Power BI Legacy CSV/Excel Import Deprecated (2026): What to Do

What's changing

The legacy "Import Excel or CSV" workflow in Power BI Service is being phased out in favor of the current Get Data / Power Query-based Text/CSV import. Existing semantic models created with the old workflow keep running for a transition window, but stop being creatable, and Microsoft's guidance points toward migrating reports off the legacy path before it's fully removed. Check Power BI's official blog and release notes for the current exact dates, Microsoft has moved this timeline before and this page won't chase every revision.

What to do instead

1. In Power BI Desktop or Service, use Get Data → Text/CSV (not the legacy "Import Excel or CSV" entry point).
2. Set the source encoding explicitly, don't rely on auto-detect, in the import dialog's File Origin dropdown, if your CSV contains any accented or non-ASCII characters.
3. Review the Power Query editor's Changed Type step before loading, this is where the most common new errors show up (see below).

The most common new error: wrong or inconsistent column types

Power Query's automatic type detection samples the first rows of the file, not the whole dataset, by default. If a numeric column has integer-looking values at the top but decimals or larger numbers further down, or a date column has one inconsistent format buried later in the file, the auto-detected type can be wrong for the column as a whole, and later rows fail to convert or come through as null.

Fix:

• In Power Query, right-click the column header → Change Type → set it explicitly instead of trusting the auto-detected step.
• In Power BI Desktop, under File → Options and settings → Options → Current File → Data Load, check whether "Detect column types and headers for unstructured sources" is scanning enough of the file, or clean the source before import so the first rows are representative.
• Cleaning the CSV so every value in a column is consistently typed before it reaches Power Query removes the guesswork entirely.

How to check whether a specific report or dataset is affected

1. In Power BI Service, open the semantic model's Settings page and check the Gateway connection / Data source credentials section, the old import path is usually labeled distinctly from a Get Data / Power Query-based source there.
2. If the dataset was created by uploading a file directly (rather than via Get Data → Text/CSV) more than a year or two ago, it's a reasonable candidate for having used the legacy path.
3. When in doubt, the safest move is to recreate the report from the same source file using the current Get Data workflow and compare the output before retiring the old one, rather than assuming either way.

Other Power BI CSV import errors that aren't the deprecation

Not every CSV import failure after this change is related to the legacy-path retirement itself. A few other errors get mistaken for it:

• "We couldn't find any data in the specified range" usually means the source file has a title row or merged cells before the actual header row, Power Query's default range detection expects the header on the first row.
• Numbers importing as text because of a locale mismatch (e.g. a file using 1.234,56 European decimal notation read with US locale settings) is a separate setting under the Power Query import dialog's Locale option, not a type-detection sampling issue.
• A refresh that worked yesterday fails today with no file changes often points to a changed or expired data source credential rather than anything about the CSV's structure.

Doing this in How To CSV

Run the export through Column Validator first to catch mixed types and inconsistent values per column before they ever reach Power Query's type detection, or convert to XLSX first if you need Excel's own stricter type formatting to carry through the import. If the file itself might be malformed rather than just inconsistently typed, paste it into CSV Lint for a line-by-line syntax check before troubleshooting further inside Power Query.