Spreadsheet Risk
Excel File Too Large and Slow? What Actually Causes It
Slow workbooks are rarely slow because of row count. The usual causes are volatile formulas recalculating constantly, formulas referencing whole columns, formatting applied to a million empty cells, and array formulas over huge ranges. Most files can be made fast again in an afternoon — but if the file is slow because several people are working in it, speed is not the real problem.
7 min readUpdated
When the file is slow because it became a system of record, Datanova helps businesses move it into a system built for that workload.
The real causes, in the order worth checking
| Cause | What it looks like | Fix |
|---|---|---|
| Volatile functions | The whole workbook recalculates on every edit | Replace NOW, TODAY, INDIRECT, OFFSET, RAND with static or non-volatile equivalents |
| Whole-column references | SUMIF(A:A,…) across dozens of formulas | Reference the used range, or convert to a structured table |
| Formatting bloat | The file is 40 MB with 5,000 rows of data; Ctrl+End jumps to row 900,000 | Delete the unused rows and columns entirely, then save and reopen |
| Array formulas over huge ranges | Noticeable freeze on any edit | Narrow the ranges, or move the aggregation to a pivot |
| Chained cross-file links | Long pauses on open; prompts about updating links | Consolidate the sources, or import rather than link |
| Conditional formatting rules | Hundreds of overlapping rules, often duplicated by copy-paste | Consolidate the rules and apply them to bounded ranges |
Check formatting bloat first. It is the most common cause of a file whose size makes no sense relative to its data, and it is the easiest to fix.
The quick diagnostic
- Press Ctrl+End. If the cursor lands far below or to the right of your actual data, you have formatting bloat — delete those rows and columns, save, close and reopen.
- Compare the file size to the data. A workbook with a few thousand rows should not be tens of megabytes.
- Search the formulas for NOW, TODAY, INDIRECT, OFFSET and RAND. Every one of those recalculates on every change anywhere in the workbook.
- Search for whole-column references (A:A, B:B). Each one asks Excel to consider a million cells.
- Set calculation to manual temporarily. If the file becomes responsive, the problem is formulas, not size.
When Excel is genuinely at its limit
A worksheet stops at 1,048,576 rows and 16,384 columns, and in practice things get uncomfortable well before that. If your data genuinely is at that scale, the answer is usually not a faster spreadsheet:
- Power Query for the loading and shaping, so the raw data never sits in the grid
- The Data Model and Power Pivot, which handle far more rows than a worksheet
- A database with a BI tool on top, when the data keeps growing and several people query it
These keep Excel as the interface while moving the volume somewhere built for it, which is often the cheapest good answer for an analysis workload.
When slow is a symptom, not the problem
Fixing performance is worth doing and frequently enough. But sometimes the slowness is what a different problem feels like from the inside:
- The file is slow because five people have it open — the problem is concurrency, not calculation
- It grew huge because it accumulated every month's data as a new tab — the problem is that it became a database
- It is slow because it holds transactions rather than analysis — the problem is that it became a system of record
- It crashes at month end because that is when everyone uses it at once — the problem is scheduling around a single file
In each of those, an afternoon of optimization buys a few months of relief and the same conversation comes back. Worth knowing before you spend the afternoon.
Frequently asked questions
How many rows can Excel handle?
Why is my Excel file so large with so little data?
Does converting to a table make Excel faster?
Should we split the file into several smaller ones?
Optimizing the same file for the third time?
In a free consultation we look at why the workbook is struggling and tell you whether it is a formula problem worth an afternoon, or a structural one that optimization will not fix.