Data cleaning examples: six real jobs, before and after
Six files that turn up over and over, what is wrong with each one, and the exact rows the engine proposes to change. Every table below is the same diff the demo produces.
CRM migration cleanup
Six years of accounts and contacts, entered by four generations of sales team. The same company exists three times under three spellings, and the migration will faithfully carry all three across.
After
Duplicate accounts merged into the most complete record, with an audit log the AE team can read before they lose a record they were using.
| Original rows | Proposed | Rule |
|---|---|---|
| Ridgeline Partners / Ridgeline Partners LLC | One account, the fuller row kept | Fuzzy company match, legal suffix |
| marcus.hall@ridgeline.com / m.hall@ridgelinepartners.com | Both kept on the merged record | Cluster merge, contact fields preserved |
| Last Contacted 4-Mar-26 / 03/04/2026 | 2026-03-04 | Date format |
Every one of those rows is a proposal you approve or reject before anything is written.
Campaign list hygiene
A list assembled from a webinar export, a content download and a purchased segment. The same person appears with a work address, a personal address and a typo, and the bounce rate is about to tell everyone.
After
One row per person, addresses lowercased and trimmed, undeliverable strings flagged rather than silently sent to.
| Original rows | Proposed | Rule |
|---|---|---|
| Jose.Munoz@Acme.com | jose.munoz@acme.com | Whitespace and email casing |
| j.munoz@acme.com (same person, second row) | Cluster, one row kept | Fuzzy name and company match |
| sarah@@northwind.co | Flagged, not sent | Invalid address shape |
Every one of those rows is a proposal you approve or reject before anything is written.
Multi-source financial reconciliation
Three exports from three systems. One writes dates day-first, one month-first and one as a spreadsheet serial number. Amounts arrive as text with currency symbols, and negatives are in brackets.
After
Every date in ISO 8601 and every amount a plain number, so the three files finally join and the totals tie out.
| Original value | Proposed | Rule |
|---|---|---|
| 46085 | 2026-03-04 | Spreadsheet serial date |
| $12,400.00 | 12400.00 | Currency format |
| (1,250.50) | -1250.50 | Parenthesised negative |
Every one of those rows is a proposal you approve or reject before anything is written.
Event and lead capture cleanup
A badge-scan export where names, companies and emails were typed by hand at a booth. Everything is shouted in capitals, half the accents are broken, and the same visitor was scanned twice.
After
Names recased and repaired, duplicate scans clustered, and the list usable the same morning instead of at the end of the week.
| Original value | Proposed | Rule |
|---|---|---|
| JOSE MUñOZ | Jose Muñoz | Encoding repair and casing |
| ACME CORP . | Acme Corp | Whitespace and punctuation |
| Two scans, same badge | One row, both timestamps kept | Duplicate cluster |
Every one of those rows is a proposal you approve or reject before anything is written.
Supplier and product catalog merge
Two vendor lists that describe overlapping products under different SKUs and different supplier spellings. Nobody knows how many suppliers there really are, so nobody can negotiate.
After
Supplier names standardized and matched, SKU casing and padding made consistent, and the real supplier count visible for the first time.
| Original value | Proposed | Rule |
|---|---|---|
| northwind trading co. / Northwind Trading Company | One supplier | Fuzzy company match |
| sku-00421 / SKU 421 | SKU-00421 | Identifier casing and padding |
| 0042 | 00042 | Leading zero restored |
Every one of those rows is a proposal you approve or reject before anything is written.
Address file standardization
A shipping file where states are written four ways, postcodes lost their leading zeros in a spreadsheet, and the country column mixes codes and names.
After
One form per state and per country, postcodes restored to their real length, and a file a validation service or a mail house will accept.
| Original value | Proposed | Rule |
|---|---|---|
| Calif. / CALIFORNIA / california | CA | US state standardization |
| 2139 | 02139 | Leading zero restored |
| United States of America / usa | US | Country standardization |
Every one of those rows is a proposal you approve or reject before anything is written.
Nobody made these files badly on purpose
Every one of them was produced by a system doing exactly what it was told, over several years, by people who each had a reason for typing it that way.
The damage is entered, not introduced
A booth scan, a form with no validation, an export through a spreadsheet. The file is the sum of every hand that touched it.
The fix has to be reviewable
Somebody will ask why a record moved. A proposal you approved answers that; a script that ran overnight does not.
It happens again next quarter
The same export arrives with the same faults, which is why the work is worth saving as a recipe rather than repeating by hand.
The rules are boring and that is good
Dates, casing, leading zeros, legal suffixes. Nothing here needs to be clever, it needs to be applied consistently and shown to you.
Run one of these on a real file now
The three samples below are the messy versions of the cases above. Or drop your own CSV and see what it finds.
1. Load a file
Your file is read on your own machine. Nothing is uploaded, so nothing can leak.
Or start from a messy sample
Status
Loading the sample file.
File: CRM contact export
Rows
--
Columns
--
Changes
--
Duplicate rows
--
Date formats
--
Blank emails
--
Load a file to see the diff
Drop a CSV or pick one of the messy samples. Every proposed change appears here as a line you approve or reject.
2. Review every change
Nothing approved yet.
Nothing is applied to your file until you approve it, and your original file is never modified.
Proposed changes
Original Proposed
Showing the first 400 changes. 0 more are in the audit log download.
Duplicate clusters
Merge or keep both
Want this on bigger files, on a schedule, with your team?
Saved recipes, scheduled cleans and shared seats are part of the paid plans. Start with your email and we will send you a code.
3. Take the result
The downloads work without an email. The cleaned CSV contains only the changes you approved.
Clean the file, and be able to show what you changed
Try the demo on your own CSV first. It runs in your browser, it costs nothing, and no card is required.
No card required. Your file never leaves your computer.