Excel Power Query Takes Too Long to Refresh? Fixes

How-to basics · Updated November 19, 2026

When Excel Power Query takes too long to refresh, the cause is usually a slow source, a transformation that processes more data than necessary, or the time Excel needs to load results into a sheet or the Data Model. Find which stage is slow first, then reduce the work at that stage; changing refresh settings alone rarely fixes an inefficient query.

Find out where the refresh is slowing down

A long refresh does not automatically mean Power Query itself is at fault. A query may wait for a database, download files over a network, perform transformations on your computer, or load a large result into Excel. Refreshing several dependent queries can add all of these stages together.

Start with a controlled comparison. Save the workbook, note how long Refresh All takes, and then refresh the slow query on its own from Data > Queries & Connections. If one query dominates, investigate its source and steps. If each query is quick alone but Refresh All is slow, look for shared sources, dependent queries, or several large loads running together.

  • Open Data > Queries & Connections and check which query or connection remains active longest.
  • In Power Query Editor, review the Applied Steps list. Note where previews begin to slow down or errors appear.
  • Compare the refresh time when connected to the company network or VPN with the time on a reliable local connection, if your source permits that comparison.
  • Separate the time spent retrieving data from the time spent loading it. A connection-only query can help you assess transformation and source work without putting the full result on a worksheet.

Power Query previews are not a perfect stopwatch for a full refresh: previews may use a limited sample or cached results, while loading can require processing the complete result. A preview that appears quickly therefore does not prove the final query will refresh quickly.

Check the source before changing the query

The source often sets the lower limit on refresh time. A local workbook or text file may be quick, while a database, SharePoint folder, web endpoint, or shared network drive can involve authentication, server processing, file enumeration, and data transfer. A slow source can make even a simple query feel stuck.

Check whether the source itself responds slowly outside the query, and whether its size or location recently changed. A folder query that used to read a few files may now be scanning thousands. A database report may have become slower because the underlying table grew or the server is busy. For a network file, copying a permitted test file to a local folder can help distinguish network delay from transformation time; do not replace the production source unless the workbook is meant to use that local copy.

For databases, ask the data owner whether the required fields and filters can be supplied by a view or source-side query. Filtering at the server can reduce both the amount transferred and the work Excel must do. For files, exclude unrelated files and temporary files early. For web sources, check whether the service imposes rate limits or returns data in pages; repeatedly requesting many small endpoints may take longer than one supported bulk request.

Credentials, privacy checks, and connection prompts can also delay a refresh or prevent it from completing. Confirm that the sign-in method is still valid and that each source is reachable. Do not disable privacy protection just to make a refresh proceed: privacy levels help govern how data from different sources may be combined.

Reduce rows and columns as early as possible

Power Query evaluates a sequence of steps. If the query reads a large table and only discards most of it near the end, unnecessary data may be retrieved, transferred, or processed. Remove columns you do not need and filter out irrelevant rows near the beginning of the query, as long as the filter preserves the intended result.

For example, a sales query used to report the current year may not need ten years of records. Filtering to the required date range and keeping only fields such as date, product, region, and sales amount can reduce memory use and speed later grouping or merging. The right filter depends on the reporting requirement: excluding records merely to make refresh faster can silently make a report wrong.

This is where query folding matters. With a compatible source, Power Query can translate supported steps into a request that the source system executes. A database may then return only the filtered rows and selected columns instead of sending the full table to Excel. Folding depends on the connector, source, and transformations; it is not available for every source, and not every step can be translated.

In Power Query Editor, right-click a step and look for View Native Query when the connector supports it. Its availability is a useful clue, not a universal test: some connectors do not show a native query even when they optimize operations. If a step breaks folding, later work may run locally and require more data to be downloaded. For a database query, ask a database administrator before changing native SQL or source objects.

Review expensive transformations and merges

Some transformations naturally require more work. Sorting a large table, grouping many rows, removing duplicates, expanding nested records, and merging large tables can consume substantial time and memory. These operations are not inherently wrong; the goal is to avoid applying them to data that could have been reduced first.

  • Filter rows and remove unused columns before a merge, group, or sort where doing so preserves the required output.
  • For a merge, check that the matching columns have compatible data types and that both tables contain only necessary fields. Unexpected duplicate keys can multiply rows and inflate the result.
  • Inspect custom columns and repeated steps for work that can be performed once or at the source. Avoid adding complicated per-row calculations before filtering if those calculations are not needed for excluded rows.
  • Use referenced or connection-only staging queries when several outputs need the same cleaned source, rather than maintaining separate copies of the same transformation logic.

Do not assume that every step should be moved or deleted to improve speed. Changing step order can change results, especially when filters, sorting, grouping, and row numbering interact. After simplifying a query, compare row counts, totals, and key fields against a known correct result.

Determine whether loading is the bottleneck

A query can finish retrieving and transforming data but still take time to write results. Loading hundreds of thousands of rows to a worksheet involves more than displaying a small preview; Excel must create or update the table and may also refresh formulas, PivotTables, charts, or other dependent content. Loading the same result to the Data Model has different memory and processing costs.

In Excel, open Data > Queries & Connections, right-click the query, and choose Load To when that option is available. Choose a worksheet table only when users need the rows directly on a sheet. If the query is an intermediate step, Connection Only can avoid creating an unnecessary table. If it feeds a PivotTable or a model-based report, loading to the Data Model may be more appropriate, but it is not automatically faster for every workbook.

Check whether the output is larger than the report needs. A worksheet has a finite row limit, and a query intended to return more rows than a sheet can hold should not be loaded as an ordinary worksheet table. Summarizing at the source or in Power Query may be more useful than placing a massive detail table in the workbook. A large model can also increase workbook memory use and affect other users who open the file.

Use refresh settings carefully

For a query connection, right-click it in Data > Queries & Connections and choose Properties to review options such as refreshing when the file opens, refreshing at an interval, and background refresh where available. These settings control when or how refresh runs; they do not make the underlying source or transformation faster.

Background refresh can let Excel remain usable while a connection refreshes, but it may make it harder to tell when dependent work is complete. If a later action relies on refreshed data, wait for refresh completion before checking the result. Turning background refresh off can make Excel wait for the operation to finish; that changes the user experience, not the amount of work.

If multiple tables refresh at once, parallel loading may reduce elapsed time when the computer and source can handle the extra activity. It can also increase memory use or burden a busy server. Power Query options and wording vary by Excel build, so check File > Options > Data for available query and loading settings rather than assuming a particular option exists. Change one setting at a time and compare results.

Account for Excel, memory, and workbook size

Large refreshes need enough available memory for the source data, intermediate steps, and loaded results. Other open workbooks, browser tabs, and applications can make a refresh slower or unstable. Closing unneeded programs and restarting Excel can be a useful diagnostic, but persistent delays usually call for reducing the data or investigating the source rather than routinely restarting.

Excel’s 32-bit and 64-bit editions have different memory constraints. If a workbook regularly processes large datasets and Excel reports memory problems, check which edition is installed under File > Account > About Excel. Moving to 64-bit Excel may help with memory capacity, subject to organizational compatibility and installation rules; it does not fix a slow database, excessive rows, or inefficient transformations.

Workbook calculations can also make the overall wait appear longer. A refresh may trigger dependent formulas, PivotTables, or charts to update. To distinguish this from query time, observe the query status in Queries & Connections and note whether Excel continues calculating after the connection finishes. Avoid turning off calculation without understanding the effect on the workbook, since displayed formula results may then be out of date.

A practical order for fixing a slow refresh

Make changes in a sequence that preserves a reliable comparison. First record the current refresh time and verify the output. Next isolate the slow query, confirm the source is responsive, and reduce unnecessary rows and columns. Then inspect merges and other expensive transformations, and finally reconsider where results are loaded and which refresh options are needed.

  1. Save a copy of the workbook and record the expected row count or a few validation totals.
  2. Refresh one query at a time to identify the slowest part of the workbook.
  3. Check source access, network conditions, credentials, and whether the source has grown.
  4. Filter and select only needed data early; check folding where the connector supports it.
  5. Review merges, grouping, sorting, and custom calculations, then validate the result.
  6. Remove unnecessary worksheet loads and compare refresh time after each change.

Query Diagnostics, when available in your Excel build, can provide more detail about evaluations and source activity. In Power Query Editor, look under the Tools tab for Query Diagnostics commands. Use the output to investigate rather than treating one duration as a universal verdict: diagnostics themselves add work, and results depend on the connector and the query.

Frequently Asked Questions

Why does Power Query refresh take longer than the preview?
The preview may show only a sample or use cached data. A full refresh can retrieve and transform all qualifying rows, load the result, and update dependent workbook content.

Does clicking Refresh All make Power Query slower?
Refresh All can take longer because it refreshes multiple connections and dependent queries, not just the query you are looking at. Refresh queries individually to find the one that dominates the total time.

Will disabling background refresh speed up a query?
Usually it changes whether Excel waits for the refresh to finish, not how much work the query performs. It can make completion easier to track, but it is not a general performance fix.

Why is my Power Query fast on one computer but slow on another?
Differences in network access, VPN, source permissions, available memory, Excel edition, and other running applications can all affect elapsed time. Compare the same workbook, source, and refresh conditions before drawing a conclusion.

Should I turn off privacy checks to speed up refresh?
Not as a routine fix. Privacy settings protect how Power Query combines data from sources with different privacy levels. Review the source permissions and privacy configuration with your organization instead of removing safeguards without understanding the consequences.

Related

Try it: the Date Difference Calculator: days, months, years and weekdays between two dates, with the DATEDIF formulas.

More in How-to basics

Sources: Function behaviour and menu paths: Microsoft Support documentation for Excel. Method & sources

Spotted an error in this guide? Let us know.