Personnel
staff are using statistics to their advantage with the help of sophisticated
software analytics, writes Keith Rodgers. From historical pattern analysis to
future predictions, the programs’ functions are playing a key role in HR’s bid
to prove itself in the business arena
Dave
Gibson, manager of HR systems at US farm vehicle and tools supplier Deere &
Co, has more than 20 years’ worth of data on file covering some 80,000 past and
present employees. A goldmine of statistical information? In theory, yes, but
until recently its relevance to real-world business decision-making was similar
to most archive material in large corporates – strictly limited.
In
the past, the only way line managers could identify historical trends or build
predictive HR models was to commission the HR systems team to write one-off,
customised programs. Real-time analysis was out of the question self-generated
reports were restricted to run-of-the-mill operational outputs.
Proven
benefits
Today,
however, managers at Deere are empowered to do their own analysis. Using tools
and applications from SAS Institute ñ one of the leading business intelligence
providers in the HR field the company is able to dig down into the database to
support decision-making in a range of areas, from ongoing healthcare and
pension provision to statistical analysis of employee turnover. Although the
company has some way to go in meeting its own goals in areas such as skills
sets and knowledge management, the investment has already yielded tangible
returns.
Deere,
which employs 37,000 people in 50 countries, is one of a growing number of
organisations embracing sophisticated software analytics in their HR
operations. Once the domain of highly-skilled technical staff, these kinds of
business intelligence applications are increasingly being packaged in a way
that makes them accessible to line managers across organisations.
HCM
issues
Although
the current growth rate of HR analytics is relatively small ñ industry analyst
AMR Research puts it at just 5 to 8 per cent – the area is generating a large
amount of interest from software developers and users alike. Historical trends
analysis gleaned from the mountains of employee-related data that HR
departments are required to retain – throws new light on core Human Capital
Management issues such as employee retention and the impact of structured
training on career development.
At
a more sophisticated level, data modelling takes both current data snapshots
and historical pattern analysis and uses them as the basis for forward
predictions, allowing companies to second-guess future skills shortages and
calculate the impact of long-term policy decisions. As leading players in the
HR sector shift their attention to broad business planning analysis, the
central role of human capital management in the wider business environment is
becoming clearer.
New
perspectives
The
strength of these kinds of analytics lies in their ability to interpret data
from numerous angles, pulling together disparate streams of information to form
multi-dimensional views. Effectively, that automates many of the tasks that
previously fell to statisticians, allowing business managers to view data
relationships in ways that were previously impossible.
There
are, however, several caveats. For one thing, success is determined by the
accuracy and cleanliness of the core data, which has to be pulled from
operational systems across the company into one centralised warehouse.
Additionally, while users no longer need statistical backgrounds to interpret
the information, they still need a basic understanding of exactly what data
they are looking at.
Interpretation
debate
According
to Gibson, data accuracy is less of an issue at Deere then it will be in many
other organisations. The company has years of experience in centralising data,
which is primarily pulled from its SAP back-office system, and the only
problems it has encountered are poor network connections to some of its smaller
sites.
Interpretation,
however, is more of an issue. Some managers have failed to differentiate between
data on active employees and past records, drawing conclusions based on wrong
samples or in one case, on factory locations that were closed in the 1980s. In
many cases, those problems can be bypassed by standardising reporting, and
since it first began rolling out the SAS HR Vision application 18 months ago,
the company has started building pre-defined reports into the system covering
inquiries in areas such as average tenure at the company or absenteeism.
Because
the reporting methodologies are standardised centrally, comparisons between
different factories and regions are known to be statistically valid. Beyond
that, dependent on their security access rights, users are free to draw up
their own reports: four hours of training covers the basics for most users,
although the company offers advanced classes for more complex demands. In most
IT rollouts, reporting demands tend to spiral as users begin to get to grips
with the potential their system offers. When that is handled centrally, it
increases the pressure on the core IT team, but devolved reporting allows users
to customise their feeds pretty much at will.
Future
planning
The
HR Vision system now underpins both tactical and strategic HR decisions at an
enterprise level. Corporate healthcare planning, for example, can be built
around current levels of eligibility and future estimates based on demographic
data. Because US plans tend to limit orthodentistry care, for example, the
company can assess the potential needs of both employees and dependents and
then model the total cost implications of raising the level of insurance cap.
Likewise, the company routinely collates data on why employees quit and is now
beginning to identify trends, some of which ñ in areas like daycare provision,
for example ñ can be addressed with practical remedies.
Equally
as important, this kind of HR analytics allows Deere to predict long-term
trends. With 25,000 pensioners and their spouses on its books, most of them
enjoying healthcare benefits, it can extrapolate future costs and model
expectations of how that base will grow. Similarly, it is able to analyse the
impact of future retirements on the company’s overall skills sets, allowing it
to plan successions and ensure managers are being groomed to take over, as well
as ensure that it is aware of potential gaps in specific functions such as
engineering or accounts. In the last year, for example, it has been focusing on
its supplier management skills sets after identifying weaknesses in its
purchasing operation. Thanks to a hiring freeze in the 1980s, the company also
has a bi-modal workforce, with one group of employees loosely falling into the
45-60 year age group, and a second in the 18-32 band. Clearly, the two groups
have different interests and needs, and statistical analysis allows Deere to
tailor programs where necessary.
Compiling
competencies
Sign up to our weekly round-up of HR news and guidance
Receive the Personnel Today Direct e-newsletter every Wednesday
In
the longer-term, meanwhile, Gibson has long been planning to improve the
organisation’s understanding of individual competencies. "That is one
thing HR has not spent time on and should ñ there is a need to identify
competencies," he says. Currently,
the company’s knowledge is restricted to information about employees’ formal
education and data on the roles they have previously held at the company. In a
large-scale project likely to take some six to nine months, however, Gibson’s
vision is to build a list of competencies attached to every position, matched
against more detailed data on employees’ competencies gleaned from information
such as the training courses they have attended.
Gibson
concedes that the company has done no formal return on investment analysis
following the implementation of HR Vision, but the tangible benefits have
become apparent across the line management. In some respects adoption of the
technology is an ongoing process as the company learns more about its own
requirements and the trends it wants to analyse, so the return becomes broader
as take-up extends into new areas. "Once you find out you can do
something, you look at other areas where you can get reports. That is kind of
inherent. You light one match ñ and then you can see the whole room."
