Free online CSV & spreadsheet tools
Clean, convert, and chart your data — each tool does one job well, with no spreadsheet formulas, no install, and no upload. Your file is processed entirely in your browser, so it never reaches our servers.
Remove duplicates from CSV
Delete duplicate rows by any column, keep the first or last match, and download a clean file.
Open tool →Clean phone numbers
Format a messy column of phone numbers to the E.164 standard, with no upload.
Open tool →Clean email column
Lowercase, trim, validate, and dedupe an email column in a CSV — entirely in your browser.
Open tool →Split full names
Break a full-name column into separate first and last name fields, with optional middle name.
Open tool →Combine columns
Merge two or more CSV columns into one — pick a separator and name the new field.
Open tool →Find and replace in a CSV
Swap text in a column like Excel's Find & Replace — match case or whole cell, with no upload.
Open tool →Convert CSV to Excel
Turn a CSV into a formatted .xlsx workbook in your browser — no upload, no install.
Open tool →Convert Excel to CSV
Export the first sheet of an .xlsx or .xls file to a clean CSV — entirely in your browser.
Open tool →Convert CSV to PDF
Lay a CSV out as a paginated PDF table with repeating headers — printable, and nothing uploaded.
Open tool →Visualize a CSV as a chart
Summarize a spreadsheet into a bar, line, area or pie chart and export the image — nothing uploaded.
Open tool →Prepare a CSV for analysis
Decide what one row represents before deleting duplicates: repeated purchases by one customer are separate transactions. Standardize category labels and matching fields, preserve identifiers as text, then compare row counts and inspect a chart for unexpected groups. Keep missing values distinct from zero unless your analysis defines them as equivalent.
These tools handle a file-level cleanup. They do not replace a reproducible database or scripting workflow, fuzzy record matching or a pipeline for files that exceed available browser memory.
What is data cleaning?
Data cleaning means correcting or removing inaccurate, inconsistent, incomplete or repeated records so a file is fit for its next task. Data cleansing and data scrubbing describe the same work. Cleaning does not prove that a contact is reachable or that a value is true.
Keep an original copy. Define the required columns, trim unwanted whitespace and standardize the fields you will match. Then remove duplicates using an appropriate key. Review missing values rather than inventing replacements, and check row counts, flagged records and a sample of the final export before importing it.
Validation checks whether values meet rules; it can run during entry or after a cleanup. Cleaning repairs existing values. Improving the source form or export can prevent the same errors next time.
Where your file is processed
Sigmera cleans files inside your browser. Server-based tools upload the file for processing; check their retention policy before using customer or recruitment data. Desktop tools process files locally but require installation. Cloud spreadsheets store files online; a desktop spreadsheet can work with a local file.
Choose a tool by the job: formatting, exact duplicate removal and splitting columns are separate from fuzzy matching or clustering. Browser capacity depends on available memory and the size of the file. Preview results before exporting, and keep the original file.
Once a table is clean, turning its rows into Word or PDF files is a separate step. Lapu AI's guide to document automation from spreadsheet data explains how to use your own templates to generate documents.
What data cleansing tools do
Data cleansing tools fix the four faults that make a fresh export unusable: duplicate rows, values in an inconsistent format, text stuck in one column that belongs in two, and whitespace you cannot see. Desktop CSV software makes you install it first, and most online CSV tools make you upload the file first. These do neither — each one is a page that reads the file in the tab.
The order matters more than the tools. Run a messy file through this sequence:
- Strip stray characters. Trailing spaces, stray quotes, and placeholder text like “N/A” all defeat an exact match later. Use find and replace on a CSV column.
- Standardise formats. Phone numbers to E.164 with the phone number formatter, addresses lowercased and trimmed with the email column cleaner.
- Fix the column layout. Break one field into two with split full names, or merge two into one with combine columns.
- Deduplicate last. Only now will the CSV duplicate remover catch the records two spellings apart. The same logic applies to the Remove Duplicates button in Excel.
- Hand it off. Export as a workbook with CSV to Excel, or sanity-check the result as a chart before you import it anywhere.
Every step above previews in full for free — the download counter only ticks when you export the finished file. For the reasoning behind the sequence, see what data cleaning actually involves.
Online CSV tools vs desktop CSV software
Desktop CSV software puts an application on your machine and works on a copy of the file on disk. Most online CSV tools avoid the install by doing the opposite — they take an upload and do the work on a server. Sigmera is a third shape: the page is online, the processing is not. The engine loads into the tab and reads the file through the browser’s own File API, so there is no installer to get past IT and no transfer to account for.
A spreadsheet is still the right tool when you need formulas, pivot tables, or a chart you will keep updating. It stops being the right tool at the edges. An Excel worksheet holds 1,048,576 rows by 16,384 columns, and it reformats values as it opens them — leading zeros dropped from postcodes, long account numbers rounded into scientific notation. A single-purpose tool that treats every field as text leaves those alone. If you do want the workbook at the end, run the passes here and convert between Excel and CSV as the last step.
The other reason to reach for a CSV tool rather than a text editor is quoting. RFC 4180 says a field containing a comma, a double quote, or a line break has to be wrapped in double quotes, with an embedded quote escaped by doubling it. A blind search-and-replace ignores that and splits "Acme, Inc." into two fields, shifting every column after it. Each tool here parses to those rules and re-quotes on export. For the spreadsheet equivalents of the same passes, the guide to separating names in Excel walks through Text to Columns, and the Excel cleaning guide covers the remaining spreadsheet steps.
Encoding: the fault that survives a cleaning pass
A CSV does not record which character set it was written in. The bytes are all a reader gets, so a file exported as UTF-8 and read by something that assumes a legacy 8-bit encoding turns Zoë Müller into Zoë Müller. No dedupe or format pass catches that, because the mangled text is still valid text. It is a decoding fault, and it happens before any cleansing tool sees the data.
One marker settles it. A byte order mark at the start of the file is unambiguous — the Unicode Consortium notes that “where a BOM is used with UTF-8, it is only used as an encoding signature to distinguish UTF-8 from other encodings” (Unicode, UTF & BOM FAQ). Without one, every reader is guessing. Sigmera checks the BOM first, then tries strict UTF-8 and falls back to Windows-1252 for a file that fails it, because Windows-1252 round-trips every byte rather than dropping the ones it cannot decode. When the guess is wrong — a Central European export, a UTF-16 dump out of an old system — the workspace takes an explicit override: UTF-8, UTF-16, Windows-1252, Windows-1250, ISO-8859-1 or ISO-8859-2, applied on a re-read of the same file.
Sort the encoding out before anything else in the sequence above. Deduplication compares strings, so a list where half the rows were decoded one way and half another will not match its own duplicates. The neighbouring failure is the spreadsheet rewriting values as it opens them: Microsoft’s own note is that “when Excel opens a .csv file, it uses the current default data format settings to interpret how to import each column of data” (Microsoft, Import or export text files) — which is how postcodes lose a leading zero and account numbers arrive as scientific notation. That damage is done at open time, and opening a CSV without letting Excel reformat it is the step that prevents it. Cleaning a file that was already altered on the way in fixes the format and keeps the wrong values.
There is a third class of fault none of these tools claim: data that is well-formed and untrue. A phone number can be valid E.164 and belong to nobody; an address can parse cleanly and not exist. A cleansing pass judges shape, not truth — the check for truth is a live source, a verification service, or a person who knows the record. What a CSV can and cannot carry sets the boundary for the rest.
How to check a cleansing pass actually worked
Every data cleansing tool reports success the moment it finishes. That report says the operation ran, not that the result is right — and a pass that removed the wrong rows looks identical to one that removed the right ones. Four checks separate them, and none of them takes longer than the pass itself.
Read the count, and have expected it first. Excel’s Remove Duplicates ends with “a message… to indicate how many duplicate values were removed, or how many unique values remain” (Microsoft, Remove duplicate values). The number is only useful against a guess you made before clicking. A 4,000-row export that loses 1,800 rows to deduplication did not have 1,800 duplicates — it was matched on too few columns.
Keep the original until the check passes. Microsoft’s instruction is direct: “Because you are permanently deleting data, it’s a good idea to copy the original range of cells or table to another worksheet or workbook before removing duplicate values.” In a spreadsheet the original is whatever you remembered to copy. Here it is the file on your disk, which no tool writes to — the cleaned version exists in the tab until you export it.
Run the pass twice. A second run over a genuinely clean file changes nothing. If the duplicate remover still finds matches on the second pass, the normalisation ahead of it did not finish — two spellings became one only after the first dedupe had already walked past them. That is the sequence at the top of this page failing in a way a single run hides.
Spot-check the edges, not the middle. The first twenty rows of an export are almost always fine. The damage sits in the rows nobody scrolls to: the accented surname, the one-word company name in a person column, the mobile number saved with an extension, the field long enough to have been truncated upstream. Sort by length or filter for non-ASCII characters and read those. Charting the cleaned file in the CSV viewer surfaces the same thing faster on a numeric column — a category that should not exist shows up as its own bar.
The spreadsheet route has the same four checks and one extra hazard, since the Remove Duplicates button edits the sheet in place; deduplicating in Excel without losing rows covers what to copy before you press it.
What each tool changes in your file
The first question to ask any data cleansing tool is what it overwrites. Here the default answer is nothing: every pass that rewrites values writes them into a new column beside the original, so the before and after sit next to each other in the preview and in the export. Only one tool removes anything.
| Tool | What changes, with default settings |
|---|---|
| Remove duplicates | Removes rows. Matches on the whole row or the columns you pick, ignores letter case unless you switch that off, and keeps the first copy. |
| Clean phone numbers | Adds phone_cleaned beside the source column, in E.164. Optionally moves numbers that fail validation into their own column. |
| Clean email column | Adds email_cleaned (trimmed, lowercased) and an email_valid flag beside the source column. |
| Split full names | Adds first_name and last_name (and optionally middle_name) beside the full-name column. The original stays. |
| Combine columns | Adds one new column at the end of the file. Source columns stay; empty cells are skipped. |
| Find and replace | Adds a _cleaned copy of the column with the replacements applied, matched anywhere in the cell or on the whole cell. |
That layout is what makes the checks above cheap. With the original and the cleaned value in adjacent columns, a spot-check is a scroll rather than a comparison between two files, and a row the pass got wrong is visible next to the value it started from. When the cleaned column is the one you want, delete the original in the importer or the spreadsheet afterwards — the decision stays with you rather than being made by the tool.
Step-by-step guides
Each tool includes examples and limitations. For the spreadsheet method, choose a guide:
Frequently asked questions
- What are data cleansing tools?
- Data cleansing tools find and fix the errors in a dataset that stop it being usable: duplicate records, values in the wrong format, text stuck in one column that belongs in two, and stray whitespace. Sigmera splits that work into one tool per problem so you can run only the passes your file needs.
- Are these online CSV tools free?
- Opening a file, configuring a clean-up, and previewing the full result is free and unlimited on every plan. The free plan includes 3 downloads a month with no credit card; Pro is $15 a month for 30 downloads and Max is $39 for unlimited. Rows and file size are unlimited on all three.
- Do I need to install any CSV software?
- No. Every tool is a web page that loads a JavaScript engine into the tab and reads your file with the browser's own File API. There is nothing to install, nothing to update, and no admin rights needed — which is why this works on a locked-down work laptop.
- Is it safe to clean a customer list in a browser tool?
- With Sigmera, yes, because the file is never transmitted. Parsing and cleaning happen in the browser tab on your own machine, so the CSV containing names, emails, or phone numbers never reaches our servers. There is no processor and no data transfer, which removes most of the GDPR questions that a normal upload-based tool raises.
- Which tool should I use on a messy export?
- Work in this order: find and replace to strip stray characters, then the phone and email cleaners to standardise formats, then split or combine columns, then remove duplicates last. Deduplicating before you normalise misses matches, because two spellings of the same record are still two different strings.
- How large a file can these tools handle?
- There is no row cap or file-size cap on any plan. The practical limit is your own machine's memory, since the whole file is held in the browser tab — in exchange, a large file costs nothing in upload time because it never leaves the device.
- What is the difference between online CSV tools and CSV software you install?
- Installed CSV software runs on your machine but needs an installer and, on a work laptop, admin rights to get there. Most online CSV tools skip the install by running on someone else's server, which means your file has to be uploaded first. Sigmera's tools are a third shape: web pages that load the cleaning engine into the browser tab, so there is no installer and no upload — the file stays on the device that opened it.
- Will these tools break a CSV that has commas inside a field?
- No. Every tool parses the file to the CSV convention documented in RFC 4180 and re-emits it to the same rules, so a value like "Acme, Inc." stays inside its quotes and stays one field. That is the main thing a CSV tool gives you over a find-and-replace in a text editor, which will happily split a quoted field down the middle and shift every column after it.
- Is data cleansing the same as data cleaning?
- They mean the same thing and the tools are interchangeable. "Data cleansing" is the older, more formal term and still shows up in enterprise software and job descriptions; "data cleaning" is the plainer phrase most people search for now. Neither implies a different technique — both describe removing duplicates, correcting formats, and filling or dropping bad values so a dataset can be used.
- Does GDPR require you to clean customer data?
- Indirectly, yes. The accuracy principle in Article 5(1)(d) of the GDPR says personal data must be "accurate and, where necessary, kept up to date", and that every reasonable step must be taken to erase or rectify inaccurate data without delay. A contact list with dead addresses, duplicate records, and mistyped phone numbers is not accurate data, so periodic cleansing is part of meeting that obligation rather than housekeeping. Doing it in the browser keeps the personal data on your own machine while you fix it.
- Why did accented characters turn into symbols after an export?
- Because the file was decoded with the wrong character set. A CSV carries no declaration of its encoding, so a UTF-8 export read as Windows-1252 renders ë as ë and every accented name looks corrupted. A byte order mark at the start of the file removes the ambiguity; without one, the reader guesses. Sigmera checks the BOM first and falls back to strict UTF-8, and you can override the encoding (UTF-16, Windows-1252, Windows-1250, ISO-8859-1, ISO-8859-2) and re-read the same file if the guess is wrong. Fix this before deduplicating, or the mis-decoded rows will not match their duplicates.
- How do I know a data cleansing pass worked?
- Predict the numbers before you run it, then read the ones you get back. If deduplication removes far more rows than you expected, the match key was too loose. Run the same pass a second time: on a genuinely clean file nothing changes, and a second pass that still finds duplicates means the normalisation before it did not finish. Then spot-check the rows at the edges — accented names, one-word values, unusually long fields — rather than the first screen.
- Should I keep a copy of the file before cleaning it?
- Yes, and Microsoft gives the same advice for Excel's own Remove Duplicates: "Because you are permanently deleting data, it's a good idea to copy the original range of cells or table to another worksheet or workbook before removing duplicate values." With Sigmera the copy is automatic in practice — the tools read your file and never write to it, so the original on your disk stays exactly as it was until you export a cleaned version as a new file.
- Do these data cleansing tools overwrite my original column?
- Not by default. The phone, email, and find-and-replace passes write their result into a new column named after the source — phone_cleaned next to phone, for example — and leave the original values beside it so you can compare row by row. The name splitter adds first_name and last_name next to the full-name column, and the column combiner adds its new field at the end. Only the duplicate remover takes rows out, and it keeps the first copy of each match.
- Can a data cleansing tool fix wrong data, or only badly formatted data?
- Only badly formatted data. These tools judge shape: is this a duplicate row, is this phone number in E.164, is this column holding two values that belong in two. They cannot tell you that a correctly formatted number reaches nobody or that a well-formed address does not exist. Verifying that a record is true needs a live source, a verification service, or someone who knows the account — cleansing gets the file into a state where that check is possible.