Making analytics count
14 May 2015
Peter Reilly, Principal Associate
This blog post was originally published on the Symposium website
For those of you who follow cricket, you might have noticed a recent challenge to evidence-based decision making. When Peter Moores, the England coach, was asked why his team had lost to Bangladesh in the World Cup he was condemned for saying: ‘We shall have to look at the data.’ To former cricketers this was all wrong, as his ‘instinct’ should have told him what the problem was. And before agreeing with me that evidence should trump instinct, think of how many managers run their empires on the basis of a ‘gut feel’ for the situation.
Customers of our organisations often are at the sharp end of evidence-free decision making – queuing up at airports, packed on station platforms; failing to get a doctor’s appointment as needed; unable to go in to a sports ground on time – all because of staff shortages. Of course, sometimes this poor service is deliberate in putting profit before service, or political in not wishing to invest in public service. But often it is because of poor planning and the absence of quality data.
When researching good practice in HR analytics for the Ministry of Defence over the last few months, I was struck by the fact that although organisations may not use a formal definition of the term, there was general agreement that analytics is about improving the quality of business decisions and solving business problems. It is not about churning out endless management reports stacked full of statistics that rarely get read by recipients. It is not about HR’s navel-gazing concerns about its own functional performance. It is all about calculating the number of border staff, train drivers, doctors or security officers that are required to meet varying flows of customers by plane load, by the hour or the season.
It addresses questions on how to improve workforce productivity, reduce accidents, restructure effectively, change culture, etc. So it might concern itself with the standard HR metrics on absence, turnover, time to hire etc, but these are deployed in connection with data on the impact that these people management issues have on business performance. Indeed, it is important to note that HR analytics reverses the approach that characterises so much HR work of producing data then trying to find a problem to solve; instead, you start with the problem and look for the data to answer the question.
Given the business-driven subject matter of HR analytics, much of these data are held outside HR. They sit in Marketing (on customer preference), Finance (on detailed costings), in Health and Safety (on accident patterns) and in the business (on things like output, quality, speed etc). It is this combination of people and organisational data that persuades some in HR analytics that they should move to a cross-functional structure where information and skills can be pooled. Certainly, the organisations I have talked to have tended to recruit staff into HR analytics from outside the function on the basis that few in traditional HR either have the skills or the mindset to do this sort of analytical work. Many it is true are ‘converted’ to HR by focusing on the people-related business problems but their own background inclines them to work with other disciplines and to see issues in a rounded, holistic business-centred way.
Having said that, it is obvious that HR analytics requires a range of skills:
- Frontend consultancy, relationship management – helping the client specify the problem
- Data and systems awareness to know what data there are, their quality and how they can be extracted
- Statistical, analytical problem solving
- Back-end visualisation, communication and reporting results.
The first and last of these skill sets should well be within the compass of HR and indeed should be part of the sort of capabilities possessed by HR business partners themselves. Effective HR analytics involves co-working between the business partner acting on behalf of the business customer and someone from the analytics team with the aptitude for the way in which consultants are able to get to the nub of the problem at the beginning and with the ability ‘to tell a story’ at the end such that it leads the customer to take action.
In the middle are the skills likely to be the preserve of HR analytics, the technical ability to turn the raw data into meaningful results. But again how organisations do this varies. Some hire deep experts who can undertake a whole range of complex tasks. Others are content to employ those with good statistical skills whilst being prepared to contract out the difficult stuff to third parties like universities. The latter approach may well reflect the fact that most analytical tasks do not require fancy statistics; predictive modelling is usually a small proportion of the workload.
However, what is key in undertaking the simple or more complicated statistics is having sound data – accurate and well defined – and IT systems that link data together and have good reporting features. So whether or not HR analytics teams are structurally connected to those responsible for data and systems, they are terribly reliant on their effectiveness. In similar vein, management information and reporting may or may not be part of a bigger HR analytics team but ensuring that the standard stuff is well communicated to managers is essential. Disputed facts, variations in reported truth, and inconsistent messages do not just irritate customers but clog up the wheels of the HR analytics operation. Rather than doing structural equation modelling, the analytics team is arguing about the definition of an FTE or clarifying that data differences are due to different extraction dates.
The Symposium conference that I am chairing in June will give attendees the opportunity to find out how to set up and run an effective HR analytics team. It also allows you to think about the connection between workforce planning and analytics where good quality data are clearly vital, and between talent management and analytics where the link may be less obvious as the focus may be on selection and development and less on calculating a desirable flow of talent into leadership positions. Employee engagement is perhaps the best example of where the relationship between people and organisational performance has been established and the analytical underpinnings are obvious.
The conference programme also demonstrates that for many in HR analytics there is a progressive move from getting data and reporting them, towards understanding and describing the current situation, diagnosing today’s ills before looking to understand and respond to future challenges. Even the best organisations find this progress a slow one as they build capability both in their own team and in ‘intelligent customers’; they create a sense of organisational confidence in what HR analytics can do and they demonstrate a decent rate of return on the investment in this area.