People data that is correct and protected
Headcount each area counts its own way, incomplete records, access nobody reviews and data kept longer than needed. How to spot and fix data quality problems in Human Resources.
The 5 most common problems
HR data drives each person’s pay and it is personal data: an error shows up straight away and has bigger consequences.
Headcount does not reconcile across areas
HR, payroll and finance give different numbers of people because they count with different criteria or cut-off dates.
How to spot itCompare, on the same date, headcount in the HR system, payroll and finance.
Incomplete or out-of-date records
Job, site, manager or grade not updated after a change. It affects payroll, the org chart and permissions.
How to spot itShare of records with all mandatory fields, and changes recorded late.
Joiners and leavers that do not reach every system
Someone who has left still has system access, or a new starter is missing from this month’s payroll.
How to spot itActive user accounts of leavers and joiners without payroll.
Access to personal data without control
Too many people can see salaries, absences or appraisals, and nobody reviews the permissions.
How to spot itNumber of people with access to sensitive data and the date permissions were last reviewed.
Data kept longer than needed
Candidate CVs from years ago and former employees’ files with no deletion date.
How to spot itRecords past the retention period set for their type.
Recognise any of these?
Measure your area with the free checklist: 20 checks with automatic scoring.
Download the checklist →Which indicators to track
Eight indicators for employee records, joiners and leavers, and access. Thresholds are a starting point: adjust them to your company.
| Indicator | How it is calculated | Starting threshold |
|---|---|---|
| Unexplained headcount differences | Unexplained difference in people between HR, payroll and finance ÷ headcount | 0% |
| Complete records | Records with all mandatory fields ÷ active employees | above 98% |
| Changes recorded on time | Changes recorded before their effective date ÷ changes | above 95% |
| Access closed on leaving | Leavers with access closed on their last day ÷ leavers | 100% |
| Joiners complete on time | Joiners with a complete record before day one ÷ joiners | above 95% |
| Permissions reviewed | Systems holding people data with permissions reviewed in 6 months ÷ systems | 100% |
| Data past retention | Records past their retention period ÷ total | 0% |
| Duplicate employees | People with more than one record ÷ employees | 0% |
Where to start, in three steps
Each step links to the guide and, where there is one, to the ready-made template.
Reconcile headcount
Compare HR, payroll and finance headcount on the same date, and measure how many records are complete.
Access and retention
Review who can see sensitive data, close leavers’ access and decide how long each type of data is kept.
Rules and owners
Write the rules for employee records and for joiners, movers and leavers, and name who answers for each piece of data.
Frequently asked questions
Who owns employee data?
HR answers for the employee record, payroll for pay data and IT for applying access as decided. Each piece of data needs a clear owner.
How long should employee and candidate data be kept?
It depends on the type of data and legal obligations (employment, tax and social security). Set a period for each data type and review it with your advisers or whoever handles data protection.
Why does headcount differ between HR and finance?
Usually because of different criteria: people versus full-time equivalents, different cut-off dates or temporary contracts counted another way. Agree a common definition and cut-off date.
Can I use employee data in AI tools?
Carefully: it is personal data, and some workplace uses, such as recruitment or decisions about workers, are high-risk under the AI Act. First check the legal basis, use only the data you need and consider anonymising or pseudonymising it.
Where does your data governance stand?
The free assessment tells you what to prioritise in five minutes.