Data cleaning for analysts who've outgrown Excel

Clean 10M-row files right in your browser.

Mungr is a data cleaning and preparation tool that runs a real analytical database (DuckDB, compiled to WebAssembly) locally in your browser. You can munge, wrangle, and tidy millions of rows of CSV data without uploading, installing, or waiting on IT.

Register Free →

No credit card · Free up to 5GB · Your data never leaves your browser

100% client-side
DuckDB + WASM
Tested at 10M+ rows
Built for HIPAA & GDPR
messy_export.csv · 9.8 GB on disk cleaning…
A
B
C
D
1
J.DOE@MAIL..COM
13/45/2024
 1,204  
NULL
2
m.tan@corp.io
2024-02-02
88
active
3
m.tan@corp.io
2024-02-02
88
active
4
Müller, A.
2024-02-09
312
active
0 rows processed · 0 bytes uploaded
The usual options

If you work with regulated data, none of the usual tools actually work for you.

Every path fails a different way. And if you're in healthcare, finance, or legal, the cost isn't just frustration. It's a compliance violation.

OptionBreaks aroundWhat happens
Excel / Google Sheets Heavy formulas & large files It freezes and slows to a crawl long before you hit the sheet's row limit, especially with real-world formatting and formulas.
OpenRefine / local scripts >1M rows It needs installing and IT sign-off before anyone can touch it. Then it runs out of memory the first time someone hands it a real export.
Cloud data platforms No row limit They handle the scale, but at real cost. And you still have to upload raw data off your device to get there.
Healthcare, finance & legal teams Every file Uploading data anywhere, cloud tool or not, risks a HIPAA, GDPR, or client-confidentiality violation before you've cleaned a single row.

What if cleaning your data never required uploading it in the first place?

The fix

One tool for data preparation, warehouse-scale performance, and zero data exposure.

You get a real query engine in your browser, not a toy.

Handle 10M+ rows without the spinning wheel

DuckDB compiled to WebAssembly is a real analytical database engine running locally in your browser, not a JavaScript parsing hack.

Tested at 10M+ rows on 16GB RAM laptops (Lenovo & Dell) with no crash

Your data never leaves your device, PII included

Processing happens entirely client-side. Smart Scan flags PII against known categories like HIPAA Safe Harbor so you can mask, redact, or hash it before it's ever shared.

Nothing to upload means nothing to violate HIPAA or GDPR with

Zero install, zero setup

It's a web app. Open a tab, drop a file, and start cleaning. No pip installs, no Docker, no IT ticket.

Time to first cleaned row: under 30 seconds
Data munging · data wrangling · data cleansing

What's the difference between data munging, data wrangling, and data cleansing?

The short answer: they all describe the same job, preparing raw, messy data until it's ready for analysis. Here's how the terms break down, and where Mungr fits.

Data munging

Munge it into shape

Data munging means transforming raw data into a usable format: fixing date formats, mapping aliases, combining fields, and resolving inconsistencies. It's the "get this mess into a shape I can work with" step of data preparation.

Data wrangling

Wrangle it into order

Data wrangling covers the whole loop: discovering problems, cleaning them, restructuring, and validating the result. Many analytics teams use "data wrangling" as the umbrella term for all of data preparation.

Data cleansing

Clean out the errors

Data cleansing is the cleaning part specifically: removing or fixing incorrect, duplicate, incomplete, or wrongly formatted records. Think null handling, deduplication, and standardizing values before analysis.

Whether you call it data munging, data wrangling, or data cleansing, Mungr does all three in your browser with zero uploads. It's the data cleaning tool for analysts who refuse to send sensitive files to a server. Read the data cleaning 101 guide, see how Mungr stacks up against other data cleaning tools, or read the docs.

Compliance by architecture, not by policy

Mungr finds the PII. You decide what happens to it.

Nothing ever leaves your browser, so HIPAA and GDPR concerns aren't a separate workflow. They're handled before your data goes anywhere.

facilities.csv Cleaned Smart Scan
phoneVARCHAR
facility_typeVARCHAR
bed_countBIGINT
accreditedBOOLEAN
phoneVARCHAR
⚠ Sensitive PII detected

Phone number (HIPAA Safe Harbor #5). Mask or redact before sharing externally.

Mask Redact Hash
Null values0 (0%)
Unique values25
Total values25
Smart Scan Results Found 2 issues across 10 columns in 558ms
1 compliance
phone: Phone detected (HIPAA Safe Harbor #5)

25 of 25 rows (100%) match phone pattern. Consider masking before sharing. 25 rows

1 integrity
bed_count: High null rate (68%)

17 of 25 values are missing. Consider filling with a default or filtering these rows. 17 rows

Dismiss Apply Selected
Clean once, reuse forever

Turn your cleaning steps into a recipe your whole team can reuse.

Build the sequence of steps once, save it as a recipe, and export it as JSON. Next time the same kind of export lands on your desk, a clean file is one click away.

1 Trim whitespace on all columns
2 Standardize case on email
3 Standardize date format on admitted_date
4 Mask PII in phone HIPAA Safe Harbor
5 Fill nulls in bed_count with 0
6 Extract with regex from zip_code
+ 9 more steps…
recipe_facilities_cleanup.jsonCopy JSON
{
  "recipe": "facilities_cleanup",
  "steps": [
    { "op": "trim", "cols": "*" },
    { "op": "case_standardize", "col": "email" },
    { "op": "date_standardize", "col": "admitted_date" },
    { "op": "pii_mask", "col": "phone", "rule": "hipaa_safe_harbor" },
    { "op": "fill_null", "col": "bed_count", "value": 0 }
  ]
}

Share the JSON in Slack, Notion, or your team's recipe library. Anyone on the team can load it and apply the same clean-up to their own file, and nothing about the underlying data has to be shared for the recipe to work.

10M+
rows cleaned and tested, no crash
0
bytes of your data ever sent to a server
5GB
free data processing, no credit card required
Pricing

Start free. Upgrade when you outgrow 5GB.

No card required to start. Cancel anytime.

Monthly Yearly save ~20%
  Free Propopular Team
Price $0 $29/mo $23/mo $49/mo per seat $39/mo per seat
Processing Up to 5GB total Higher processing headroom Higher processing headroom, pooled across seats
Install Nothing to install. Runs in your browser. Nothing to install. Runs in your browser. Nothing to install. Runs in your browser.
Data location Never leaves your device Never leaves your device Never leaves your device
Recipe history Last 3 recipes Unlimited Unlimited
Collaboration None Saved pipelines Shared recipe library & admin roles
Minimum None None None
Register Free Get Pro Get Team

Need more room without a subscription? Buy a one-time top-up of 5GB for $5 , no expiry rush, usable whenever you need it.

Student, academic, or nonprofit? Apply for a discount

Need a fairer price where you live? Ask about Purchasing Power Parity (PPP) pricing at hello@getmungr.com

Under the hood

Built for real-world messy data, not toy demos.

Every feature exists because a real file broke a real tool first.

01

Native CSV & Parquet

Read columnar Parquet files directly at columnar speed, with no conversion step needed.

02

19 data cleaning operations

Standardize dates and casing, trim whitespace, handle nulls, deduplicate, and extract fields with presets or your own regex.

03

Automatic PII detection & redaction

Flags identifiers against known compliance categories like HIPAA Safe Harbor, then masks, redacts, or hashes them in one click.

04

Smart Scan

One click surfaces compliance and data-integrity issues across every column, with a recommended fix for each.

05

Streaming architecture

Processes data in chunks instead of all at once, so a 10M-row file doesn't need 10M rows of RAM.

06

Recipes you can share

Save your cleaning steps, export them as JSON, and let your team apply the exact same clean-up to the next file in one click.

07

Works offline

Once it loads, no connection is required to keep cleaning. Your data was never going anywhere anyway.

Before you ask

The questions every data analyst and data engineer asks us

We'd rather answer these now than have you wonder later.

We don't process, see, or store your data. It never leaves your browser, so there's nothing for us to monetize even if we wanted to. The free tier is subsidized by Pro and Team plans for people who need more headroom and collaboration.
Yes, that's the whole point. DuckDB-WASM uses columnar, streaming execution, so memory use stays roughly flat even as row count climbs. We've tested it at 10M+ rows on standard 16GB RAM laptops (Lenovo and Dell) without a crash.
You'll get a clear warning before you hit it, with an option to upgrade. We never silently fail or corrupt your work mid-clean.
Open your browser's dev tools, go to the Network tab, and load a file. You'll see zero outbound requests carrying your data, because there's nowhere for it to go.
Yes. Mixed UTF-8/Latin-1 encodings, ragged rows, stray delimiters, and nested Parquet schemas are exactly the cases Mungr was built to survive, not just the clean happy path.
Compliance here isn't a policy you have to trust. It's a consequence of the architecture. Since your file is never uploaded, there's no transfer to log, no vendor to sign a BAA with, and no server-side breach surface. Smart Scan also flags identifiers like phone numbers and SSNs against HIPAA Safe Harbor categories so you can mask or redact them before sharing a cleaned export with anyone.
When you load a file, Smart Scan checks each column against known PII patterns (phone numbers, SSNs, emails, and more) and tells you exactly which HIPAA Safe Harbor category it matches, with links to the relevant guidance. From there you choose to mask, redact, or hash. Nothing happens automatically without your say.
Yes. Any sequence of steps can be saved as a recipe and exported as JSON. Share it however your team already shares things (Slack, Notion, a shared drive), and anyone can load it and apply the exact same clean-up to a similar file in one click. The recipe travels. The data never has to.
Yes. Data munging, data wrangling, and data cleansing all describe the same job: turning raw, messy data into something ready for analysis. Mungr is exactly that, and it does it in your browser with zero uploads, which most data munging tools can't claim.
Data munging means transforming raw data into a usable format. Data wrangling is the broader loop of discovering, cleaning, restructuring, and validating data. Data cleansing focuses on fixing incorrect, duplicate, or missing records. They overlap heavily, and Mungr covers all three with transforms for dates, casing, nulls, duplicates, PII, and more.
No relation. Mungr (spelled M-U-N-G-R) is a data cleaning and data munging tool for analysts and data engineers. We clean data, not portfolios. The name comes from combining "mung" and "r", not from any person.

Stop fighting your file size.
Start cleaning.

Free up to 5GB. No install. No upload. No credit card.

Register Free →

Set up in under 60 seconds · Your data stays on your device, always

Clean 10M rows, privately.

Register Free