The converted spreadsheet looks right and every total says zero

You convert a statement, open the spreadsheet, and it looks perfect. Then you select the amounts column, type a SUM, and get zero. Nothing is broken and nothing is corrupted — the figures were handed over as text that looks like numbers, and a spreadsheet will not add up text no matter how numeric it appears.

What actually went wrong

A spreadsheet cell holds either a number or a piece of text, and it treats them completely differently. 1204.55 is a value it can add, average, sort and chart. "1204.55" is a short string that happens to contain digits — it can be displayed and searched, and that is about all.

A PDF does not store either one. It stores the marks on the page: the characters $, 1, ,, 2, 0, 4, ., 5, 5, positioned in a row. Nothing in the file says "this region is a currency amount." Any converter has to decide what that run of characters means, and the safe, lazy choice is to call all of it text.

That is why the result looks flawless. Every character is in the right cell. The only thing missing is the part you cannot see.

How to tell in about five seconds

1. Click a single amount and look at how it sits in the cell. Spreadsheets right-align numbers and left-align text by default. A column of amounts hugging the left edge is text.
2. Select the column and read the status bar. Excel shows Sum and Average for numeric selections. If it shows only Count, there are no numbers in there.
3. Type a SUM below the column. Zero is the answer you get from a column of text.
4. Look for the little green triangle in the corner of the cell — Excel's own "number stored as text" warning. Helpful when it appears, but it does not always.

Why fixing it by hand is worse than it sounds

The usual advice is to select the column and convert to number. On a tidy export that works. On a real converted statement it tends to stall on the things that made the document worth converting in the first place:

• Currency symbols. $1,204.55 will not convert while the $ is still in the cell, so the symbols have to come out first — and then the formatting you wanted is gone too.
• Thousands separators. Sometimes handled, sometimes not, depending on the locale the spreadsheet thinks it is in.
• Accounting negatives. (84.20) means minus 84.20 on a statement. Converted naively it becomes text, or worse, positive 84.20 — a sign error inside a total, which is the kind of mistake that does not look like a mistake.
• Trailing spaces and non-breaking spaces. Invisible, and enough on their own to keep a cell as text after every other fix.
• Cells that genuinely are text. Reference numbers, dates, account codes. Force-convert the whole column and you will damage these while repairing the others.

So the honest version of the manual fix is: several passes, a find-and-replace you have to reason about, and a real chance of quietly changing a figure. On one statement that is an annoying afternoon. On twelve months of them it is a mistake waiting to happen.

What a conversion should hand you instead

The distinction worth insisting on is between what a cell contains and how it is displayed. Those are separate things, and a converter that understands the difference can give you both at once.

$4.50 in the PDF should arrive as the number 4.5, carrying a currency format that displays it as $4.50. It still reads exactly like the original document, and SUM works the moment you type it. Nothing was traded away.

PDF to Excel does that: amounts come through as figures with their original formatting attached, accounting parentheses become real negatives, and percentages become actual percentages rather than the characters 1, 2, %.

The things that should stay text — on purpose

Converting everything possible into a number is its own failure, and a quieter one. Some columns must stay as text or the data is damaged.

• Dates. Deliberately left alone. 03/04 is the 3rd of April in Calgary and the 4th of March in much of the world, and a PDF carries nothing that settles which. A converter that guesses is right most of the time and silently wrong the rest — so the text is preserved and you convert it yourself, knowing which convention your document uses.
• Reference numbers. #4412 is an identifier, not a quantity. As a number it loses its leading zeros and stops matching your records.
• Anything with a letter in it. N/A, 12 units, Ref 88 — part number, part label, and not safely either.

How to convert a PDF to Excel

1. Open PDF to Excel and add your document.
2. Choose Find tables. Each table found is listed with its page and size, and you can preview it before committing.
3. Untick anything you do not want — a stray header row, a table from a section that is not relevant.
4. Download as .xlsx for one sheet per table, or .csv for a single flat file.
5. Run the SUM check on one amount column. Five seconds, and it tells you whether the rest is trustworthy.

When nothing is found — and why that is the right answer

Sometimes a converter reports no tables. That reads like a failure and usually is not one: a page of ordinary paragraphs has no table on it, and the alternative is worse than nothing.

Tables without ruling lines have to be detected from the spacing between words, and that is genuinely hard, because a paragraph is also words separated by spaces. Push a detector to find tables everywhere and it will find them in your prose — returning a sheet of sentences chopped into columns at every space. That output is not merely useless, it is confidently useless, and it takes a while to notice it means nothing.

So being told plainly that a page had no table is better information than a spreadsheet full of shredded text. If you can see a table that was missed — one held together by alignment alone, with no lines anywhere — there is a Look harder for tables without lines option that lowers the bar, at the cost of some plain pages coming through as rows that mean nothing.

If the PDF is a scan

A scanned statement or photographed receipt holds a picture of a table, not a table. There are no characters in the file, so there is nothing to put in a cell and a converter will correctly find nothing.

Run OCR first to create a real text layer, then convert that. The chain is scan, OCR, spreadsheet — and check the figures more carefully than usual afterwards, because OCR reads text well but isolated numbers are the least reliable thing on any scanned page. A misread 3 for an 8 in a column you are about to total is exactly the error that survives to the bottom line.

Multi-page tables

A table running across several pages comes back as one sheet per page rather than a single stitched sheet. That is deliberate: joining them requires assuming the columns are identical on every page, and repeated headers or a subtotal row part-way through break that assumption quietly. Separate sheets let you see what came from where and paste them together yourself once you have looked.

Keep the PDF

A converted spreadsheet is an interpretation of the document, and for anything with financial weight the PDF stays the record while the spreadsheet is a working copy. Nothing here is kept once your download is ready, so the only copies that exist are the ones on your own machine.

If you need the whole document editable rather than just its figures, PDF to Word is the other direction to take it.

FAQ

Why does SUM return 0 on my converted spreadsheet?

Because the figures came across as text rather than numbers. A spreadsheet will not add text, however numeric it looks. Select the column and check the status bar - if it shows only Count and no Sum, there are no numbers in it.

How do I know if my numbers are text or real numbers?

Three quick checks: numbers right-align by default while text left-aligns, so a left-hugging column of amounts is text; selecting the column shows Sum and Average in the status bar only if it is numeric; and typing a SUM below it returns 0 for text. Excel's green triangle also flags it, but not reliably.

Can I just convert the column to a number in Excel?

On a tidy export, yes. On a real statement it usually stalls on currency symbols, thousands separators, accounting parentheses like (84.20) that mean negatives, and invisible non-breaking spaces - and force-converting the whole column damages the cells that genuinely should be text, like reference numbers.

Why are the dates still text?

On purpose. 03/04 is the 3rd of April under one convention and the 4th of March under another, and a PDF carries nothing that says which. Guessing would be right most of the time and silently wrong the rest, so the original text is preserved for you to convert knowing your document's convention.

Why did it say no tables were found?

Because that page most likely has no table on it. Tables without ruling lines are detected from spacing, and a paragraph is also words separated by spaces - so a detector pushed to find tables everywhere returns your prose chopped into columns, which is worse than nothing. If a real table was missed, the "look harder for tables without lines" option lowers the bar.

Can I convert a scanned PDF or a photo of a statement to Excel?

Not directly - a scan holds a picture of a table, with no characters in the file to put in a cell. Run OCR first, then convert the result, and check the figures carefully afterwards since isolated numbers are the least reliable thing OCR produces.

Does a table spanning several pages come out as one sheet?

No, you get one sheet per page. Stitching them means assuming the columns match on every page, which repeated headers and mid-table subtotals break quietly. Separate sheets show you what came from where so you can join them deliberately.

Can I get a CSV instead of an Excel file?

Yes, choose CSV at the download step. Every table goes into one flat file, each marked with the page it came from, and it opens cleanly in Excel, Google Sheets and Numbers.

What happens to my document?

It is processed and discarded the instant your download is ready - never stored, never queued, and never read by a person.