Data quality is one of the biggest obstacles to overcome in creating meaningful people analytics. In this article, I will go over 5 tips to get better data quality through smarter system design.
Unreliable, incorrect and untrustworthy data keeps companies from realizing the benefits of their HRIS systems and often results in HR departments executing time-consuming data cleanups. These “fire drills” can be reduced by addressing the root causes of the errors in the data. In this article, you will find multiple tips on how to improve your overall data quality.
If you want to improve the data coming from your system, start by reviewing your system configuration. To determine if system configuration changes are needed, a company can gather and track the repeated errors within the system. They also can look at their past audits to see how the system’s configuration contributed to the error.
Companies often cite user error in data entry for poor integrity, but data entry error should never be considered the root cause of a repeated issue. These situations either need improved configuration or additional training.
Rule #1: High-quality data is a byproduct of proper system configuration.
There are many simple configuration changes that can quickly improve the integrity of data entered into the system. Below we will list four of the most common configuration changes:
If accurate information is not available to all HR staff at the time of entry, then the field cannot be mandatory. While this field might be necessary, a mandatory designation will yield inaccurate or false data. Necessary fields should not always be mandatory fields.
Example: Birthdays are a mandatory field in the system, but your German works council does not allow this data to be stored. Local German HR will need to enter a fake date to bypass the mandatory field (01/01/1900). If an employee were to transfer from Germany to the UK and the birthdate field is not corrected, then inaccurate data creeps into other parts of the system. This spread of inaccurate data can cause a once localized exception to bring the entire field under question. Allowing the field to be left blank is a better solution for high data integrity.
Rule #2: Eliminate mandatory requirements for fields not needed in every country or not always available at the time of entry.
The chart below illustrates the difference between blank data and using dummy values to complete a field. At first glance, a fully completed field looks good in a system, but this often hides inaccuracies and make errors difficult to decipher as shown with the highlighted entries.
The two highlighted “dummy” values could easily be overlooked in the Mandatory column. The 01/01/1990 typo could even be a valid date! The whole field is difficult to decipher and becomes untrustworthy. Blank values though are easy to spot and add clarity to what data should be trusted. Re-think any configuration or mandatory field that causes dummy data to be entered. Another option when available is to configure a “data not available”.
Fields which display similar or overlapping information are frequently a cause of errors in systems.
Example: Multiple fields may be used to designate if an employee is part-time or full-time in a system. These fields may include employee status, employee type, hours, FTE, employee benefits eligibility, work schedule, and other position information. These fields can easily become out of sync with one another and raise questions on data accuracy within the system. Errors are especially likely when changes are made during an employee’s tenure.
Rule #3: Eliminate and consolidate fields with duplicate information
While there is often a business need for these specific individual fields, look for alternatives to identify the required information.
Try to reduce the number of entries required during local HR’s updates. If data feeds other systems, like benefits or finance, try creating rules within integration files to calculate required information based on hours and employee type. Also, remove part-time and full-time designations from the system, these can be determined by looking at the FTE/hours fields. In this example, configure the employee status column to display Active Regular, Active Temporary or Active Contractor. It is also important to remember that full-time and part-time definitions tend to vary between countries. Ensure this field is well-defined and understood by both local HR and Finance departments.
While systems frequently include default field options, many organizations do not actively track each field. Some organizations leave unnecessary fields in their system “to start gathering data now”. These fields quickly become sources of bad data—especially if the field has no current purpose and is not included in reports or integration files.
Example: A system has a default field for “Number of Children” which seems potentially useful in the future. The field is left in the system as self-reported. Some employees fill it out. Some include all children, and others include children they list on their insurance. It’s not maintained or reported by anyone.
Rule #4: Remove fields without an immediate purpose: Do not include fields for future use that are unnecessary today.
The CEO then sees the fields and asks for an analysis of children’s impact on turnover. The poor data quality of this field readily becomes apparent and eliminates any opportunity for meaningful analysis. It’s impossible to determine which employees omitted the question versus those who do not have children. It’s difficult to know if the information is up-to-date or accurate as the number of children is a changing figure. Since the data has never had a purpose, there is no way of knowing the data quality. Also, a work council, or data protection officer, may have approved this field, but may not be comfortable with it in the context of turnover analytics.
Having the field available insinuates it is an actively maintained field but a company in this situation would have to launch a large audit. When a new field is added with a purpose, instructions can be included with detailed guidelines on criteria for including/excluding children and force a number to be selected, even if the number is 0. Data is easier to gather accurately than cleaning up a field with a mix of good quality and bad quality data.
Value adding systems
Where does local HR look when they need data to answer a question? Is it in your HRIS system? A separate spreadsheet? Payroll? If the answer is anything other than your global HRIS system, you need to investigate the reason. A system that is updated as an administrative task for local HR and provides them with no value will result in low data quality.
Rule #5: Create a system that provides value to local HR
Steps to ensure your HRIS system provides value to local HR:
- Ensure the system is configured with local field requirements. Local fields such as car allowances, national IDs and or race/ethnicity should be provided in the system. Eliminating additional spreadsheets and tracking will bring local HR back to your system as the primary, accurate system.
- Example: Your Indian employees receive many forms of total compensation including base pay, car allowances, food allowances, and mandatory bonuses. This information is tracked by local HR in a spreadsheet which they must provide to payroll. The base salary information in your system becomes outdated because HR forgets to update the field regularly in sync with their spreadsheet, it’s merely a mandatory field that provides them little meaning without the additional fields from the spreadsheet. Adding additional specific fields to the system along with reporting allows your system to become the primary system of record.
- Allow employees and/or managers access to view data in the system. Allowing local employees and managers to see the data will add an additional check for accuracy and provide additional value to local HR.
- Example: Local HR adds a new hire with the incorrect title. This error may not be noticed until a promotion/review period or another time it is brought to the manager’s attention. If the system allows managers and employees to view their job title anytime the error will be seen quickly and can be corrected immediately.
- Set up connections or integrations with local HR tasks such as local payroll, benefits and government mandated reports. These connections bring value to local HR and require that the system data must be accurate. By connecting to other systems and reports, it will also lead to additional auditing and tracking of these fields to ensure they are maintained. Frequently, when a new integration file is set up with local payroll it’s quickly realized that national IDs or employee IDs do not match up, meaning prior reports may have been missing key data.
- Example: An integration file is set up for France’s payroll. During this integration process, many employees are not connecting to the payroll provider. With a further audit, they discover a large number of typos have been made on the national ID field (France CNIs). They can correct the errors immediately and the mistake is not made in the future because it’s vital for employees’ payment. Errors in salary are now corrected quickly because they affect employee payment.
- Regularly share HR metrics, reports and analyses with local business leaders. HR business partners need to know their data is being used in strategic ways. In wanting to ensure they are providing value to local leaders, they will need to make sure their data is accurate.
- Example: Leaders receive local turnover reports quarterly and organizational forecasting recommendations are made based on these reports. Both local HR and local leaders want to ensure their reports are accurate so they can receive the proper resourcing budgets. This leads to an emphasis on data quality in the system.
A well-designed HR system can be the foundation for high-quality data. The effort put into designing a system that reduces errors will save companies from performing time-consuming audits and wasting money on analyses with low-quality data. Every audit is an opportunity to address the root cause and modify the configuration. Investing now in data integrity can yield an ongoing return in high-quality data and meaningful analytics that are trusted and respected within the entire organization.