I will clean and deduplicate your messy spreadsheet
About this Gig
Messy spreadsheet? I clean it up so you can actually use it.
I've merged 160,000 product records from two e-commerce sites into
one clean file 98.6% of fields complete, and the missing 1.4%
documented rather than filled in with guesses.
WHAT YOU GET
· Duplicates removed exact ones deleted, near-matches flagged
in a separate column so you decide, not me
· Formats standardized dates to one format, phone numbers to one
format, names to consistent case, trailing spaces gone
· Clean .xlsx or CSV out, plus a change log of every field I touched
HOW I WORK
1. You send one file Excel, CSV, a Google Sheets export, or a
scanned PDF. If it's a scan, send one page first so I can tell
you what accuracy to expect.
2. I clean it, then check my own output against your original:
row count before and after, what was removed, what was fixed.
You get that report with the file.
WHAT I WON'T DO
I won't invent a value I can't verify. If a field is genuinely
missing, you get a list of exactly which rows, and you decide.
I won't touch formulas, macros or pivot tables unless you tell me to.
I won't promise 100% accuracy no one honestly can. What I do
promise is that every problem I find gets report
My Portfolio
FAQ
What format do I get back?
A cleaned .xlsx file, plus a CSV copy if you want one. Same column structure as your original unless you ask me to change it. Google Sheets users get a Sheets version. You also get a change log: what was fixed, what was removed, what couldn't be fixed.
What counts as one row?
One row in your finished file. Blank rows and duplicate rows you ask me to remove don't count. If your file has merged cells or multi-line cells that should be split into separate rows, tell me before ordering — that changes the count.
Can you handle Chinese data?
Yes. Chinese names, addresses, phone numbers and mixed Chinese-English fields are all fine. I read both. For dates, tell me which format you want: 2024-03-12 or 2024年3月12日.
How do you remove duplicates?
Two passes. First, exact duplicates: every field identical — removed. Second, near-duplicates: same person or record with small differences (John Smith / john smith, trailing spaces, phone written two ways) — flagged in a separate column, not deleted. I don't delete near-matches without asking. You
I only have a scanned PDF - can you help?
Yes, but it's a different job. A scanned PDF is an image, so the text has to be recognised first. That's doable, but the result always needs a manual check — OCR makes mistakes on numbers and names. If that's what you have, message me with one page first. I'll tell you how accurate it's likely to be

