Browse categories
Explore
Fiverr Pro
English
$
USD
I will extract multi column and scanned PDF data into clean excel or csv
Creative Writer, Editor and Collaborator
A US born writer living in the Philippines with 30 science fiction & fantasy novels written, and another fifty for which I've served as editor, co-author, producer and proofreader.
About this Gig
Copy-paste a multi-column PDF and the columns come out scrambled. OCR does the same. I fix that: your PDF data lands in a spreadsheet in the order a human reads it.
I specialize in the PDFs that break other tools: old or outdated documents, multi-column layouts, scanned pages, and tables that run across pages.
What I handle:
- Multi-column layouts read in true reading order, not jumbled lines
- Scanned/image PDFs via OCR
- Tables split across pages, merged into one, with repeated headers and footers stripped
- Hundreds of files batch-processed into the same consistent columns
- Invoices, statements, reports, catalogs, price lists, legacy records
No silent errors: every job runs validation checks (row counts, totals, dates). Anything uncertain is flagged for human review instead of being guessed.
You get: clean CSV and/or Excel, a short accuracy note, and in Premium a reusable script so you can run future batches yourself.
Honest disclosure: I use Python automation and, for messy scans, OCR and AI-assisted extraction. I review the output. Want no AI? Tell me before ordering.
Confidential, deleted on request. No document falsification.
Technology:
Excel
•
Google Sheets
•
Python
FAQ
Why do my PDFs come out scrambled when I copy-paste or OCR them?
Multi-column and older PDFs store text in layout order, not reading order, so tools mix the columns together. I rebuild the reading order before anything goes into the spreadsheet.
Can you handle scanned PDFs and hundreds of files?
Yes. Scans go through OCR. Large batches are processed by script into the same columns every time, so file 1 and file 300 line up.
How do you make sure the data is right?
Automated validation checks row counts, totals and dates. Anything the checks can't confirm is flagged for human review, not silently guessed. I won't claim 100% on illegible scans.
Is AI used?
The core work is Python automation. For messy scans I may use OCR and AI-assisted extraction, and I say so in the delivery note. If you want no AI, tell me before ordering and I'll use code-only or manual methods (quote may change).
Is my data safe?
It's used only for your order and deleted on request. I won't alter or fabricate documents.
