Why PDF to Excel loses formatting, and what 'numbers as numbers' means
When people say a PDF lost its formatting in Excel, they usually mean one of two very different things. One is how it looks: fonts, colours, borders. The other is what each cell means: whether 1,240 is a number you can add up, or just three characters that happen to look like one. The second matters far more, and it is where most of the damage happens.
Looks versus meaning
Excel decides what a cell is the moment a value lands in it. A number is something you can sum, sort and chart. Text is just text. A date is a number with a date format on top. If a value arrives as text, nothing looks wrong, but a column total quietly leaves it out.
Appearance mostly does not carry over, and that is normal. In a test on this site, a sample PDF table with a bold "Total" row produced a workbook where that row was not bold. The workbook gives its own table a bold header with a light fill and a frozen first row, rather than copying the PDF's styling. Expect the workbook to hold the table's content, not the page's design.
Meaning is what you should protect. The rest of this guide covers the four ways it gets lost.
1. Numbers that arrive as text
PDFs print numbers for people to read, with thousands separators, currency signs, brackets for negatives and percent signs. Excel needs those turned into real numbers. We ran the rules this site uses over a list of example values:
| As written in the PDF | What Excel gets |
|---|---|
1,240 |
The number 1240, shown with a thousands separator |
48,213.7 |
The number 48213.7, with one decimal |
$18,450.00 |
The number 18450, shown as a dollar amount |
€1,200 |
The number 1200, shown as a euro amount |
(1,250.00) |
The number -1250, shown in brackets |
12.5% |
The number 0.125, shown as 12.5% |
-3.2% |
The number -0.032, shown as -3.2% |
A quick test in Excel: select a column and look at the status bar. If it shows a sum, the cells are numbers. If it only shows a count, they are text.
2. Dates that flip
The date 03/04/2025 is 3 April in much of the world and 4 March in the United States. Nothing in the PDF says which. If a converter picks one, your spreadsheet looks tidy and may be wrong for half the rows, and you will not notice.
This site does not guess. Dates are converted only when they are unambiguous: year-first dates like 2025-03-31 (the YYYY-MM-DD form that W3C's note on date and time formats describes), and dates with the month written out, like 15 July 2025 or July 15, 2025. Anything like 03/04/2025 stays exactly as written. If you know the document's convention, converting those cells yourself takes a minute, and you will know the result is right.
3. Codes that lose their zeros
An account code such as 00417 is not a number, even though it is made of digits. Treat it as one and Excel shows 417. The same happens with postal codes and product IDs. Here, a value that starts with a zero stays text, and so does a very long run of digits (16 or more), because those are identifiers rather than quantities. In our test with the live tool, the code 00417 came out unchanged, and the rule applies to values such as 007 in the same way.
4. Tables that come apart
Merged cells and rows that wrap onto a second line are the classic ways a table breaks. Excel's own PDF import documents this: Microsoft notes that multi-line rows might not be identified properly and can need cleanup afterwards (Power Query PDF connector). Any converter has the same difficulty. After converting a table that has merged cells, check that each row still carries its own region, category or label, because a blank cell in a sorted or filtered column can hide which group a row belongs to.
The limit of "only when sure"
Being careful has a cost. Some real numbers stay as text because we would rather leave them alone than misread them. In our test, 1.234,56 (European style, with a dot for thousands and a comma for decimals) and 12 345 (a space as the thousands separator) both stayed text, and so did a plain label like N/A. Text cells are easy to spot and fix. A silently wrong number is not.
What to do after converting
- Sum a column and compare it with the printed total.
- Look for cells left as text where you expected numbers, and convert them once you know the document's number style.
- Check dates left as text against what you know about the source.
Our step-by-step example of copying a PDF table to Excel walks through all of this with a real sample, including what to check afterwards.
What happens to your file
Table extraction uploads your PDF to our servers, where open-source software processes it. It is not sent to any third-party AI or document API. The uploaded PDF is deleted as soon as processing finishes, and the results stay for 60 minutes so you can download them. If a file must not leave your device, do not upload it. See the privacy page for the details.