The symptom
This is a common, easy-to-miss glitch, and the fix is quick and doesn't require becoming a formula expert. You export payroll data as a CSV each period. You have a formula — a SUMIF or SUMIFS — that adds up wages by month. You open the file and the monthly total says zero. Nothing changed. It worked last week.
Zero looks like a quiet month. That is why this failure hides for so long.
What is actually happening
When you open a CSV in Excel, date columns usually look right — they show 18/09/2026 or 18-Sep — but underneath they are often stored as text, not as real dates. Excel's date functions need a real date (a number stored underneath) to do range comparisons. Your formula says "add up anything in September 2026" and gets nothing, because it cannot compare a text string to a date range.
The giveaway: real dates line up to the right side of the cell, like numbers. Dates stored as text line up to the left, like words. If your date column is left-aligned after opening the CSV, it is text.
Three ways to fix it
Option 1 — paste-special multiply (a common workaround; check it on your version): In any empty cell, type the number 1 and copy it. Select your entire date column. Right-click → Paste Special → choose Multiply. This converts text-dates to real dates in place. Your formula should work immediately after.
Option 2 — Text to Columns: Select the date column. Go to Data → Text to Columns. Click Next twice without changing anything. On the last step, choose Date as the column data format and pick the order that matches your file (check a known date in the export to be sure which order it uses). Click Finish. Nothing moves; the dates just become real.
Option 3 — wrap the formula: If you cannot change the source data, you can wrap your date comparison in DATEVALUE. Instead of comparing to "1/09/2026" directly, write DATEVALUE("1/09/2026"). This converts the text in the formula rather than in the column. It is the least reliable option — it breaks if the text format in the export ever changes — but it is useful when someone else owns the sheet and you cannot touch the raw column.
Options 1 or 2 are better. Fix the data once so the formula never has to guess.
The part that matters most — making failure obvious
Any of these fixes will work until the CSV format changes, a new payroll provider formats dates differently, or someone runs the file on a machine with different regional settings. The zero comes back and nobody notices until month-end.
Add one check cell and give it a conditional format that turns it red when it is not zero:
=SUMPRODUCT(--ISTEXT(A2:A5000))
Replace A2:A5000 with your actual date column range.
When this cell shows zero, every date in the column is a real date and your monthly totals are trustworthy. When it shows anything else, the dates have come in as text again — the cell turns red and whoever opens the file knows to stop before trusting any number.
Put this cell at the top of the sheet, not buried on a working tab. The rule is simple: red means fix before you read the totals.
The keep / clean / connect / convert read
This is a clean problem. The data is fine — the right amounts and dates are in the file. They just need one small step of cleaning before formulas can use them.
You do not need to buy anything, change payroll software, or rebuild the report from scratch. The fix takes about two minutes. The check cell takes two more and means you will not spend an afternoon looking for a quiet month that was never quiet.