Paste a broken export or pick a CSV: names that open as question marks, broken quotes, ragged rows, stray spaces and duplicate people are all handled, and you download a cleaned CSV you can import, the candidate email patterns in their own file, and a written reason for every row that was dropped or rewritten. Everything runs in this tab - your file is never uploaded and the page makes no network request.
You have already paid for this file, so nothing on this page asks you to pay again. Drop it below and take the cleaned CSV, the patterns and the report now - the same cleaner, in this tab, no upload and no email. What you paid for is the human pass: if the file is one the rules cannot fix (a merged column, two header rows, an export your import still rejects after cleaning), mail it to perun.intelligence@gmail.com and a person finishes it within 24 hours. 14-day refund if the delivered file does not match the description.
No time to run it here? Email the file to perun.intelligence@gmail.com with the row count and the error your import shows - the first answer is free, before any payment. A person reads it, not a form.
An .xlsx workbook is read directly: the first worksheet is unzipped in this tab and its rows go through the same cleaner, still with no upload. .xls (Excel 97-2003) and .ods are different containers and are not read here - if you pick one, the page says how to save it as CSV in three clicks.
Needs an email column — Email, Email - Work (Pipedrive) and Work Email (HubSpot) all count. Name, company and topic are used when present; Firstname/Lastname, Organization and Note are recognised too. Comma, semicolon and tab-separated files work, including a straight copy-paste from Excel or Google Sheets. A file saved by Excel itself (UTF-8, UTF-16 or Windows-1250/1252) is decoded by its real encoding rather than guessed as UTF-8, and the cleaned CSV downloads with a BOM so Excel opens it without mangling names.
Nothing run yet.
Not a screenshot. before.csv is a deliberately broken export: the same person twice under two spellings, a quoted comma inside a company field, a row with a column missing, and two addresses that cannot be emailed. The other two files are what the engine on this page produced from it, unedited.
before.csv · cleaned.csv · report.md · the email it comes back with — download all four and check the rows. A paid repair is these same three files on your export, with a person reading the report before it is sent back.
The cleaner handles what a rule can decide. It cannot unsplit a column that was merged, choose between two header rows, or rescue a file your import still rejects after cleaning. And if the file is fine but the shape is wrong - one row per line item where the API wants one object per order - the CSV to nested JSON converter nests it in the tab, free. That job is one export, done by a person: the cleaned file plus a report naming every row dropped and every address rewritten, back within 24 hours. 14-day refund if it does not match this description.
$49 · up to 5,000 rows — one export, one pass.
$79 · up to 25,000 rows — a month of form exports, or one CRM migration batch.
$99 · up to 100,000 rows — the full house list; the clean-up is automated, so size costs time, not accuracy.
What the receipt says: the charge comes from our billing account dragana.karans@gmail.com — that is this site, not a stranger. Files, questions and the refund go to perun.intelligence@gmail.com.
How the handover works: pay for the tier that matches your file, then email the CSV to perun.intelligence@gmail.com (printed again just below, and on the thank-you page); the cleaned file and the report come back to that same address, normally the same day. Nothing is uploaded to this page and nothing is stored here. Not sure which tier your file needs? Say the row count in that mail before paying - the answer is the tier you need, never a half-cleaned file.
Before you pay, or if the button will not open: write to perun.intelligence@gmail.com and describe the file (row count and what your import says). A person answers, normally the same day - and if the file is something this repair cannot fix, the answer is no rather than a charge.
Is this CSV repair free? Yes - it runs in the tab with no account and no sign-up. The $49 offers are different things: one export fixed by a person (up to 5,000 rows, 24h, a report), or the n8n pack that does this inside your own automation with Sheets logging.
Does it upload my list anywhere? No. There is no fetch, no XHR, no storage - the file is read in this tab and disappears when you close it.
Does it remove duplicate emails? Yes, by email identity: Marko.Petrovic@Acme.com and marko.petrovic@acme.com are one person, and the row with more filled fields is the one kept. The report states how many rows were merged.
Why does my Excel CSV open with question marks or broken letters? The file was saved in a legacy encoding (Windows-1250 on a Serbian/Polish/Czech Windows, Windows-1252 in the West) and was then read as UTF-8. This page decodes the bytes by their real encoding instead of guessing, names the encoding it used, and downloads the cleaned CSV with a BOM so Excel reopens it with the names intact. It also reverses the damage in the other direction - text some tool read as GB18030 (the default on a Chinese Windows, and the bug in n8n's HTTP Request node), where é is stored as a CJK character - because those bytes are still in the cell too.
Which exports does it read? Comma, semicolon and tab separated, including a straight paste from Excel or Google Sheets; Pipedrive Email - Work, HubSpot Work Email and Mailchimp Email 1 headers are recognised.
Does it tell me which addresses will bounce? No - it cleans, it does not verify. Guessed patterns are candidates you confirm before sending, and every rejected row is listed with the rule that rejected it rather than being dropped silently.
Honest limits: patterns are candidates, not verified addresses - you confirm each one before sending. Nothing is sent and nothing is stored; closing the tab discards everything. This tool dedupes inside one file only - it does not read rows already in a spreadsheet. The draft text is a template, not AI.