Pwc - Understanding The Effectiveness of Virtual Reality Soft Skills Training in the Enterprise - Study

1
The Effectiveness of Virtual Reality Soft Skills Training in the Enterprise
Public Report
June 25, 2020
The Effectiveness
of Virtual Reality
Soft Skills Training
in the Enterprise
A Study

2
Table of Contents
1
3
2
4
Why study VR for soft skills training?
05
Which type of HMD should be purchased?
22
Data & HMD security
27
Sanitation considerations
32
Study inspiration and design
10
Employees trained using VR learned faster
36
About the study
06
Deployment model
24
HMD connectivity
29
Software platform
33
Our initial hypothesis
11
Employees trained using VR were more confident
37
Summary of findings
07
How many HMDs should be purchased?
25
HMD management (logical)
31
The course topic
14
Employees trained using VR had a stronger
emotional connection to the content
38
Employees trained using VR were more focused
39
VR can be more cost-effective at scale
40
Validating our original assumptions
42
Other observations
46
Designing the v-learn course
17
Measurement
19
Introduction
04
Technology and
Operations Considerations
20
Study Concept Development
09
Findings
34

3
Comparing Costs to
Determine ROI
48
5
Comparing the costs of key tasks
50
Cost drivers
53
Cost per learner
54
6
8
7
Onboard new VR learners effectively
61
Create compelling learning content for VR
63
Collaborate cross-functionally
64
Reinforce learning through debriefing
65
Create templates to support scaling
66
Build a blended learning curriculum
66
VR is ready to scale in the enterprise
68
Invest in the learning modality suited to
the learning objective
67
Considerations and Insights
59
Acknowledgements
72
Conclusions
69

4
Introduction
SECTION 1

5
Why study VR for
soft skills training?
Digital upskilling is part of our digital journey at PwC to enable our
digital mindset and improve the way we work. We continuously evaluate
new learning technologies and methods to determine how we can train
employees faster, smarter and more cost-effectively.
PwC’s Emerging Technology Group has explored the business value
of virtual reality (VR) for several years. In the training world, VR has
some compelling use cases. Until recently, VR work in the enterprise
has focused on job skills simulation training: flight simulators, safety
procedures, equipment operation and maintenance, etc. Industries using
VR for safety, repair and maintenance simulation training are seeing
improvements in process efficiency, but we wondered whether virtual
reality would be as effective for training leadership, soft skills or other
human-to-human interactions? Does it have advantages over traditional
classroom or e-learning methods?

6
Between February and October 2019, PwC’s Emerging Technology Group,
US Learning and Development Innovation team, supported by Oculus for
Business and Talespin, collaborated to plan, design, build, deploy and
evaluate the results of a soft skills training module. The VR pilot studied
the impact of using VR to train new managers on inclusive leadership,
a specific soft skills course that is part of PwC’s focus on training our
leaders about diversity and inclusion.
Selected employees from a group of new managers in 12 US locations
took the same training in one of the three learning modalities: classroom,
e-learn, or v-learn. Learners were allowed to take the course in only one
of the three modalities. The team surveyed and provided learners a pre-
assessment prior to and a post-assessment immediately after the course.
Users also completed another assessment 30 days later to determine
retention and provide any additional learnings they may have discovered.
Between November 2019 and February 2020, the team analyzed the data,
including evaluating the cost associated with deploying each of the three
courses, to compare the value of each of the three modalities. We also
investigated what would be necessary to effectively deploy VR training at
enterprise scale, with a strong focus on cost-effectiveness and return on
investment (ROI).
About the study
6

7
Summary
of findings
V-learn, using virtual reality to train employees on
various skills, was more effective than classroom
and e-learn training modalities at teaching soft-
skills concepts. The v-learners were up to 275%
more confident to act on what they learned after
training—a 40% improvement over classroom and 35%
improvement over e-learn. V-learners were up to four
times more focused than e-learners. They completed
training on average four times faster than classroom
training and 1.5 times faster than e-learn. V-learners
were 3.75 times more emotionally connected to
the content than classroom learners and 2.3 times more connected than e-learners. On top of
those benefits, v-learn was estimated to be more cost-effective than classroom or e-learning
modalities when delivered at scale.
V-learning will likely be an
accelerator that helps drive
a new age of enterprise
training and education
by delivering a cost-
effective, immersive, and
efficient experience to train
employees on soft skills.

8
We found that VR is ready to deploy at enterprise scale. The team was able to provision, deploy
and manage a large fleet of VR headsets (head-mounted displays, or HMDs) with a very small
team. We determined that while VR training would not replace classroom or e-learn modalities
anytime soon, it should be considered as part of a blended learning curriculum when training
specific types of skills. When you combine classroom, e-learn and v-learn together, you provide
your employees with an industry leading approach.
V-learning will likely be an accelerator that helps drive a new age of enterprise training and
education by delivering a cost-effective, immersive, and efficient experience to train employees
on soft skills.
275%
more confident to act on
what they learned after training
4x
faster than classroom
training on average
4x
more focused
than e-learners
3.75x
more emotionally connectedto the
content than classroom learners
Back to Table of Contents
8

9
Study Concept
Development
SECTION 2

10
Study inspiration
and design
In September 2018 at an Emerging Technology conference in Boston, the team experienced
a VR demonstration that used a virtual human avatar to train a human resource professional
on how to legally and empathically terminate a worker’s job. The “Barry Demo,” created by
Talespin Reality Labs, provided an immersive, engaging and emotional experience in which a
trainee, using his or her own voice, would interact with a virtual human.
In the Talespin VR demonstration, learners play the role of a manager required to fire Barry,
a fellow employee. The demo includes the dialog, heavy emotions and stress normally found
in this type of situation. Even in a digitally created room with a computer-generated character,
users feel emotional and physical responses while communicating to Barry that he is fired.
When interacting with Barry, the user must choose from a number of effective and ineffective
conversations, with different termination pitfalls, in addition to Barry’s verbal and physical
reactions. The resulting experience feels genuine, and the training is not only about what is
said, but also how the user responds to Barry’s reactions.

11
This VR training experience felt like a game changer, but we needed quantifiable data. The
PwC Emerging Technology team knew the technology, but didn’t have the knowledge or
experience to understand the psychology and science behind it. They reached out to PwC’s
Learning and Development group and were introduced to a team of learning scientists who
were looking at VR as a potential training tool. They understood the science behind what
learners were feeling, but didn’t understand the complexities of the technology. It was a perfect
match.
Both teams worked together for more than 10 months to socialize the use case, secure the
additional funding necessary, and then design, build, deploy and analyze the results of the
study. PwC collaborated with Talespin to use their software platform. They also provided
guidance and support on how to build an effective soft-skills training course in VR. Oculus
for Business provided hardware support, as PwC was one of only a few companies that had
access to early versions of their hardware and software management platform. Collaboratively,
we worked as a single team to build a solution that would likely improve a user’s VR learning
experience.
We needed the study to answer two specific questions:
For PwC, this would mean comparing VR training, what we called “v-learn,”
to our two most common training methods: classroom and e-learn.
Is VR soft-skills training more effective than
traditional training methods?
1
Is VR soft-skills training more cost-effective to
deploy than traditional training methods?
2

12
Our initial
hypothesis
Our hypothesis was that training using VR is more effective in achieving
learning outcomes than traditional training methods (classroom or non-
VR digital experiences).
We assumed v-learn would help provide a more flexible remote learning experience
to our employees than traditional classroom methods. Our people are our greatest
asset, and having them travel to attend live training can be impractical and costly.
It would be important to enable employees to have the flexibility to take the training
away from the workspace, as that is easier to fit in with their schedules and saves
time and energy.
Learner flexibility
We assumed users would enjoy v-learn more than traditional learning methods. We
wanted our learners to feel like their time was spent doing something valuable and
enjoyable. If learners felt that the training was effective, it could help build employee
satisfaction and lead to better employee retention.
Employee satisfaction
To determine if soft-skills training in VR is more effective, the outcomes
would focus on the following:

13
We assumed v-learn would provide a more comfortable and less stressful
environment to practice new skills than traditional methods. Practicing skills in VR
offers the ability to try different paths to achieve better results. Learning something
new involves being awkward, making mistakes, course-correcting and trying again.
By giving people the opportunity to repeat a task in a natural way, practice builds
confidence, which improves quality and helps to reduce mistakes.
Comfortable learning environment
We assumed v-learn would help build more confidence in the execution of learning
outcomes than traditional learning methods. Confidence builds employee satisfaction,
leads to better employee retention, improves quality and reduces mistakes.
Confidence building
We assumed v-learn would result in fewer instances of being distracted while
learning (i.e., less multitasking than traditional methods), and distractions result in
lower comprehension and retention.
Improved attention
“Emotion has a substantial influence on the cognitive processes in humans, including
perception, attention, learning, memory, reasoning and problem solving.”[1] We
assumed v-learn would possibly evoke deeper emotional connections than traditional
methods, which significantly improves retention of the information learned. Higher
retention also means learners may require training less often.
Higher information retention

14
The next task was selecting a curriculum to compare the three modalities.
We focused on one of PwC’s top-priority topics: Inclusive Leadership and
Belonging at PwC.
In 2018, PwC mandated that every manager and above take classroom training on inclusive
leadership. This training, designed and built by PwC, focused on how familiarity, comfort and
trust (FCT) influence hiring, staffing and performance reviews. During the training, learners
are asked to understand personal and team member behavior that could potentially be caused
by unconscious bias. The goal of the training is for learners to commit to using only objective
criteria in decision-making.
Some participants evaluated the training as “phenomenal” and reported feeling encouraged by
PwC’s commitment to diversity, equity and inclusion. Others felt the training was awkward and
worried that their comments shared during discussion might be misconstrued, which may have
created reluctance to participate during the course. Even though the classroom training was
mandatory, there are always a significant number of employees who are unable to attend due
to client conflicts, so this training was also available as a webcast session.
We selected this preexisting training curriculum for the comparison study because it was
applicable to a wide audience, and we felt it would be significant if we could make a bigger
impact on this issue through a more effective training method. The US PwC’s Office of Diversity
and Inclusion was also consulted. This enabled us to create an e-learn and v-learn variant of
the curriculum that would be designed to meet the same learning objectives as the classroom
course.
The course
topic

15
Classroom
For the study, the classroom version was identical to the late-
2018 class, in which groups of up to 60 PwC staff members
gathered at their local office for a two-hour session facilitated
by a PwC partner and two directors (or senior managers).
Facilitators were given preparation documents to confirm training
standardization across 40 office locations. The facilitators led
the groups through a series of videos, reflection activities and
discussion topics.
Thirty minutes of the class was dedicated to encouraging
commitment to inclusive leadership behaviors and exploring
how potential unconscious bias might develop and contribute to
non-inclusive leadership behaviors. One hour of the class was
dedicated to identifying inclusive and non-inclusive behaviors in specific PwC situations.
The facilitator would play a video demonstrating non-inclusive leadership behaviors in
team discussions around hiring decisions, project staffing and project performance credit.
After playing the video, facilitators asked learners to identify undesirable behaviors and
highlight which inclusive behaviors should have been employed instead. The goal was to
help learners self-identify where they could better employ inclusive leadership behaviors.
E-learn
The online module allowed learners to take the training at
any time by using their laptops. The training was 45 minutes
and guided learners through the same activities, videos and
reflections found in the classroom version. Learners participated
by watching, selecting the next button and answering multiple
choice questions. The largest challenge with the e-learn training
was the inability to control learner attention. They could easily be
distracted and focus elsewhere during the “watch” portions of the
training.

16
V-learn
We needed to design and develop the new v-learn course. We leveraged the existing
classroom and e-learn content, but needed to employ it in a significantly more interactive
simulation. The classroom and e-learn course experiences were linear: A video was shown,
the learners asked some questions, then the next scenario was presented. A VR experience is
achieved by placing a VR head-mounted display (HMD) over the eyes of a learner, replacing
the view of their physical surroundings with a digitally generated display, which leaves the
learner feeling he or she is in a new environment.
The course design could have used a virtual classroom, where learners could passively
watch and listen to the classroom presentation. However, we determined this linear approach
would not leverage any advantages of the VR modality. We hypothesized that placing the
learner directly in the scenarios covered in the curriculum and giving them the ability to act
as they might in real life would be more rewarding for them. The same three scenarios would
be presented (who to hire, who to staff and who got the performance differentiator) in virtual
reality, but instead of watching a video, the v-learner would be one of the team members
participating in the discussion.
In each of the scenarios, learners would be presented with situations and questions where
they needed to act and make a choice. Not all of the choices provided were correct, so a
learner could make mistakes. This wasn’t necessary in the classroom or e-learn training, but
it was critical for the v-learn. The team would need to create a branched narrative for each of
the possibilities because if the learner chose incorrectly, the interaction on the screen would
have to react to the learner’s choice. At the end of a scenario, the learner would be provided
feedback on the decisions he or she made and would be given the opportunity to try again.
We hypothesized that placing
the learner directly in the
scenarios covered in the
curriculum and giving them
the ability to act as they might
in real life would be more
rewarding for them.

17
Designing the
v-learn course
Unlike the classroom and e-learn courses, which were built from existing
content, the v-learn course took almost three months to design.
We understood how to create a branched narrative, but we contracted with Talespin to help
guide us through the design process and the VR development.
At the same time, we needed to understand how users would react to the overall VR
experience, so we explored different approaches to help enable a positive learner experience.
This impacted the design of the course, so close contact was maintained between the project
workstreams.
In the final v-learn course, learners find themselves in a virtual office (modeled after
actual PwC offices). They are speaking to a group of animated virtual characters that are
representative of PwC colleagues sitting at the conference table or are attending via a
conference call. Learners play themselves as they carry out professional conversations with
characters to determine whom to hire, whom to staff, and which ones to credit for success.
In each of the three scenarios, the learner speaks directly to virtual characters, reading
dialogue aloud from a list of options. The head-mounted display microphone collects the
spoken choice, and a voice-to-text service matches spoken words to the option.

18
The program then records the learner’s choice. Virtual human
characters then react to what the learner says, and the
conversation progresses with more options. Each option allows
learners to observe and practice the inclusive leadership
behaviors identified as key learning points of the curriculum.
During the course of the conversations, learners may state that
another character is showing bias for a particular candidate, ask
questions to collect information or share their own candidate
recommendations. They must read their chosen dialog aloud,
versus saying “A,” “B,” “C” or “D.” Having the learner say, “I think
Marie should be given the performance differentiator,” gave the
simulation a more realistic and conversational experience. It also,
subconsciously, reinforced the desired behavior.
Each module was five-to-seven minutes long, depending on the learner’s choices and speed.
After each conversation, learners could review their dialogue choices. Weak and neutral
options selected by the user are highlighted, and wording reflective of more inclusive behaviors
is presented, along with explanations as to why the recommended dialogue is more inclusive.
Learners are allowed to repeat and practice each scenario as many times as desired.
Photo courtesy of Talespin

19
Measurement
In late 2018, the firm had already trained more than 13,000 leaders on inclusive
leadership. The v-learn training was scheduled to begin in August 2019, so we
needed to find enough leaders to participate in the study.
Each year at PwC, more than 700 senior associates are promoted to manager. We
determined that if we combined those associates with new senior experienced hires
and leaders who missed both the classroom and e-learn course, the total potential
study participants could be greater than 1,600. We identified our test group and
divided them into three groups. Each group was asked to take one of the following
training sessions: classroom, e-learn or v-learn.
We also conducted experience surveys and performed face-to-face and phone interviews to
capture the learner experience. Additionally, during the v-learn creation process, the team
engaged groups of test users to improve the effectiveness of the learning course and gauge
the sentiment about the virtual reality experience. Iterative improvements were made to
the course throughout the summer based on UX testers’ feedback. These findings helped
enhance our understanding of population readiness for VR.
Learners were given identical assessments and surveys before
and after completing their course. Assessments included:
A pre-assessment, which evaluated the ability to make inclusive
leadership decisions before taking any course.
A post-assessment, which evaluated the ability to make inclusive
leadership decisions after taking their assigned course.
A retention assessment, which evaluated the ability to make inclusive
leadership decisions 30 days after taking their assigned course.
19

20
Technology and Operations
Considerations
SECTION 3

21
Technology and
operations
considerations
It is one thing to create virtual reality software simulations, but it’s a completely different
challenge to scale the deployment of head-mounted displays (HMDs) in the enterprise. Other
than content development and/or licensing of content, HMDs and their operations require the
greatest investment when building VR capabilities for the enterprise.
Our initial estimate of the number of PwC staff who would be eligible to participate in the study
was greater than 1,600, and training would take place in 12 PwC office locations throughout
the US.
We needed to purchase HMDs, but we would also need to:
Manage the HMDs
Manage how software was loaded on the HMDs
Confirm the HMDs were secured both physically and logically
Determine where in the various offices the experience would take place
Confirm that the data collected was secure and met PwC’s guidelines for
employee privacy.
To address the location requirements and other contributing variables, it was necessary to get
answers to several questions: Which HMDs should we purchase? How many should we order?
How will they be managed? How will we secure them physically and logically? And how will we
protect data collected from each user?

22
Which type
of HMD should
be purchased?
There are two categories of VR HMDs: tethered and wireless.
Tethered HMDs
Tethered HMDs require the HMD to be connected to a high-performance computer via cable(s).
This is the traditional way HMDs are deployed for VR as it provides better performance. These
setups could also require lighthouses (tracking sensors that help the computer determine
where the user is positioned inside a simulation) and spaces as large as 10m x 10m to support
6 Degrees of Freedom (6DoF) experiences.
However, this category requires specialized IT support, large spaces not often found in
organizations, and greater investments. It is not uncommon for tethered HMDs ecosystems to
cost approximately $5,000 each (including the VR HMD, positional tracking, controllers and a
high-performance computer with the appropriate high-performance video cards).
Wireless HMDs
Wireless HMDs are self-contained. Everything necessary — computing power, positional
tracking, network connectivity and controllers — is wireless. Since these units are wireless,
they are easily transported from environment to environment; there is no need for VR users
to be concerned with tripping over wires while in the experience; and they do not require any

23
special IT support services. However, immersive hyper-realistic experiences are limited due
to lower graphics and CPU performance on these HMDs, and in-headset duration is limited
to a few hours, as they are powered by rechargeable batteries. Wireless HMDs require less
investment than tethered HMDs, with prices ranging from $199 to $1,499.
We had extensive experience deploying tethered solutions, but due to how much technical
support they required, we selected wireless HMDs to support the VR study.
The Oculus Quest
For this study, PwC selected the Oculus Quest, a wireless 6DoF HMD that provided a
majority of the features we needed. It was wireless, tracking was handled by the headset
itself and there was no need to set up external tracking sensors (lighthouses). It also had
a built-in microphone so we could use voice control within the training application in lieu of
the controllers (if we chose to do so), it was comfortable to wear once adjusted properly and
battery life was adequate to meet our needs.
Other HMDs had similar features, but Oculus also had the Oculus for Business solution, which
let our team manage our fleet of HMDs remotely. This software helped enable our team to turn
on and off specific features, manage access, monitor the health of the device and control which
applications were loaded on the Oculus Quests.

24
Headset deployment model
To execute the study, we needed to understand how many HMDs would
be needed. We already knew that the training would take place in 12
different PwC office locations, plus we were asked to keep the training
window down to six weeks. We knew that office space was limited, and
we had a limited budget to purchase HMDs. We considered the following
deployment models:
1:1 means one headset to one user, so every user would receive
his or her own headset. The 1:1 model was rejected almost
immediately as it would be cost-prohibitive.
1:1 model
1:Many is when one headset is shared with many users.
The 1:Many model made more sense, as we could place a limited
number of HMDs in each location and enable trainees to check them
out just prior to attending the training.
Selecting the 1:Many model required the team to create a way to
schedule HMDs to align with user scheduling to confirm there was
an HMD for a specific user at a specific time.
1:Many
24

25
How many HMDs
should be purchased?
To determine how many HMDs to purchase,
we had to capture the following variables:
ID VARIABLE VALUE
A The number of training cities (PwC office locations) 12
B
The number of cubes/rooms/offices per office location that we could lock down
for the study
4
C Total number of eligible trainees 1,600
D Total number of eligible trainees by location Varies
E
The total duration of the course (from the time the trainee arrived until the time
he or she returned with the headset)
1 Hour (rounded)
F
How many spare HMDs would be held as backup in case there was a hardware
and/or software failure (mostly likely due to an HMD not being charged)
20%
G
How many days will the training be made available? The requirements to train
10,000 employees in 30 days are significantly different from needing to train
10,000 employees over 12 months.
20
H How many hours per day will the training locations be available?
4-8 (varies by day of
the week)

26
In addition, it was important to understand that PwC partners and staff normally work at client
sites and are only in the office on Monday or Friday. However, some teams work from a PwC of-
fice five days a week. Finally, participation in the v-learn course was voluntary, so we assumed
we would not get all eligible trainees to attend.
The simple calculation should have been one headset per training area, times the number of
locations per city, times the number of spares. Or, if using the table above, from the ID column:
B * A * (1+F). Using the variables above (12 x 4 x 120%), 57.6 HMDs would be needed. Howev-
er, after evaluating the situation, we erred on the positive side and determined we should ship
an average of six HMDs to each location. This provided the total number of HMDs (72) that
would need to be purchased.
It is critical to remember the three “hows”: How many locations, how many users and how fast
do you need to train users are the key variables in determining how many HMDs you need to
purchase.
We have deployed a 1:Many model, and as of February 2020, we are supporting 250 HMDs
across 25 locations. We could host a one-hour training simulation for up to 2,000 PwC partners
and staff a day, and train the 55,000 partners and staff in the US firm in less than five weeks.
For PwC, the key variable is the number of locations. We expect to increase the inventory over
time as more content is onboarded.
It is critical to remember the
three “hows”: How many
locations, how many users and
how fast do you need to train
users are the key variables in
determining how many HMDs
you need to purchase.

27
Data and
HMD security
Data security
HMDs can generate a lot of data. If the HMDs are not secured and managed, data can po-
tentially be tied to a specific user. For example, data might be simple, such as user ID, time
of session or even the location where the headset was used. Data collected can also be quite
extensive, such as gaze heatmaps (what was the user looking at, for how long, and what was
the user doing before and after that gaze). It could also include the time to respond to specif-
ic stimuli (questions, interactions, voice biometrics) or even the user’s emotive state (happy,
anxious, fearful, etc.).
When considering data security on a headset, it is important to understand that there are
normally two types of data security that need to be addressed: the security of the underlying
operating system that is running on the hardware, and the security of the data collected by an
application running on top of that operating system (i.e., data collected from inside the appli-
cation). For the study, the Oculus Quest HMDs were running a more secure enterprise build of
their consumer version of the headset,and their management software (Oculus for Business)
helped enable the team to manage the headset features in a secure way.

If the HMDs have access to the corporate network and enterprise data shares, it will be import-
ant to also have logical security. It could be as simple as a PIN code or enabling an authorized
user’s mobile device in close proximity to the headset. It could also be as robust as LDAP-
based user ID and password. Future HMDs may also have built-in biometric sensors (retina
scan or fingerprint) to validate users, as entering in user IDs and passwords can be cumber-
some in virtual reality.

28
For the study, any trainee could put on a headset and have access to it. However, only the
v-learn course was loaded on the Oculus Quest, and users were limited in what they could
do. We placed the Oculus Quests in “kiosk mode,” which means when a user put on the HMD,
the v-learn course would launch automatically. Controllers are required to exit the kiosk mode,
but since participants used voice interaction, controllers were not provided when a participant
checked out the HMD.
Each study participant was provided with a unique PIN to access the v-learn course when
logging into the experience. No user identifiable information was passed to Talespin. The only
thing Talespin knew was that a specific PIN accessed the course at a certain time from a spe-
cific HMD. This was a low-cost yet secure way to help support high data security.
It’s important that data collected be encrypted, secure and managed appropriately. For the
study, we worked closely with Talespin to confirm the data collected was encrypted end to end.
Data was also stored encrypted, and even if that data were decrypted, no user identifiable
information was passed outside of PwC’s firewall. At the end of the training period, Talespin se-
curely sent PwC the user data, which we then decrypted and associated individual users with
their unique PINs prior to analyzing the data.
HMD security
Our initial purchase was for 100 HMDs. 76 were earmarked for the study, and the balance was
for other uses. These HMDs needed to be received, provisioned, configured, placed into inven-
tory, and stored in multiple PwC office locations throughout the US. HMDs should be stored like
any other electronic equipment (low humidity, constant temperature, etc.). Like any other asset
of value you also should be concerned about the HMDs physical security.
For the purposes of this study, it was necessary to create several different mechanisms to
manage the physical inventory and stay organized — spreadsheets, feedback forms, shipment
logs, asset tags, scheduling software, etc. The hand controllers were paired to specific head-
sets and it was important to not mix them up as there is no easy way to quickly identify which
controller belonged to which HMD. Each HMD kit consisted of the headset, two hand control-
lers, power supply, USB-C cable, headphones, and a headphone cable. The hand controllers
were paired to specific headsets and it was important to not mix them up as there is no easy
way to quickly identify which controller belonged to which HMD. Each Oculus Quest has a
unique three letter designator stored within the serial number which also aligned to the Oculus
For Business device manager software which we used to manage the HMDs. We created asset
tags that aligned to these unique IDs and tagged each of the HMD kits. It was not the most
gratifying part of this study, but it was critical to the success of the overall effort. One part-time
team member was able to manage 800 individual assets both physically and electronically
without losing or misplacing a single asset.

29
HMD
connectivity
Local
Local does not require a network to run. Specific content is loaded on the wireless headset or,
in the case of a tethered headset, it runs on the connected computer. It is common to preload
360VR content this way due to the size of the experience. Network bandwidth requirements
are easily managed, as the content is downloaded to the headset (wired or wirelessly) before
it is accessed.created asset tags that aligned to these unique IDs and tagged each of the
HMD kits. It was not the most gratifying part of this study, but it was critical to the success of
the overall effort. One part-time team member was able to manage 800 individual assets both
physically and electronically without losing or misplacing a single asset.
When discussing VR at enterprise scale, we are making the assumption
that the enterprise HMDs are wireless. Tethered HMDs require additional
computer and technology support, cost more money to purchase, and
are more difficult to manage. Future HMDs may have cellular connectivity
(LTE or 5G), but today WiFi is the most common way to get HMDs on
the network. Bandwidth requirements are determined by where the VR
application (content) is “running”: local, network and network+.

30
Network
Network requires an active network to support the application running on the headset. This
might simply be a user login or API to support voice interactions, or more complicated use cas-
es to support Q&A, branched narrative or in-headset support. Network bandwidth requirements
need to be addressed at 100+ HMDs all on the same subnet.
Each headset requires approximately 52kbps to maintain real-time interactivity. A large cor-
porate network can easily manage this, unless you are supporting an event where the entire
inventory of HMDs are being used at the same time and in the same physical location (like a
conference or large meeting). For example, if you are supporting 100 HMDs for a conference
and they will be used simultaneously, expect to require a minimum of 52Mbps connectivity.
Network+
Network+ is a variation of the network above, but, in addition, the content is loaded in real time
from the cloud. A simple example might be WebVR content: Very little content is loaded on the
headset, and most of the heavy lifting is provided by the cloud. Network+ is very dependent on
network capacity. As an example, an interactive 360VR video streaming at an uncompressed
720P requires upwards of 50Mbps for each headset. There are many vendors that can com-
press this in real time, so depending on the implementation, performance may vary.
We implemented a network approach for the study, as the content was stored locally but
required network access for voice interaction and for the collection of course data. It was
necessary to work with network security to prevent the blocking of specific API calls — a task
that took more than three months to implement. This was not necessary for the study, but we
wanted to enable our network infrastructure to support this modality globally.

31
The most critical aspect of deploying VR at scale in a shared 1:Many
model is how you will manage the inventory of HMDs. Maintaining
software patches; monitoring power; making sure HMDs are secure;
loading, unloading and distributing content; and making sure users do
not install unauthorized applications or go to unapproved websites is a
labor-intensive activity — especially if the HMDs are in multiple locations
and/or countries. Just like mobile phones in the enterprise, VR HMDs
need some kind of mobile device management (MDM) to manage the
experience.
If you have only five or 10 HMDs, management can be manual, but if you
have hundreds or thousands of HMDs, you will need to purchase and
install this technology in your enterprise. Traditional MDM providers are
starting to introduce this capability for virtual reality HMDs, and there are
some standalone solutions provided by third parties. Even the headset
manufacturers are providing different levels of management. Every
organization is different, so research and due diligence are necessary
to see which solution best fits your enterprise. Last year, there were no
solutions available, but today there are many.
One item of note: The installation of security certificates on some of the
HMDs is not yet possible. If you are using a certificate-based security
approach for your enterprise and mobile phones, you may have to
address MDM capabilities outside your standard platform.
HMD management (logical)
31

32
Sanitation
considerations
In the 1:Many distribution model, it’s important to make sure HMDs are cleaned prior to
sharing the headset with the next user. Gupta, Wantland and Klein (1996) suggest that much
of the peripheral equipment used in VR are potential fomites (“a harmless object that is able to
harbor pathogenic organisms and may serve as an agent for the transmission of infections”)[2].
The study discusses how airborne pathogens and skin flora may thrive in environments similar
to those of HMDs and hand controller devices. HMDs are enclosed, and after being immersed
in a simulation for a while, they can become warm and cause the user to sweat, which can be
absorbed by the padding that sits between a user’s face and the HMD. It’s easy to see how
germs can be transferred to another user. Before sharing our HMDs, we sanitized them before
each use.
The team used two different approaches to sanitization. We followed Oculus’s approach
to sanitization[3] as it was simple and low cost. We also used a device that was designed
specifically to sanitize HMDs by bathing it in UVC light. UVC light is used to sanitize medical
equipment and should not be confused with UV light. UV Light operates in a slightly higher
band and does not destroy harmful viruses and other contaminants. Radiation in the UVC
range of 250-280nm destroys bacteria, viruses and other microbes. UVC light penetrates
the cells of microorganisms and disrupts the structure of the DNA molecules rendering them
inactive and no longer harmful. The process was quick (one minute) and left no chemical smell.
Using both sanitization processes supported our 1: Many distribution model. Each HMD was
sanitized after each use prior to letting another user check it out.

33
We employed Talespin’s CoPilot and Runway software platforms to host
the VR portion of the study. Talespin also provided the development and
design resources necessary to support our effort to create the v-learn
course.
Software platform
CoPilot uses artificial intelligence and virtual humans to simulate
realistic conversations in virtual reality, enabling learners to develop
and practice critical soft skills in a safe and controlled environment.
CoPilot
Runway is where performance data is accessed and analyzed. It is
supported by a native cloud infrastructure, robust administrative
functions (including tracking, skills mapping, assessment,
analytics and distribution), and a scalable and secure solution that
is designed from the ground up to support native v-learning.
Runway
33

34
Findings
SECTION 4

35
Findings
The findings below are based on the results of this specific pilot, which was designed for
PwC employees, and results will vary with other programs or populations. Most companies
should see a positive impact, but how much would be dependent on the factors discussed in
each section.
Findings at a glance:
Employees trained using VR completed training faster
Employees trained using VR were more confident
Employees trained using VR had a stronger emotional
connection to the content
Employees trained using VR were more focused

36
We were able to train employees up to four times faster in VR than in the classroom
and 1.5 times faster than e-learn. We were surprised when our digital natives initially
struggled with the VR headsets, but, once they were comfortable, most found them easy
to use. Even when we accounted for the additional time needed to onboard new users
in VR headsets for the first time (on average 10 extra minutes), we found it is still three
times faster than classroom and 1.15x faster than e-learn.
E-learn
Classroom
VR
45 minutes
2 hours
29 minutes
Time to
complete
training
Source: PwC VR Soft Skills Training Efficacy Study, 2020
Employees trained using VR
completed training faster
We were able to train employees up to four times faster in VR
than in the classroom and 1.5 times faster than e-learn.
36

37
Users trained with VR were up to 275% more confident to act on what they learned after
training—a 40% improvement over classroom and 35% improvement over e-learn. They
were almost two and a half times more confident in discussing diversity and inclusion
issues after taking the training in virtual reality and nearly three times more confident in
acting on diversity and inclusion issues after the VR training. That’s significant because,
when it comes to soft skills, confidence is a key driver of success. Believing in themselves
and having confidence helps learners connect better with others, while also feeling more
satisfied with the time spent training.
Employees trained using VR were
more confident
Users trained with VR were up to 275% more
confident to act on what they learned after
training—a 40% improvement over classroom
and 35% improvement over e-learn.
Improvement in confidence
discussing issues of diversity and
inclusion after the training:
Classroom
E-learn
VR
Improvement in confidence
acting on issues of diversity and
inclusion after the training:
Classroom
E-learn
VR

38
V-learners felt 3.75 times more emotionally connected to the content than classroom
learners and 2.3 times more connected than e-learners. People connect, understand and
remember things more deeply when their emotions are involved. Consider the emotional
impact of viewing a photo of a wildfire, compared with reading a fact-filled analysis of
climate change and the wildfire crisis, and it’s easier to understand how strong emotional
connections could lead to more positive outcomes.
Employees trained using VR had a
stronger emotional connection to
the content
V-learners felt 3.75 times more
emotionally connected to the content
than classroom learners and 2.3 times
more connected than e-learners.
Average emotional
connection felt to
learning content
38

39
When you think about how VR works, it’s easy to understand why users would be less
distracted with this technology: The simulations and immersive experience command their
vision and attention. In our study, VR-trained learners were up to 4 times more focused during
training than their e-learning peers and 1.5 times more focused than their classroom peers.
When learners are immersed in a VR experience, they tend to get more out of the training and
will likely have better outcomes
Employees trained using VR were
more focused
VR-trained learners were up to 4 times more
focused during training than their e-learning
peers and 1.5 times more focused than their
classroom peers.
How many times were you
multitasking or distracted during
this experience?
How many minutes do you estimate
it took to get back on task?
39

40
When PwC decided to study the efficacy of VR training, we also chose to compare the cost of
developing each of the learning modalities. The most effective way to do that was to design
and build a course that could be replicated on each of the three training modalities. We found
that, while VR requires more upfront investment to build and deploy than classroom or e-learn
training, when delivered to enough learners, VR training can be more cost-effective at scale.
When we evaluated the price differences between building a VR course versus a classroom or
e-learning course, we discovered that the VR custom course cost 47% more than the classroom
course and 48% more than the e-learning course. This makes sense when you account for the
cost of creating the content, but also the cost of developing the VR experience. That may require
3D artists and software developers, which you don’t need when creating traditional classroom or
e-learning content.
However, when you consider the return on investment (ROI), a new picture emerges. VR training
takes less time than classroom and e-learning courses. So, if you take into account the cost
of the employees’ time, VR becomes a better investment at some point. And that factor just
considers the time saved. The graph below shows PwC’s total cost savings and where we broke
even on training our learner population.
Training modality cost per learner

41
Even if you include onboarding times, we estimated we would be able to achieve parity in costs
vs. classroom costs at only 375 learners. Based on these variables, we also calculated at 3,000
learners, VR costs fall to 52% less than classroom; at 6,000 learners, 58% less and at 10,000
learners; it’s 64% less. In every company, an employee’s time has value, so the more time
workers save, the faster you can achieve positive ROI.
Comparing VR training ROI to e-learn requires more learners than the classroom, as e-learn
does not require a facilitator/trainer to lead each class. We estimated we would be able to
achieve parity in costs for v-learn vs. e-learn costs at 1,950 learners. Based on these variables,
we also calculated at 3,000 learners, VR costs fall to 8% less; at 6,000 learners, 20% less and
at 10,000 learners; it’s 26% less. Just like with classroom training, the faster an employee can
return to work, the faster you can achieve a positive ROI.
52%
more cost-effective than classroom
training at 3,000 learners
58%
64%

42
Validating our
original assumptions
Our initial hypothesis was, “Training using VR is more effective
in achieving learning outcomes than traditional training methods
(classroom or non-VR digital experiences).” We felt ROI would come from
several areas. The study validated a majority of our assumptions:
Post-survey results indicated that 78% of all VR participants preferred
v-learning to more traditional modalities.
We assumed users would enjoy v-learning more than traditional learning methods.
We wanted our learners to feel like their time was spent doing something valuable
and enjoyable. If learners felt that the training was effective, it could build employee
satisfaction and lead to better employee retention. Post-survey results indicated
that 78% of all VR participants preferred v-learning to more traditional modalities.
In fact, 91% of the participants who took the classroom or e-learn course and then
were given the option to take the v-learn course preferred the v-learn over both the
classroom and e-learn course.
Employee satisfaction

43
PwC was not able to test this assumption for the following reasons:
Learner flexibility
It was cost-prohibitive to purchase an HMD for each learner, enabling them to
take the device home and take the v-learning training based on their availability.
We decided early in the planning phase that we would deploy a 1:Many model
(sharing HMDs) which would require them to take the v-learn training in a
PwC office.
Early user onboarding testing sessions also uncovered that learners had
preconceived notions of how to use virtual reality and demonstrated an inability to
successfully put on and use VR gear following written or audio-visual instruction.
This required the team to rethink the learning experience, and, after three UX
iterations, the team came up with a highly effective process with onsite personnel
providing instruction and support.
Each of the HMDs were managed: what software was loaded, WiFi configuration,
guardian settings, etc. If learners took the HMD home, the learner would require
administrative rights to the headset to enter in their WiFi credentials. Although
the process to configure network settings was not difficult, the team felt it could
cause the learner to spend more time dealing with the technology than taking the
training course.
The study plan stated that we wanted to train the learners in four to five weeks.
If learners could check out the HMDs, the team could have had a difficult time
retrieving them, which would have delayed the next learner from starting.
In the end, the team decided it would be easier to have learners trained in a PwC
office.

44
The overall simulation required learners to practice behaviors they were taught. The
experience also enabled learners to retry (practice) a module to see if they could
improve their outcome, as there were several different paths to success (and failure).
Consequently, 24% of the learners chose to do an optional retry of one or more of
the learning modules. Survey results indicated that 97% of v-learners felt time spent
on practice was appropriate. We saw immersive practice, with dynamic feedback and
optional retries, improved learner confidence. VR learners were 40% more confident
than classroom learners and 35% more confident than e-learners in employing the
skills they learned during training.
We quickly discovered retention scores were inconclusive, as the delta between pre-
and post-assessments in each modality was not significant. Indeed, the assessment
team underestimated the previous knowledge experience our test population had
on the diversity and inclusion topic. In hindsight, we should have selected a topic
that was not already in our curriculum or selected a different test group that had not
already been immersed in similar training.
We did test several associates and senior associates (1-2 levels below manager) who
had not been exposed to diversity and inclusion behaviors in leadership functions,
and the results demonstrated significant retention scores and learning in a v-learn
course. However, these were not approved study participants, therefore their data
was not part of our assessment or these results.
Despite our inability to demonstrate learning via assessment, 75% of those surveyed
reported experiencing a wake-up call moment about the inclusivity of their behaviors.
This indicates that learning did take place, even though our assessments were
unable to demonstrate it.
Comfortable learning environment
Higher information retention

45
As identified in our key findings, VR-trained learners were up to four times less
distracted during training than their e-learning peers and 1.5 times less distracted
than their classroom peers. This was self-reported, and the team did not use any
passive technology to observe this attribute. Based on experience and months of
observation, the team actually felt the self-reported statistic was lower than what we
observed. However, the statistic was significant and should result in higher learner
comprehension and retention.
VR-learners demonstrated significantly higher confidence compared to e-learn and
classroom learners. They were almost 50% more confident than their classroom
peers when discussing the topics learned during training. That’s significant because,
when it comes to soft skills, confidence is a key driver of success. Believing in
themselves and having confidence can help learners connect better with others.
Confidence builds employee satisfaction, which can lead to better employee
retention, and help improve work quality and reduce mistakes.
Improved attention
Confidence building

46
Other observations
Because it was a realistic simulation where they played themselves, v-learners reported
making decisions based on what they would have done in real life. Seventy-five percent of
learners said the training helped them identify moments from their past when they were not
as inclusive as they had previously believed. Data showed a measurable sentiment trend, with
users reporting feeling inspired and confident after completing their final module of feedback in
the VR headset.
Classroom learning provided an emotional, authentic and interpersonal environment for
learning. The success of classroom learning depends heavily on the skills of the instructor
and the contributions of class members, so learners can take the same course and have a
completely different experience because of the varying dynamics between the instructor and
fellow classroom learners. Good classroom instruction is often expensive because of the cost
of knowledgeable instructors.
E-learning offers convenience, scalability and flexibility for the learner. If the e-learning is
cloud-based and the learner can take the course from anywhere, at any time, on a computer
they already have, then it is possible to quickly roll out training at scale. However, e-learners
often multi-task and are highly distracted. They may feel less emotional connection with the
content, which is believed to lead to lower retention of the material. In our study, they reported
less satisfaction with the opportunities to practice skills in the online setting.
75% of learners said the
training helped them identify
moments from their past when
they were not as inclusive as
they had previously believed.

47
VR-based learning can yield higher confidence and improved ability to apply the learning on
the job because of the ability to practice in an immersive, low-stress environment. It is more
effective than e-learning because of the immersive nature of the practice experience. “It feels
like you are really there,” VR users often say. “It demanded my attention and my involvement, I
couldn’t just sit back and listen, I actually had to do the task.”
VR practice is likely more effective than classroom practice because often there are not enough
subject matter specialists to observe and provide feedback on each person’s performance in a
classroom setting. This means that during classroom practice, the learner may not be applying
the skills taught and, in fact, the practice could possibly reinforce incorrect behavior because of
the lack of feedback.
To address this, VR simulation is designed by asking the subject matter specialist to help
create a scenario that offers the learner the opportunity to make choices that apply the skills
being taught, along with options that depict the most frequent mistakes. In VR, we include
dynamic feedback so that when a learner makes a choice in simulation, the strong behavior
choices are called out, and the mistakes are met with corrective feedback, reasoning and an
opportunity to retry.

48
Comparing Costs
to Determine ROI
SECTION 5

49
Comparing costs
to determine ROI
We are often asked: “How much does VR really cost?” All three modalities (classroom, e-learn,
and v-learn) each require a course to be planned, designed, developed and deployed. As part
of this pilot, we sought to compare the costs and ROI of deploying the same learning content
in the three learning modalities. We explore our findings in this section.

50
Comparing the
costs of key tasks
TASK CLASSROOM E-LEARN V-LEARN NOTES
Course design $ $ $$
VR requires more options to be
considered; “happy” and excep-
tion paths must each be flushed
out to help provide a dynamic
experience.
Research $ $ $ Equivalent
Script $ $ $$
VR requires additional scripting to
account for the exception paths.
Talent acquisition $$ $$ $
Assumption that videos will be
created requires more actors to
be used.
Creative/video $$ $$ $$$
VR requires additional creative
assets to be created (3D art).
The table below highlights some of the key tasks and comparison cost
ranking between the three modalities as they relate to common tasks.
The more $$ symbols there are, the more that task costs to execute.

51
TASK CLASSROOM E-LEARN V-LEARN NOTES
Development $$ $$ $$$$$
VR requires the solution to
be coded, whereas e-learn is
normally built upon a learning
platform.
Editing/assembly $ $$ $$$
VR requires additional assem-
bly work to account for the
exception paths.
Testing $ $$ $$$
e-Learn and v-learn require
additional testing to identify
software anomalies
Legal review $ $ $ Equivalent
Logistics setup $ $ $$
VR requires additional equip-
ment (such as the HMD, head-
phones, etc.) to be managed
that is not part of a standard
employee equipment package.
Support $ $ $$ Equivalent
Facilitation $$$$ - $
Classroom training requires
facilitation that is not scalable.
The more people you train, the
more facilitators you need.
Total cost
comparison
17 16 25
VR courses in general will be
more expensive to build than
classroom or e-learn courses.
This table aligns closely with the actual costs tracked to execute the inclusive leadership
course in each of the three modalities. Classroom and e-learn were very close (within 2%) in
overall costs to develop, and the v-learn course cost 47% more than the classroom course to
build. But this is only the creation of the course. Overall cost ratios change when you begin
accounting for the deployment, execution and value of the course.

52
We monitored more than 25 variables and had more than 28 different calculations to determine
the overall return on investment. Since this was not a long-term study and it was focusing on
soft skills training, it was difficult to determine actual value savings for things like higher con-
fidence levels, fewer distractions and greater emotional connection, etc. We could infer value,
but we could not define it.
This left the team with only a few variables that had an impact of
the overall ROI:
The total cost to build a course for a specific modality.
The cost to facilitate, manage and operate a course for each specific modality.
The number of people that are trained.
The fully loaded cost per hour of each employee trained including revenue loss
(at PwC, we use the term RSR).
The average time it takes for an employee to be trained for each specific modality.
If a modality takes less time to train someone, that employee can go back to work
sooner, and less revenue is lost.

53
Cost drivers
The time spent learning
The highest cost driver is the duration employees spend on a specific training.
In our study, VR took one-fourth the time of classroom training and two-thirds
the time of an e-learn. As an example, if an employee’s fully loaded rate is
$100/hour, for our study, employee cost for classroom learners was $200, $75
for e-learners and $50 for employees that were trained with VR. At 20,000
employees, the costs to train within each modality become easy to differentiate:
In this example, it would be $4M for classroom training, $1.5M for e-learn, and
$1M for v-learn.
The more employees you train in VR, the more time you save, therefore the
less money you spend on time to train your employees.
Course facilitation as a cost driver
The second largest cost driver is facilitation. For classroom training, this is not
a scalable proposition. Each class requires a facilitator, while e-learning does
not require any facilitation, and (in our study) the v-learn modality had a part-
time facilitator in each location. As an example, the cost to facilitate training
13,000 leaders in a classroom at PwC is eight times more expensive than the
cost to facilitate a v-learn course for the same number of people. e-Learn has
no facilitation costs associated with it.
Hardware and software considerations
e-Learn and v-learn require additional hardware and software that must be
included in the overall cost. However, for our calculations, we amortized the
hardware over a period of 12 months, as we assumed the hardware could be
used for other courses or purposes.

54
Cost per
learner
For the purposes of this study, we created a value curve to see which
modality was the most cost-effective. The curve changes as you add more
employees to be trained.
Training modality cost per learner

55
The summary of the curve is simple: There comes a time when v-learn becomes more cost -ef-
fective than classroom and e-learn modalities. The fewer employees you train, the more expen-
sive the cost per learner. The more you train, the less it costs per learner. However, it is not a
linear curve. As an example, if you train 100 employees, you have to recover the entire cost to
create and deploy the course across those 100 employees.
It is important to understand the numbers used in these models were for a custom built course.
ROI will be different for every company. If the fully loaded cost per employee is $35, the num-
ber of learners you require to reach ROI is significantly higher. If you can create a two-hour
training with less investment, then your overall costs are lower, and if you purchase an off-the-
shelf course, then the cost per learner will drop across all three modalities.
Overall, there were a few variables that really moved the needle
one way or the other.
Cost of facilitation
The number of locations needed to conduct training
The number of employees needing training
The fully loaded cost of each employee to be trained
How quickly the employees needed to be trained

56
TOTAL LEARNERS 100 500 1,000 3,000
Classroom course creation $250,000 $250,000 $250,000 $250,000
Cost of facilitation* $4,000 $32,000 $40,000 $120,000
# of classes 4 32 40 120
Lost value (employee time)** $20,000 $100,000 $200,000 $600,000
Total cost of training $274,000 $382,000 $490,000 $970,000
Avergage cost per learner $2,740 $764 $490 $323
*Cost per hour per facilitator is illustrative at $400/hour as we used senior leaders to provide the training. The classroom course was 120 minutes and it was
necessary to add an additional 30 minutes for the facilitators to prep. Classroom size was 25 people and required 1 facilitator.
** For this example, we are using an illustrative cost of $100/hour to cover trainee time.
Initially, we were surprised at the cost per learner and thought that the cost of facilitation was
the key cost driver. However, even if we lowered the cost of facilitation to $0, the cost per
employee only lowers to $2,700, $700, $450, and $283 respectfully. This identified the key cost
driver for classroom training is the duration of the class (2 hours). Unless you reduced course
duration, the cost totals would not change significantly.
Employee rate** $100 per hour
Facilitator rate* $400 per hour
Facilitator time 2 hours
Learning time 2 hours
Classroom value table
Below is a table looking at an example of average cost
per learner for classroom training. As you can see, the
more you train, the lower the cost per learner.

57
*The cost per hour of the facilitator as they did not require the subject matter expertise of an instructor and was there to help users onboard to the
VR experience.
**The total average time to take the course was approximately 29 minutes, for this calculation we rounded this number up to 30 minutes.
***The time it took for the learner to check out the headset, be trained on the equipment, find the location for the training, and return the equipment averaged
12 minutes. For the purposes of these calculations, for this calculation, we rounded this time to 15 minutes.
***The cost to produce a VR training course was estimated to be 47% greater than the classroom course.
# OF LEARNERS 100 500 1,000 3,000
V-learn course creation*** $367,500 $367,500 $367,500 $367,500
Cost of onboarding $2,500 $12,500 $25,000 $75,000
Lost value (employee time) $5,000 $25,000 $50,000 $150,000
Facilitation cost $3,750 $3,750 $3,750 $3,750
Equipment (amortized) $169 $843 $1,687 $5,060
Total cost of training $378,919 $409,593 $447,93 $601,310
Average cost per learner $3,789 $819 $448 $200
V-learn value table
Looking at the v-learn training course, we needed to
account for different types of costs including the VR
headset (which we amortized), onboarding time to the
VR Headset, and the facilitation needed to help users
get fitted and trained on how to use the VR headset.
Just like classroom training, we did not account for any
additional infrastructure costs.
Facilitation cost was the same for each grouping of learners because we added more HMDs
to complete the training in approximately 6-7 days. If you reduced the number of HMDs, you
would increase the number of days needed to train, increasing the cost of facilitation. In this
model, we used 2, 10, 20, and 60 HMDs respectfully. If you remove facilitation, which we would
expect to happen once users became more familiar with using HMDs, (this would also reduce
onboarding time to $0), the impact is $3727, $787, $419, $174 respectfully.
Employee rate $100 per hour
Facilitator rate* $75 per hour
Learning time 0.5 hours
Onboarding time** 0.25 hours

58
*The cost of the equipment and licensing fees for the laptop/computer, LMS, and software necessary to manage the course deployment.
TRAINEES 100 500 1,000 3,000
E-learn course creation $250,000 $250,000 $250,000 $250,000
Lost value (employee time) $7,500 $37,500 $75,000 $225,000
Amortization / licenses* $2,125 $10,625 $21,250 $63,750
Total cost $259,625 $298,125 $346,250 $538,750
Average cost per learner $2,596 $596 $346 $180
E-learn value table
Looking at the costs of the e-learn, the curve begins
close to the classroom costs per learner, but it takes
much longer to cross the cost-per-learner curve with
the v-learn modality. It is no longer necessary to ac-
count for facilitation, but there was still the need to
account for the cost of the computer and the learning
management system license fees. In this example, we
did not account for any additional infrastructure costs.
Using the notional data in this example to illustrate cost differentiation, even at 3,000 learners,
e-learn was more cost effective than v-learn. However, once you remove the cost of onboarding
users and the cost of facilitation, v-learn catches up and actually becomes more cost effective
to deploy at 3,000 learners (in this model)—$180 per learner vs. $174 per learner (or about 4%
less expensive). If you extrapolate that to enterprise scale (10,000 learners as example), the
e-learn drops to $121 per learner and v-learn to $53 per learner.
Employee rate $100 per hour
Training time 0.75 hours

59
Considerations & Insights
for Deploying VR Training
SECTION 6

60
Considerations
and insights
Using and deploying virtual reality at enterprise scale uncovered many lessons that we
would not have learned by doing a proof of concept in the lab. This pilot required the
team to develop new skills and processes that simplified the deployment. Below are
some insights we learned. We explore each of these insights in this section.
Onboard new VR learners effectively
Create compelling learning content for VR
Collaborate cross-functionally
Reinforce learning through debriefing
Create templates to support scaling
Include VR as a part of a blended learning curriculum
Invest in the learning modality suited to the learning objective
VR is ready to scale in the enterprise
1
2
3
4
5
6
7
8

61
1. Onboard new
VR learners
effectively
Introducing a VR experience to an employee who has never had a VR
experience can be challenging based on several factors: culture, experience
with other modes of technology, topics to be trained and their role within the
organization.
Sending a “what to expect” communication will help learners feel less self-conscious
and more excited about the experience. Share enough information to address any
of a user’s perceived fears, questions or concerns. Make sure you share in detail
how they will be trained to use the VR headset, what to know about the experience,
safety, hygiene, etc., before they arrive for the training.
Pre-communication
The following are recommendations that should apply to most first-time users. We have discovered
that it is only necessary the first or second time a user engages with VR. Once they have had
an experience, it’s very easy. The important point to remember is to make sure the experience
is positive. You really only get one chance; if a user has a poor experience, the chances of
them being receptive to a second VR experience are extremely low. When addressing new user
onboarding in VR, there are three things you should consider:

62
For first-time learners we encourage someone to show them how to fit the HMD and
make sure it is adjusted appropriately. Do not assume users know how to put on a
headset, how to adjust it or how to be safe while using it. This single change to our
VR onboarding process for new learners increased our user satisfaction by more
than 100%. For first-time learners, we normally let them use a small office to reduce
the self-conscious feeling they experience the first time they put on a headset. This
helps put them at ease.
Arrival and fitting
The first thing a learner should see is a simple graphic that shows them how to
adjust their headset, how to improve clarity and how to adjust focus. Please notice
the learner didn’t put on the headset and immediately go to a “desktop” of apps
and settings. For first-time learners, help them begin by making their experience
comfortable and clear. Ideally, a learner will then progress to the actual simulation.
This “kiosk” way of engaging users helps make it simple, so they do not need to
understand how to navigate to the different apps, settings, etc. They put it on, make
sure everything is clear and the simulation starts.
Many employees get bored and distracted while being instructed on core VR skills,
so we suggest having most of the introduction done inside the experience (i.e., after
they put the headset on). It is not uncommon for users to possibly feel self-conscious
and discouraged if they are not receiving validation that they are engaging with VR
equipment and technology as intended, so if possible, let them know they are doing
great. Once a simulation starts for the first time, learners need a minute or so just to
look around and absorb what they are seeing.
Since learners may feel disoriented or dizzy from repeated side-to-side head
movements to explore, keep the initial screens simple so they can focus on one thing.
Our simulations used voice (not controllers) so the first thing users do is learn to talk
with the virtual avatars. If you are using hand controllers, it will be important to help
learners understand how to use them properly. No matter which user interface you
use, keep distractions to a minimum in the first sequence (unless you really want to
provide some eye candy wow). Give them some time to experience it – try not to rush
them. Give users a way to jump into the simulation right away if they do not need
extra time to get oriented.
The VR experience

63
VR offers a real opportunity to improve your employees’ vital
workplace performance through experiential learning.
If you want to create leadership skills that can impact confidence,
make emotional connections, reduce distraction, and truly make VR a
differentiating part of your recruiting, training and retention strategy,
then you should create content that is compelling and aligns to the
modality. Do not just create one path for your learners to follow.
Create realistic scenarios that learners can relate to. Create settings
that mimic your environment. Introduce them to virtual avatars
whose behavior is modeled after common personas.
VR holds a leading opportunity to help improve your organization’s
workplace skills by providing scalable hands-on practice and
feedback in a realistic and accessible on-the-job setting.
2. Create compelling
learning content for VR
63

64
3. Collaborate
cross-functionally
The business needs to be supportive of having the training offered to a large number of
people through this medium. Training functions and HR needs to be involved to help track,
organize, schedule and support training in VR. IT needs to provide support from a technical
and deployment perspective. You will need equipment procurement, asset tracking, user
management, access to WiFi and integration with learning management systems.
HR
IT support
IT operations
Security
Infrastructure
L&D
Designers
3D artists
Learning scientists
Data analytics
Developers
Topical subject matter specialists
VR development is not owned by one group. To get a VR program
prioritized, funded, supported and deployed, you need to coordinate and
collaborate across several groups:

65
We refer to this as “team level reinforcement,”
and its adoption helps drive the behaviors you
are looking to promote. Doing it via a debriefing
session was the leading approach we have
employed. It is a combination of v-learn and
classroom training.
Yes, v-learning is the most expensive approach,
but it can also provide the highest value.
Book times for individuals to take the v-learn
course. Then schedule debrief sessions in work
teams with the team lead as the facilitator. The
team debrief sessions help learners further
personalize the training to their team and make
action plans that help allow them to hold one
another accountable.
Single learner training is often more effective in VR. However, if that
learner can then share that experience and discuss what they learned
immediately afterward, the learning is exponential.
4. Reinforce
learning through
debriefing

66
5. Create templates to
support scaling
Since deploying our initial v-learn course, we’ve established tools and
templates for easier script writing and content collaboration, which help
new authors focus on creating experiences and utilizing VRs strengths.
Templates were the key reason we have been able to reduce the time it
takes to build VR course content.
6. Include VR as a part of a
blended learning curriculum
VR is not an appropriate platform to host every type of training, and we
do not believe that it will completely replace other modalities. What we
do think is that VR is a new modality that should be considered when
training specific types of skills. This multimodal approach is the future.
VR is ideal for practicing what you learn in a safe environment, in a
dynamic way, while e-learn is great for learning how to use software, or
quickly jump in for a learning burst. Classroom training is great when it is
necessary to collaborate and discuss a topic.
When you combine each, you are providing your employees with the
leading approach. This new modality will likely help transform a new age
of training and education.
66

67
7
. Invest in the
learning modality suited to
the learning objective
When users need to practice a soft or hard skills task, VR provides a safe
training environment.
Trainees are not judged for failing a task or saying something awkward. They can experience
new techniques without putting anyone or anything in harm’s way. A VR simulation never tires,
and a virtual avatar can be asked the same question, or perform the same task, 1,000 times
without losing patience.
The more thought that goes into a course design, the more the learners benefit. However,
it does come at a cost. It is necessary to deploy VR in the correct scenarios. VR is not ideal
at simulating training for a software application or very linear content. These are preferred
for e-learn or classroom modalities. We recommend VR be used for the practice portions of
training. It can also be effective to provide environmental (360) context to information. Skills
performed in specialized environments with specialized actions are better practiced in VR than
in other modalities. The key for VR is having enough learners to help justify the cost.

68
8.VR is ready to scale
in the enterprise
In the past, VR was just too expensive, too complicated and too
challenging to deploy outside of a small group. Today, enterprise headset
ecosystems are under $1,000, and these HMDs can be managed like any
other enterprise mobile device. Large (and small) studios are developing
compelling content, vendors are creating software packages to enable
non-VR developers to create their own content in a cost-effective way,
and big learning management system players are enabling VR content to
be more easily integrated into their platforms.
The value VR can provide is unmistakable when you apply it the correct
way. This VR pilot gave us 10 months of learning experience with VR
design and deployment. Since the pilot was completed in November 2019,
we’ve created and deployed v-learning on sales conversations, difficult
conversations and cyber training to over 4,000 PwC partners and staff.
We moved from a 10-month process to a three-month process, and we are
tracking for six weeks for soft skills training and two to three weeks for
hard-skills training modules.
68

69
Conclusions
SECTION 7

70
A combined team from PwC’s Learning and Development Innovation team, PwC’s Emerging
Technology Group, Oculus for Business, and Talespin worked together to launch a 10-month
pilot to understand if VR can be used to train enterprise employee soft skills.
This pilot answered specific questions: Is VR soft skills training more effective than traditional
training methods, and is VR soft skills training more cost-effective to deploy than traditional
training methods?
The answer to both questions is “yes.”
V-learn, the ability to use virtual reality to train skills, is more effective at training soft skills
concepts than classroom and e-learn training modalities. Learners are more confident, less
distracted, have a stronger emotional connection, and when deployed to enough learners,
v-learn for soft skills training can be more cost-effective than classroom or e-learning
modalities.
Many lessons were learned and captured. The key driver to achieve a positive ROI would
be to make sure you have enough learners to train, as VR course content requires greater
investment to develop than classroom or e-learn content. The difference in value will eventually
balance out, and you can achieve a higher ROI because VR training takes less time than
classroom or e-learn training. The more learners that train in VR, the easier it is to make up the
difference in development and deployment costs.
Conclusions
Is VR soft skills training more
effective than traditional
training methods, and is
VR soft skills training more
cost-effective to deploy than
traditional training methods?
The answer to both questions
is “yes.”

71
VR training is not the answer to everything and will not replace classroom or e-learn training
modalities anytime soon. It is not an appropriate platform to host every type of training, and so we
do not believe that it will completely replace other modalities. VR is a new modality that should be
considered when training specific types
of skills. Companies should consider a
multimodal approach in the future, as VR
can be ideal for practicing what you learn in
a safe and dynamic way. When you combine
classroom, e-learn, and v-learn together,
you are helping provide your employees
with an industry leading approach. This
new modality will likely help transform this
new age of training and education in the
enterprise.
VR can be ready to deploy at enterprise scale. Since we’ve completed the pilot, we have
deployed additional courses and have reduced our time to three months and are tracking to
reduce it to six weeks. The cost to deploy is also dropping, as we have become more efficient
at managing the equipment and have significantly improved the learners experience with VR.
Companies should consider a
multimodal approach in the future,
as VR can be ideal for practicing
what you learn in a safe and
dynamic way.

72
Acknowledgements
SECTION 8

73
Acknowledgements
What a journey: 10 months from start to finish, thousands of hours of planning, designing,
building, deploying, managing, analyzing and writing about this pilot. This effort was supported
by no less than 40 people, including our collaboration with Oculus and Talespin.
A special “Thank You” to the team that made this possible:
PwC US Executive Sponsors:
PwC US Executive Producers:
PwC US Lead Producers/Authors:
PwC US L&D Innovation Team:
PwC US L&D Research Team:
PwC US L&D Producers:
PwC US Emerging Technology Team:
PwC US IT Support:
PwC US Marketing and Editors:
Talespin:
Oculus:
Cleanbox Technology:
Scott Likens, Tom Puthiyamadam
Daniel Eckert, Rayellen Smith
Daniel Eckert, Andrea Mower
Jill Kirtland, Steven Murphy, Krishma Patel, Rayellen Smith
Mary Lefaiver, Dhan Lall
Ceil Adam, Michelle Hershiser, Jill Kirtland, Krishma Patel
Tom Foth, Sarah Warning, Tony Anderson, Sami Al-Shatri,
Rosalyn Nguyen, Kevin Frank, Daniel Eckert, Andrea Mower
Dominic Caldura, Kerwin Larrobis, Fernando M. Augusto, Redwin
Padua, Debasish Hira, Gopi Yalamanchili
Whitney Romanoff, Amanda Westbrooks, Eileen Feretic
Eli Weisman, Rob Stemm, Kyle Jackson, Jasmine Bulin,
Remmelt Blessinga
Danielle LaCoste, Marie Brunet, Maria Fernandez Guajardo,
Isabel Tewes, Alicia Berry, Andy Coughlin
Amy Hedrick
We would be remiss if we didn’t thank all of the PwC employees that either helped with user
testing or were actual participants in the study. It is because of them that we are achieving our
purpose: building trust in society and solving important problems.
73

74
© 2020 PwC. All rights reserved. Not for further distribution without the permission of PwC. “PwC” refers to the
network of member firms of PricewaterhouseCoopers International Limited (PwCIL), or, as the context requires,
individual member firms of the PwC network. Each member firm is a separate legal entity and does not act as
agent of PwCIL or any other member firm. PwCIL does not provide any services to clients. PwCIL is not responsible
or liable for the acts or omissions of any of its member firms nor can it control the exercise of their professional
judgment or bind them in any way. No member firm is responsible or liable for the acts or omissions of any other
member firm nor can it control the exercise of another member firm’s professional judgment or bind another member
firm or PwCIL in any way.
Chai M. Tyng, Hafeez U. Amin, Mohamad N. M. Saad, and Aamir S. Malik. (2017). The Influences of
Emotion on Learning and Memory. Frontiers in Psychology 2017: 1.
Gupta, S.C., Wantland, C.A. and Klein, S.A. (1996). Cyberpathology: Medical Concerns of VR
Applications. Journal of Medicine and Virtual Reality 1996: 1 (2) 8-11.
https://business.oculus.com/support/665720147270975/
1
2
3

Pwc - Understanding The Effectiveness of Virtual Reality Soft Skills Training in the Enterprise - Study

More Related Content

What's hot (20)

Similar to Pwc - Understanding The Effectiveness of Virtual Reality Soft Skills Training in the Enterprise - Study (20)

More from Abdallah Yakoub (20)

Recently uploaded (20)

Pwc - Understanding The Effectiveness of Virtual Reality Soft Skills Training in the Enterprise - Study