CSV is the format everything accepts: accounting systems, databases, import wizards, statistics packages, anything with a "choose a file" button. So "convert this PDF to CSV" is usually not really about CSV at all — it is about getting a PDF's contents into something another system will swallow.

That framing matters, because CSV is an unusually unforgiving destination. It is a text file with commas in it and no way to say what anything means, which is liberating until a description contains a comma or a reference number starts with a zero.

What a CSV file actually is

One line per row, values separated by commas, and nothing else. No sheets, no column types, no formatting, no formulas, no merged cells, no second tab. A spreadsheet program displays a CSV as a grid, which makes it easy to forget that the grid is the program's interpretation rather than something in the file.

Two consequences follow, and between them they cause nearly every CSV problem anyone has:

A value containing the separator has to be quoted. Widget, blue becomes "Widget, blue", and a value containing a quotation mark doubles it. Tools that get this wrong produce files that are off by one column from the first description containing a comma onwards.

Nothing in the file says what a column holds. 00123 is a five-character string in the file and a number as far as the importer is concerned, so it arrives as 123. 04/08/2026 is the fourth of August or the eighth of April depending on the machine reading it. The file does not have an opinion; every reader supplies its own.

Decide the columns before you extract anything

Because CSV has no structure beyond its rows and columns, the column set is the design of the whole job, and it is the thing that is expensive to change later.

Source file
Supplier
Reference
Date
Currency
Net amount
Tax amount
Gross amount

Source file first. One column, and it is what lets anyone looking at a suspicious figure open the document behind it. In a CSV there is no cell comment and no second sheet to hide provenance in, so if it is not a column it does not exist.

Keep one value per column and no units inside a figure. 1240.00 with a separate GBP column imports cleanly everywhere; £1,240.00 imports as text in most places and as 1.24 in a few. Extracting specific fields from PDFs goes through how to name fields so each has exactly one answer — that discipline matters more for CSV than for Excel, because there is no type information to fall back on.

Check the PDF has text first

A PDF either contains text objects or pictures of text. No CSV conversion of any kind is possible from the second kind until the characters have been recognised from the image.

Extract Text from PDF answers this in your browser in a few seconds, per page, with nothing uploaded: if text comes back the document is readable, and if none does it is a scan and converting scanned PDFs to Excel is the relevant route — everything it says applies equally when the destination is CSV.

This is thirty seconds of work that prevents the most common wasted afternoon in document conversion.

How to get from PDF to CSV

Four routes, and which is right depends on whether you want a table verbatim or fields from many documents.

One table, verbatim: a table converter, then save as CSV. If the PDF contains a single large table and you want it as-is, a PDF table extractor followed by "Save as CSV" is the short path. What makes this work or fail is the table itself — extracting tables from PDF covers why ruled tables succeed and whitespace-aligned ones do not.

Copy, paste, then split. Paste the text into a spreadsheet and use Text to Columns, splitting on a delimiter or by fixed width. Genuinely effective on monospaced, fixed-width reports — bank statements and system printouts especially — and frustrating on anything proportionally spaced. Be aware that PDFs store text with positions rather than reading order, so a two-column layout often pastes as interleaved lines.

Fields from many documents, then export. When you want the same handful of values from each of fifty documents, extract to a sheet with one row per document and export that to CSV at the end. This is the workflow in converting multiple PDFs into one spreadsheet, and CSV is simply the last step of it rather than a different method.

Ask for the data instead of the document. If the PDF came from a bank, a portal or an accounting system, a CSV export almost certainly exists behind a login. It will be better than anything extracted from the PDF, and checking takes five minutes.

Open it correctly, which is not by double-clicking

This is where most CSV data gets quietly damaged, and it happens after the conversion rather than during it.

Double-clicking a CSV lets the spreadsheet guess every column's type with no chance to intervene. Leading zeros vanish, references like 1-2 become dates, long numbers become scientific notation, and NA or TRUE become something other than text. None of it is flagged.

Import it instead. In Excel: Data → From Text/CSV, which opens a preview where you set the delimiter, the file encoding and each column's type before anything is parsed. Set reference and code columns to Text explicitly. In Google Sheets: File → Import, and turn off "Convert text to numbers, dates and formulas".

Two settings are worth knowing by name:

Encoding. Choose UTF-8. If supplier names come through as é where é belongs, the encoding was guessed wrong, not the file.

The separator. In locales that use a comma as the decimal mark, CSV files are frequently semicolon-separated, and the same file opens perfectly on one colleague's machine and as one long column on another's. The import dialog is where you fix that; the file is not wrong.

Checks to run before the CSV goes anywhere

Five, and they take a minute.

  1. Row count against document count. Equal, or you know what is missing.
  2. Column count on every row. A row with one extra field is a quoting failure, and it means every value after it on that row is in the wrong column. Most importers will tell you which line.
  3. Leading zeros intact. Look at the reference column specifically.
  4. Dates all one way round. Sort the column. A block of values where the day and month have swapped is immediately visible at the boundaries.
  5. Arithmetic that should hold. Net plus tax equals gross, on every row. This catches far more than it looks like it should, because misreads break arithmetic more often than they preserve it.

Then keep the original PDFs. A CSV carries no audit trail of its own, which is exactly why the Source file column exists.

When CSV is the wrong destination

CSV is the right answer when another system is going to read the file, and the wrong one when a person is. If the output is going to be read, filtered, totalled or kept, Excel holds column types, number formats, multiple sheets and the formulas that do your checking — all of which a CSV discards by design.

The short version of the trade-off: CSV for machines, Excel for people. PDF to CSV vs PDF to Excel works through the cases where that rule of thumb is not enough.

Common questions

Can one CSV hold several sheets? No. One file, one table. Several tables means several files.

Why did my reference numbers lose their leading zeros? Because the file was opened rather than imported, and the column was guessed to be numeric. Re-import and set that column to Text; the data in the file was never wrong.

Why is my CSV one long column? The separator does not match what the reader expected — usually a semicolon-separated file opened by a comma-expecting reader, or the reverse. Set it explicitly on import.

Should I use tab-separated instead? Often a good idea. Tabs appear in business data far less often than commas do, so there is less to quote and less to go wrong. Name the file .tsv so the reader does not have to guess.

Can I convert a CSV back to a PDF table? Yes, by way of a spreadsheet, and the round trip loses nothing you had in the CSV — because the CSV had no formatting to lose.

What about UTF-8 and Excel? Excel's From Text/CSV import handles UTF-8 properly. Double-clicking historically did not, which is another reason to import.

The habit worth keeping

Choose the columns first, put Source file at the front, keep currency and units in their own columns, and import rather than open.

Those four habits survive every tool change, and they are what separate a CSV that another system accepts on the first attempt from one that comes back with a row number and no explanation. The free tools page has the browser-side utilities for the PDF half of the job.