If Excel Power Query is taking a long time to refresh, first identify whether the delay is at the data source, in the query steps, or while loading results into Excel; then fix the slow stage rather than changing settings at random. In Microsoft 365 for Windows, check the query’s source, Applied Steps, and load destination, and test a refresh after each targeted change.
Why Power Query refreshes can take a long time
A refresh is more than a recalculation. Power Query connects to a source, requests data, applies the query’s transformation steps, and then loads the result to a worksheet, the Data Model, or a connection used by another query. A delay at any one of those stages can make the whole refresh seem slow.
The source may be a large workbook, a database, a folder of files, a web service, or a cloud location. Network speed, server workload, authentication, and the amount of data returned all matter. On the Excel side, complex transformations and loading millions of cells can also take time. If several queries depend on one another, refreshing them may involve a chain of work rather than one isolated operation.
It is useful to distinguish a slow refresh from a slow preview. The Power Query Editor displays a preview while you work, and that preview is not necessarily the same as loading the entire final result to Excel. A preview that appears quickly does not guarantee that a full refresh will be quick.
Find which part of the refresh is slow
Start with a simple comparison: refresh one query rather than using Refresh All. In Excel, open Data > Queries & Connections, right-click the query, and choose Refresh. If one query is consistently slow, investigate its source and steps. If Refresh All is much slower than refreshing that query alone, other queries or their dependencies may account for the extra time.
Then compare the refresh in the Power Query Editor with the time it takes to load results back into Excel. A delay before rows appear in the editor points toward the source or transformations. A quick preview followed by a long wait during the workbook refresh can point toward a large output, a Data Model load, dependent queries, or the workbook’s available memory and resources.
- Note the query name, source, approximate row count, and destination.
- Refresh one query at a time and record which one takes longest.
- In the editor, review the Applied Steps list and check where the preview begins to slow down.
- Try a small, representative date range or a few files to see whether the delay scales with input size.
- Check whether a source is on a network, VPN, cloud service, or shared drive that may be responding slowly.
For deeper analysis, Power Query diagnostics may be available from the Power Query Editor’s Tools menu. Use the diagnostic output to investigate operations and source requests, but treat it as technical evidence to interpret—not as a single universal timer that identifies every cause.
Reduce the amount of data retrieved
One of the most effective improvements is to avoid bringing in columns and rows that the final report does not need. If the source contains years of transactions but the report only uses the current year, filter to the required period early. Remove unused columns near the beginning of the query, and avoid importing extra sheets or files from a folder.
For example, a folder-combine query that processes every workbook in an archive will take longer as that archive grows. If older files are not needed, filter the file list before combining contents. Check the folder for temporary files and unrelated workbooks too; a broad folder query may try to process files that were never intended as inputs.
When using a database or another source that supports query folding, Power Query may be able to send filters and other supported operations to that source. This can reduce the data transferred to Excel. Folding depends on the connector, source, and sequence of steps. In the editor, right-click a step and look for View Native Query when available; a missing or disabled option does not by itself prove that the entire query is slow or broken.
Do not assume that every source can filter data before it reaches the computer. Some file-based and web-based sources have different capabilities, and some transformations cannot be pushed back to the source. If a filter does not fold, reducing the input files or choosing a more suitable source-side filter may still help.
Review expensive or unnecessary query steps
Power Query records transformations as Applied Steps, so review the sequence rather than deleting steps blindly. Sorting a large table, grouping rows, merging tables, expanding nested records, or invoking a custom function for every row can be demanding, especially when the input is large. The exact cost depends on the source and data shape.
Place filters and column removal early when doing so is valid for the intended result. Avoid sorting if the output does not require a particular order: sorting consumes work and does not guarantee that later operations will preserve that order. If the query merges two tables, check that the join columns have matching data types and that the lookup table is not unnecessarily large. A type mismatch can produce incorrect matches as well as extra troubleshooting.
Also look for duplicated work. Several independently built queries may each connect to and transform the same source. Referencing a staging query can make a workbook easier to maintain, but it does not guarantee that the underlying data is fetched only once or cached for every dependent query. Refresh behavior depends on the query structure, connector, and load settings, so test the workbook rather than assuming that a reference eliminates all repeated processing.
Be cautious about adding a buffering step as a generic speed fix. Buffering can use memory and may prevent some operations from folding to the source. It is useful only in particular query designs, not a universal solution to slow refreshes.
Check query dependencies and load destinations
In Data > Queries & Connections, inspect the query list and identify which queries load to worksheets and which serve as intermediate sources. A query used by another query may still need to be evaluated as part of a refresh. Refresh All can therefore take longer than the time for a single visible table.
For intermediate queries that do not need to appear in a worksheet, consider changing the load destination. In Power Query, use Home > Close & Load > Close & Load To to choose an appropriate destination, such as creating a connection or loading to the Data Model where suitable. A connection-only query avoids placing that query’s result in worksheet cells, but it does not automatically remove the processing needed by downstream queries.
Worksheet output has a hard limit of 1,048,576 rows per sheet. Even below that limit, a very large table can take time and memory to load, and it can make the workbook cumbersome to open and use. If the report does not need every row displayed, filter the output or use a suitable summary. Loading to the Data Model can be more practical for some analyses, but it still requires resources and is not guaranteed to make refresh faster.
Investigate credentials, privacy, and connection settings
A refresh can pause or fail because a source needs updated credentials, a network connection is unavailable, or a privacy prompt requires a decision. In Excel, check Data > Get Data > Data Source Settings to review saved permissions and credentials for the relevant source. The exact options depend on the connector and sign-in method. If access recently changed, confirm that the account still has permission to read the source.
When a query combines sources, Excel’s privacy settings can affect how Power Query evaluates the data. Set privacy levels to reflect the real sensitivity of each source. Do not lower or ignore privacy protections simply to try to make a refresh faster: that can change how data from different sources is handled and is not a safe general performance fix.
Refresh properties may also affect when work happens. In Data > Queries & Connections, right-click a query and choose Properties to review options available for that connection, such as refreshing when the workbook opens or refreshing in the background where supported. Background refresh can let you continue using Excel while a refresh runs, but it does not make the underlying query finish sooner. If the workbook is being edited during a refresh, wait for completion before relying on the displayed results.
Practical fixes to try in order
Make one change at a time and compare refresh times under similar conditions. This makes it easier to tell whether an adjustment helped and reduces the risk of changing the query’s meaning.
- Refresh the slow query by itself and identify its source and load destination.
- Remove unneeded columns and restrict rows or files as early as the required result allows.
- Review Applied Steps for large sorts, merges, grouping, custom functions, and duplicated transformations.
- Check that source files, database access, VPN or network connections, and credentials are available and responsive.
- Load intermediate results as connections where appropriate, and avoid worksheet output that the workbook does not need.
- Close unnecessary applications and check whether Excel is short on available memory or system resources, especially with large workbooks.
- Save a separate copy before making substantial query changes, then compare both the refresh time and the output for correctness.
Excel for the web and Excel for Mac do not have identical Power Query capabilities, connectors, or menus to Excel for Microsoft 365 on Windows. If a query was built on Windows, some editing or source options may not be available in another environment; check the capabilities of the specific Excel build and connector before treating a missing option as a refresh fault.
Common misconceptions about slow refreshes
A larger refresh interval or background-refresh setting does not optimize a query. These options change when a refresh occurs or whether Excel waits in the foreground; they do not reduce the amount of source data or transformation work. Similarly, closing and reopening Excel may clear a transient issue, but it is not a durable fix if the same source or step is consistently slow.
Another misconception is that a query must be broken if its preview takes a while. Large or complex sources can take time to preview, and the editor’s preview does not necessarily represent the time needed for a full load. Conversely, a fast preview does not establish that the final worksheet or Data Model load will be fast.
Finally, “Refresh All” is not always the best test. It may refresh multiple connections and dependent results, so its total time cannot be attributed to one query without isolating the parts. Diagnose first, then choose the optimization that preserves the intended data and output.
Frequently Asked Questions
Why is Excel Power Query stuck on Evaluating?
It may be waiting for the source, processing a large input, running a demanding transformation, or resolving a connection or privacy check. Refresh the query alone, check source access, and inspect the Applied Steps to narrow down the stage causing the delay.
Does Refresh All refresh every Power Query?
Refresh All can refresh workbook connections, including Power Query connections, and may involve queries that feed other queries. The exact work depends on the workbook’s connections and dependencies. Refresh one query from Queries & Connections to compare.
Will loading a query to the Data Model make it refresh faster?
Not necessarily. The Data Model can be a suitable destination for large analytical tables, but Power Query still has to retrieve and transform the data, and the model itself uses system resources. Choose the destination based on how the data will be analyzed, then measure the result.
Can I close Excel while a query is refreshing?
Do not assume the refresh will finish safely after Excel closes. Wait for the refresh to complete and confirm the workbook is saved before closing, especially when the result is important or the source is remote.
Why is a query slower on one computer than another?
Differences in network access, source permissions, available memory, processor resources, Excel build, and connector support can affect refresh time. Compare the same query, source, and input size, and check whether one computer is using a VPN or a different sign-in account.
Related
- Excel Number Format K: Show Values in Thousands
- Excel Number Format: Keep Leading Zeros in Your Cells
- Excel Number Format: Hide Zeros Without Deleting Them
Try it: the Formula Explainer: paste any formula and see each function and reference in plain English.
More in Errors & fixes
- Google Sheets Apps Script Not Working? Fixes to TryIf Apps Script is not working in Google Sheets, first check whether the script is attached to the…
- Excel Error Message When Opening a File: Causes and FixesAn Excel error message when opening a file usually means Excel cannot read the file as presented, the…
- Excel Lost Conditional Formatting? Find and Fix ItExcel conditional formatting usually has not disappeared completely: the rule may apply to a different range, be hidden…
- Excel: Remove #N/A Errors From VLOOKUP Results SafelyTo replace a VLOOKUP #N/A error with a readable message, wrap the lookup in IFNA: =IFNA(VLOOKUP(E2,$A$2:$C$100,3,FALSE),"Not found"). This…
- Excel Page Break Not Going Away? Fix Persistent LinesIf an Excel page break will not go away, first switch to View > Normal; if the line…
- Excel RATE Function Not Working? Fix Common ErrorsIf Excel’s RATE function is returning an error or an unexpected number, check that the cash-flow signs are…
Sources: Function behaviour and menu paths: Microsoft Support documentation for Excel. Method & sources
Spotted an error in this guide? Let us know.