Shortfalls in data quality, completeness or accessibility should not prevent us from presenting workforce reporting and executing people analytics.
The very process of reporting with questionable data shines a light on the process of collection, and assuming we are asking the relevant questions that have the potential to provide value-add insight, there will be a demand for people analytics that insists data issues are addressed.
Of course where data isn’t trusted, we should provide a health warning, but within reasonable margins: does it matter whether the resignation rate of a critical resource has sharply increased to 15% or 17%? Regardless of the number, there is a problem that needs to be analysed and addressed.
Our sensitivity to data quality will change according to the question being asked. Therefore, transparency about quality and errors is an important part of people analytics. We should report our findings along with our assessment of the data quality, and allow the decision-maker to use their judgement when using the insights.
It’s important to accept (and educate those who use people data in our business) that the majority of people data will never be perfect.
Subjectivity
Much of our data about people is based on opinion, and therefore a degree of subjectivity. For example, a manager’s view of employee annual performance may be weighted towards more recent contributions, or a customer’s perception of service provision may be skewed by factors not in the control of the front-line employee.
Timing
Much of data is extracted from ‘live systems’, therefore the ‘truth’ today may be out of date tomorrow.
Data capture
The process of data capture can present a significant risk, particularly when the process is unclear, ignored or unmonitored; or key performance indicators measure efficiency (in the form of speed) rather than effectiveness (in the form of quality).
Reporting accuracy
Finally, unless we have the structures, competencies and reconciliations that ensure our reporting and analysis reflects the data held on source systems, we will lose credibility and therefore the permission to present our analysis.
HR practitioners analysing people data must accept that a proportion of our information is based on opinion, and is therefore variable, and the fluid nature of our systems and data repositories will sometimes present slightly disconnected versions of the truth.
We must also highlight (not necessarily take ownership or responsibility for) data capture process errors, while ensuring that the data we use is auditable back to its origins.