Strange case of 'wrong' HRIS reports

As far as I know, there are only a few things that HR leaders (at the corporate level) across companies agree on. One of them seems to be that the reports from Human Resource Information Systems (HRIS) that they receive/generate are often misleading and/or wrong. Fortunately, in most cases, nothing really bad happens because of these 'wrong reports', as they are not used for making any serious decisions. They just get converted into charts and tables and land up in presentations. These presentations are used mainly to describe (or even rationalize!) the past and not to predict the future. So no real damage happens ! Of course, if one tries to use these data to support decision making (i.e. as the basis for Human Capital Analytics), then it can lead to wrong decisions. Since Human Capital Analytics seems to be one of the key opportunities to create significant organizational value in the people domain*, we need to look more carefully at the reasons that make these reports wrong/misleading - so that we can fix them in a sustainable manner.

When errors are discovered in HRIS reports, the most common tendency is to view it as a 'data entry mistake'. While errors do creep in at the data entry stage, often the main issue lies at the levels of 'data interpretation' & 'business rules'.

Often there is no common understanding of how different data elements/terms are interpreted. Let us look at a very simple example - a request to pull out a 'list of all the staff in IT'. While this seems simple, there are many possible interpretations here (especially in a global organization). For example, this could mean a list of

(a) all staff who are doing an IT kind of job' and/or
(b) all staff whose are being paid from the salary budget of the IT department and/or
(c) all staff who are in the reporting-tree of the Head of IT

Now, to get this simple report correct, it is not enough just to use the correct definition of the term 'IT team' while generating the report. The real challenge is to build that definition into a business rule that covers all staff staff movements in IT (e.g. hire, transfer etc.) and to ensure that the rule gets followed correctly across all the countries in which the firm operates. For example, if the definition of 'team' is along reporting lines [as given in (c) above], then whenever an IT person gets hired in the firm in any country, it has to be ensured that the new hire reports directly to someone in the reporting chain of the global head of finance. In a global firm, this simple business rule might not always be that easy to implement as there could be scenarios (e.g. in a country where the size of the firm is very small) where it might seem more appropriate to make the IT person report to a business person. Of course, there are other factors (e.g. 'double-hatting') that could complicate this further. It is possible to use complex definitions to take care of complex situations. However, this would also imply more complex business rules, making the communication/implementation of the business rules difficult.

Another dimension that is relevant here is the variation in HR practices across countries. For example, let us look at a case where there is a decision to make a staff member, who has been working with the firm on a 'contract basis', a member of the regular staff. In this case the practice in one country might be to do a 'terminate and rehire' while that in another country could be just to change the 'employee class' from 'contract' to 'regular'. Thus the staff would show up in a 'new hire report' if he/she is in the first country and it won't happen if he/she is in the second country. Now, if we generate a global report, we would get misleading data on the number of new hires. Another similar issue could be 'what is a promotion' (e.g. 'increase in job level' and/or 'change in pay grade' and 'increase in pay' etc.). If we look deep enough, a large number of such issues are likely to surface in any global firm.

Any attempts to 'clean up the data' on a one time basis would be useless as the data would again go out of shape very quickly (as the problems at the data interpretation/business rules level would keep on producing 'data errors'). It is also useless to try to 'reverse-engineer' a report generation criteria (by combining a large number of HRIS fields to form a complex condition) which if applied on the current data in HRIS, would produce a result similar to a target list (e.g. a list of staff names given by the Head of IT as the list of 'his staff'). Since the data/target population (e.g. IT staff) is dynamic, the 'reverse -engineered criteria' (that captures target population at this point) might not be able to capture the target population accurately at a future point of time. So even if there are hundreds of data fields in the HRIS, need for business rules still exist !!!.

Thus to generate meaningful HRIS reports

(1) there should be a set of Business Rules & Data Standards that are clearly understood and consistently applied across geographies and businesses (which would lead to 'patterns in data') AND
(2) the report generation criteria should be aligned to these Business Rules and Data Standards (so as to capture the relevant patterns in the data) AND
(3) it should be possible to express the report generation criteria in terms of HRIS fields AND
(4) the query tool should be able to pull out the data as per the criteria

In practice, the above 4 requirements/steps might not always happen in a neat sequential manner especially since the business context and the analysis requirements keep on evolving. Usually the above requirements would also mandate significant amount of selling (and even pushing!) on the part of HR to secure buy-in from the business leadership and it is easy to get carried way by the immense opportunity to provide great information/analytics to support decision making. We should always keep in mind that the purpose of HRIS/Human Capital Analytics is to enable the business to function more effectively and not the other way around !!!

*Note : The objective of Human Capital Analytics (HCA) is to provide information and analytic support to enable better people related decisions. HCA could include, inter alia analytical reports, trend analysis, dashboards, benchmarking, predictive models etc. Analytical reports analyze people related issues (e.g. attrition) from multiple dimensions (e.g. various combinations of dimensions like tenure, age, experience, gender, performance, potential, location, job family, level, salary band,time since last promotion etc.).

Depending on the context developing HCA in a particular organization could involve a wide range of tasks. These could include requirement analysis, finalizing specifications, setting up the IT/ analytics infrastructure, benchmarking, report generation,maintaining dashboards and even building predictive/multiple regression models (e.g. to predict attrition). Since people related decisions might require various types of data that might be held in different information systems (that might not 'talk' to one another) and since many many of these systems are optimized for transactions/data storage and not for data retrieval (as required for HCA), setting up of HCA usually involves developing some sort of a data warehouse/data mart.

It is interesting to note that since developing and operating HCA involves a wide range of tasks, this would also require a wide range of skill sets/roles - Business analysis, IT/HRIS, HR specialist/consulting, statistical analysis, report generation, presentation/decision support, change management, project management etc. Now, some of these tasks/roles can be outsourced to a vendor - and hence some of the skill sets could come from the vendor. If the organization decides to staff some of these roles with internal resources - for cost and context understanding reasons, this could create new challenges. These could include issues like -'would the organization be able to hire these people easily?' - even if the organization manage to do so would it be able to provide these resources career paths within the organization (as some of these are quite specialized jobs - that could be very different from the mainstream jobs in the organization).

Apart from these staffing related challenges, there could be other significant challenges involved in developing and operating HCA. These include a avilability of data (if the current HR systems/benchmarking process don't already capture all the data required for analytics, you might have to put systems/processes in place to collect the data -this would take time and resources - also it would limit your ability to do any trend analysis as previous data would be missing), Mindset change (even if you make information/analytical support available would the managers use it in the decision making process?) and ensuring investment and sustained focus (- setting up and maintaining data collection process, analytical & reporting/ data presentation infrastructure etc. could require a lot of money and resources - is the organization ready for that - it is one thing for the leaders to say that they need information - the question is whether they would pay for it and whether they would continue to do so !)