Digital Innovation
T
he hiring process for a lot
of companies is broken:
I’m an occupational
psychologist and I’ve been
blending assessments with technology
my whole career.
The problem has been that the
industry has just taken the same
assessment methodologies, science and
tests, and put them online; there wasn’t
the technology to embed a prediction of
success. But artificial intelligence (AI)
is driving the scoring of video data and
I think we will see a huge shift in the
hiring landscape.
HireVue brought in a team of
data scientists to figure out how
to apply AI to video data, creating
a video intelligence platform. By
combining the AI-driven video data
with predictive, validated industrial-
organisational psychology practices,
we can enhance human decision
making in the hiring process
and deliver higher-quality
talent, faster; historically
you couldn’t have speed and
quality of hire, but now it is
possible to have both.
Essentially, responses
to video interviews are full
of data. The content of the
verbal response, intonation
and emotions portrayed
are a few of the 25,000 data
points we collect, which are
analysed with our proprietary
machine-learning algorithms to
predict future job performance.
Applying AI to video is new, but uses
established psychological research to
predict job success.
Helping Unilever
uncover future talent
For example, Unilever brought us in
to help upgrade the recruiting system
for their future leaders programme,
which recruits recent college graduates
for seven core functions (marketing,
customer development/sales, finance,
HR, supply chain, R&D and IT),
selecting 800 candidates out of 250,000
applicants, from dozens of countries.
When they looked at their paper-
based, expensive and time-consuming
process, they found they had a four-to-
six-month timeline to hire. It included
candidates filling out a 45-minute
application, based on their resume.
These were individually reviewed
by Unilever recruiters, who then
conducted hour-long live phone
interviews with up to 50,000 applicants.
Working with partners, including
HireVue, they transformed their
process with technology; this has
been up and running in more than 55
countries since autumn 2016.
Now, a digital application, integrated
with Unilever’s talent management
solution, pulls information directly
from applicants’ business social media
“Historically,
you couldn’t
have speed
and quality
of hire”
profiles. Using Pymetric’s science-
driven gamification, applicants then
play a series of 12 games, taking around
20 minutes, and receive feedback
within 48 hours. Based on the profile
assessment, selected candidates are
identified for functional areas.
After this, they record a digital
interview on their tablet or
smartphone, involving three short
hypothetical questions and a business
case, based on real-world Unilever
scenarios. This measures applicants’
ability to solve problems; think on the
spot and process new information;
apply knowledge about Unilever and
general business thinking; and take
Unilever’s priorities and objectives
into account.
We analyse the candidate videos
for Unilever, using a combination of
their assessment criteria, machine
learning and AI. Because this solution
is grounded in data comparisons,
it eliminates any unconscious bias
from recruiters. It also uses predictive
modelling in its data analysis to
determine which candidates are likely
to have the most success at Unilever.
All applicants receive feedback at
every stage of the process, and those who
pass the video interview spend a day at
Unilever’s discovery centre, to gain a
preview of working for the company.
If offered a contract, candidates can
sign via their smartphone.
Faster, more
efficient funnel
The end result is that the hiring
timescale has reduced from
six months to six weeks, with
some candidates completing
the process in two weeks.
The completion rate for
applications has increased to
96% from 50% and Unilever
estimates it has saved 50,000
hours and £1m a year.
For candidates, it’s a fast process,
and the experience is friendly. In
their video, they’re telling their story,
answering questions relevant to the
job they would get, while learning
about the role. The net promoter score
feedback we get from candidates is
particularly high.
What’s next for the industry?
Historically, with assessments you had
to add steps to your process of hiring.
That’s all going to blow away. You’re
going to have a video assessment, a
game application assessment or virtual
reality feature (“come try out this job in
a virtual reality scenario” or “take this
video so you can tell us about yourself”)
– and we’re going to use that data and
virtual reality blueprints to generate
simulations and apply AI scoring.
Issue 2 - 2017
55