How to convert PDF to CSV
You need the data out of a PDF and into a CSV, so it can be imported into a database, an accounting tool or a script. Here is how to get a clean, comma-separated file in one click, with every column landing in its own field.
CSV is the plainest way to move tabular data between tools: just rows of values separated by commas, and almost everything can read it. The problem is getting there from a PDF. Copy-pasting collapses the columns into one blob, and manually retyping a statement or price list is exactly the kind of work nobody has time for. The fix is a tool that reads the table's real structure and writes it out field by field.
The fastest way: a one-click Chrome extension
Convert PDF to Excel is a free Chrome extension that pulls the table out of the PDF you are viewing and exports it as CSV, right in your browser.
- Add the extension to Chrome and open your PDF.
- Click the extension icon. It detects the table and rebuilds it column by column.
- Choose CSV as the export format.
- Download a clean file where each column is its own comma-separated field, ready to import.
Why not just save as CSV from a viewer?
Most PDF viewers have no CSV option at all, and the ones that export text give you a single wall of words with the columns gone. CSV only works when the field boundaries are correct: one comma per column, one line per row. The extension keeps that structure, so a value like 1,488.00 stays in the right field instead of splitting your row in half at the thousands comma.
What a clean conversion actually looks like
Here is the difference in practice. A three column invoice row in the PDF reads Consulting fee 12 1,488.00. Copy-pasted into a text editor it arrives as one string with the spacing lost, and any import treats it as a single field. Converted properly it comes out as:
Description,Qty,Amount
Consulting fee,12,"1,488.00"
Hosting (annual),1,"240.00"Two details do the work here. The thousands comma inside the amount is quoted, so it cannot break the row into an extra column. And the header line survives, which is what lets the next tool map the fields automatically instead of asking you to do it by hand.
CSV or Excel: which should you pick?
Pick CSV when the next step is an import: a database, a CRM, an accounting package or a Python script all take CSV happily. Pick Excel (.xlsx) when a person will open the file and you want formatting and formulas to survive. The same extension exports both, so you are not locked into one.
A practical rule: if a machine reads the file next, choose CSV. If a human reads it next, choose Excel. CSV carries no formatting, no formulas and no multiple sheets, which is exactly why importers like it and exactly why it is the wrong format to hand to a colleague for review.
Getting the CSV into the next tool
Once the file is on disk, the import is the easy part:
- Google Sheets: File, Import, Upload, then pick "Detect automatically" as the separator.
- Excel: Data, From Text/CSV, check that the preview shows separate columns before you load.
- A database or script: point your importer at the file directly. With the header row intact,
pandas.read_csv()or aCOPYstatement needs no extra arguments.
If the preview shows everything crammed into one column, the export lost its field boundaries. That is the symptom the conversion above is meant to prevent.
Why not just ask ChatGPT to do it?
It is a fair question, and the answer is that a chat model is the wrong instrument for a grid of figures. ChatGPT handles prose well, but with tables it paraphrases: it can reorder rows, merge or drop columns, round values, or quietly fill a gap with a number that was never in the document. On an invoice or a bank statement one altered figure is a real problem, and nothing in the output tells you which one changed.
There is a privacy angle too. Using a chat model means uploading the document to a third party, which is not ideal for financial or personal files. A browser extension keeps the file on your device.
Where AI genuinely helps is after the extraction. Get the data out exactly with a tool built for it, confirm the totals match, then hand the clean CSV to ChatGPT for summarising, trend spotting or drafting an explanation.
What about scanned or photographed PDFs?
Text-based PDFs convert to CSV for free. If your file is a scan or an image-only PDF, like a photographed invoice, an optional AI mode reads it too. It is opt-in, so you decide when to use it, and plain text PDFs never need it.
Need the full spreadsheet instead?
If you would rather keep formulas and formatting, export to Excel instead of CSV: it is the same one click, just a different format in the export step. To go the other way, see our guide on converting Excel to PDF.