in this discussion does not make their impact unimportant.
What you can expect
The missing native-SQL input at initial connection has been addressed for supported connectors. Our ongoing work includes improving how existing connection settings reopen through the gear icon, restoring missing advanced settings in Web API and Web Page connections, and addressing Combine Files failures.
For startup and loading performance, improvements are expected in the December 2026 Power BI Desktop release. That expectation applies specifically to startup and loading; we are not announcing the same release timing for every issue discussed here.
Connector and workflow coverage can differ while this work progresses. An improvement to initial connection does not automatically resolve an editing problem, and progress on one Combine Files failure does not establish that every reported symptom is fixed.
For current limitations and available workarounds, including when returning to the legacy Desktop experience is appropriate, see New Power Query experience known limitations and workarounds. In Desktop, you can clear the New Power Query experience preview checkbox without restarting. That fallback does not switch Power Query Online in web modeling to a legacy interface.
Share your feedback
Discuss your experience and share feedback in the forum: Share your thoughts on the new Power Query experience in Power BI.
Propose feature ideas or vote on existing suggestions in the Fabric Ideas portal.
Continue the Power Query Corner series
This post is part of our six-part Power Query Corner series on the new Power Query experience. Explore the other posts in the series:
Finding and connecting to data in the new experience (Preview)
Understand how your Power Query query runs (Preview)
Understanding query relationships with Diagram View (Preview)
Review wide tables with Schema View (Preview)
Read the whole recipe in Query Script View (Preview)
Cluster supplier labels and rank results (Preview)
Links will be added as each post is published.
Power BI Updates Blog articles
