HR Data Quality Checker
Read the file before you calculate anything with it. Every finding says what it would break.
Reads .xlsx and .csv inside your browser. Nothing is uploaded and nothing is changed.
The spreadsheet
Drop a spreadsheet here
Any employee list, export or report. Nothing is changed and nothing is uploaded.
The thing this exists to catch
A hire date column where some rows read 13/05/2019 and others read 05/13/2019. One of those can only be day first and the other can only be month first, so the column is two formats stacked in one place. No tool can tell which rule applies to 01/02/2019, and every tool will apply one anyway. Tenure comes out months wrong for part of the file, turnover picks up leavers in the wrong year, and nothing anywhere reports an error. The numbers simply come out wrong and look completely ordinary.
Five minutes here is cheaper than explaining a headcount figure twice.
Findings are ranked by what they break, not by how unusual they are
A problem means something downstream will return a confident wrong answer. Two dates formats in one column. An employee id used twice. A column that holds numbers in some rows and words in others.
Worth checking means it will break matching or grouping rather than arithmetic. Trailing spaces, Finance and finance and FINANCE counted as three departments, non-breaking spaces pasted in from a web page that look identical to a real space and never match one.
A note is usually nothing. A column that is 70% blank is exactly what an end date column should look like. It is said out loud so you can confirm it rather than discover it later.
It does not fix anything
There is no repair button and your file is never modified. Deciding that two rows are one person, or that a blank end date means somebody is still employed, is a judgement about your organization and not something a spreadsheet reader is in a position to make. When the finding is genuinely mechanical, spacing and capitalization and repeated rows, the Employee Data Cleaner will make the change and show you every one of them first.
Leading zeros are treated as part of the value
An employee id of 00123 is an identifier, not the number one hundred and twenty three, and a column of them is reported as text. This is the single most common way a roster stops matching: Excel opens the file, helpfully reads the ids as numbers, drops the zeros, and half your employees no longer exist in the other export.