Keep only required columns in a Power Query practice table
In Power BI Desktop Power Query Editor, select the columns you want to keep and use Remove Other Columns to discard the rest from the query result. This is useful when a source contains many fields but your report needs only a small, clearly defined subset. It changes the query output, not the original source file.
The exercise uses a fictional workshop attendance table. You will retain Session, Room, and Attendees while excluding a descriptive Note field. The point is to understand an explicit selection of fields and its consequences for later transformations. This is not a privacy guarantee or an instruction to import sensitive data unnecessarily.
Create a disposable example
In a blank Power BI Desktop file, use Enter data to create four columns named Session, Room, Attendees, and Note. Add three rows: Drawing, Cedar, 12, Morning; Mapping, Elm, 9, Afternoon; and Writing, Cedar, 15, Morning. Name the table WorkshopPractice. Open it in Power Query Editor using the transform option available in your version.
Inspect the preview before removing anything. Confirm that all three rows exist and that the column names are spelled as expected. Give Attendees an appropriate whole-number type if necessary. A column-selection operation does not validate whether the remaining counts are correct, so check the sample values independently while the table is still small.
Decide which fields the report actually needs. Suppose the planned visual compares attendance by session and room. Session and Room identify the groups; Attendees supplies the quantity. Note is unnecessary for this particular result. Writing this purpose first prevents arbitrary removal of fields that a later visual or calculation actually requires.
Keep the intended fields
Select Session, Room, and Attendees in the preview. In the desktop editor, you can usually select multiple column headers with Ctrl while clicking. Open the relevant column-removal menu or context menu and choose Remove Other Columns. Check the preview immediately afterward: Note should disappear, while the three selected fields remain.
Do not choose Remove Columns on the selected fields by accident. That command removes the selection, which is the opposite outcome. If the wrong fields disappear, undo or remove the newly added step in Applied Steps before making further changes. A small clean correction is easier to inspect than a chain of compensating transformations.
The expected row values remain Drawing, Cedar, 12; Mapping, Elm, 9; and Writing, Cedar, 15. Count the rows and add the attendance values mentally: the total is 36. Removing a descriptive column should not remove a workshop or alter an attendance number. These checks verify the meaning of the output rather than only its visible headers.
Read the recorded transformation
Look in Applied Steps for the new selection step. Its exact generated name can vary. Click the preceding step and then the new one to compare the table before and after the change. This makes the transformation's scope visible and helps a reviewer understand where the Note field ceased to be available.
A keep-list and a remove-list respond differently to future schema changes. Selecting only named required columns expresses a narrow output. Removing just Note may allow other newly introduced source columns to remain. Choose the behavior that matches your reporting contract, then test it with a disposable source change rather than assuming every schema change will be harmless.
Fields that later steps reference must still exist when those steps run. If a subsequent transformation uses Note, removing it earlier can cause an error. Review dependencies and reorder or redesign the transformations deliberately. Do not suppress an error just to obtain a refresh marked successful while losing the intended report logic.
Check the loaded model
Use Close and Apply when the preview is correct. Inspect the resulting table and any affected visuals in Power BI Desktop. Confirm that the expected fields are present and that an attendance total still evaluates to 36 for the sample. Query preview, loaded data, and report visuals are connected stages, but each deserves a relevant check.
If this is an existing report, save a copy before changing its query shape. Removing a field can affect relationships, measures, sort settings, and visuals that depend on it. The simple practice table has none of those dependencies. A production model can have many, so the ease of the sample operation should not be mistaken for proof that a real model is unaffected.
Power Query is available in several Microsoft products. This walkthrough targets Power BI Desktop; Excel and online experiences may place commands differently. Current Microsoft Learn guidance explains the shared choose/remove concept, but your application's menus and enabled features can differ. Follow your installed interface and inspect the generated step.
Try a controlled change
For a reader exercise, add a fourth workshop with 8 attendees to the disposable source and refresh the query. The expected total becomes 44 while the output still contains only the three required columns. Then consider a proposed new field called Facilitator. Should it automatically become part of this report? Explain your answer before changing the keep-list. That decision defines the intended output more clearly than a long list of fields removed without explanation.














