Photo credit: Getty Images

How bots, algorithms, and artificial

intelligence are reshaping the future
of corporate support functions
Alexander Edlich, Fanny Ip, and Rob Whiteman

Industrial companies are discovering additional sources of value in

applying advanced technology to general and administrative support
functions. The results can be impressive for businesses that can
adapt to the disruption of legacy systems.
As advanced industrial companies continue to algorithms. Our experience shows that companies
grow, support functions are coming under more and following a systematic approach to tech-enabled
more pressure to deliver value, manage complexity, transformation can reap substantial efficiency gains
and reduce cost. Many organizations have already in their general and administrative (G&A) functions.
tapped the potential of traditional levers such as The resources freed up in this way can then be
centralization, offshoring, and outsourcing. To deployed in more valuable activities such as business
succeed, today’s leaders are turning to digital counseling and scenario analysis. This article
solutions and automation to improve performance explores: the value that can be created through
and reduce costs across finance, human resources, tech enablement in administrative functions;
and IT. looks at real-life examples from finance, HR, and
IT; considers key success factors; and suggests
As technologies such as robotic process automation how companies can make the best start on their
(RPA) mature, an increasing amount of the work transformation journeys.
done by people will be transferred to bots and

1 McKinsey & Company

 Sources of value the region of $60 to $120 billion globally, albeit
Today’s better, faster, and cheaper technology is set with considerable variation between segments
to reshape support functions—and will do so without (Exhibit 1). Although the direct cost savings may
the years of pain often associated with traditional appear small when compared with those in areas
tech initiatives such as enterprise resource planning such as procurement or manufacturing, McKinsey
systems. Early results in other industries show that analysis indicates that a company’s ability to
companies can achieve 5 to 10 percent cost savings in deliver productivity improvements in G&A is one
as little as 18 to 24 months, with long-term savings of of the biggest predictors of its ability to outperform
more than 30 percent. its industry in total returns to shareholders.
Approached in the right way, then, automating
Across the advanced industrial sector, the median routine G&A tasks through a tech-enabled
spend on G&A expenses accounts for 4 to 8 percent transformation can deliver substantial impact to the
of revenue. Our estimates indicate that the value whole organization.
that could be created from tech enablement is in

EXHIBIT 1 The value from tech enablement in G&A activities varies by industry sub-segment

Margin expansion
$ billions

Automotive1 24-50

Other
mobility2 4-8

Aerospace/
4-7
Defense3

Broader
industrials
& semi- 31-56
conductors4

Total 63-121

1
Whole value chain including tier 1 suppliers, automotive OEMs, and dealers
2
Commercial vehicles and off-highway equipment (e.g., for construction and agricultural use) including tier 1 suppliers, equipment
manufacturers, and dealers and distributors
3
Includes tier 1 suppliers and equipment manufacturers
4
Includes industrials, food processing and handling, motion and controls, industrial automation, and electrical, power, and test equipment
across the value chain: component suppliers, equipment manufacturers, distributors, VARs, engineering and services providers, and product
companies

How bots, algorithms, and artificial intelligence are reshaping the future of corporate support functions 2
 Modernizing the finance function of the opportunity requires advanced cognitive
At many organizations, the finance function is automation technologies such as machine-learning
beginning to evolve toward a more integrated algorithms and natural-language tools.
consultative model that supports value-based
decision making. However, companies often have At one company that was trying to verify whether
difficulty devoting enough attention to the analysis employees were reporting vacation time accurately,
required to support this model because of the the internal audit function developed an algorithm
demands of day-to-day transactional activities. that compared declared vacation days with data
The sheer scale of these activities makes them ripe from badge swipes and computer-usage data.
for automation: in fact, our analysis shows that 27 Another company reengineered every part of its
percent of finance activities could be automated record-to-report process by redesigning activities
using technologies already available (Exhibit and organizational structures around a portfolio
2).1 About a third of this opportunity could be of technologies. Managers introduced RPA for
captured using basic technologies such as robotic tasks such as preparing journal entries and applied
process automation (RPA), a type of general- machine learning to reconcile differences between
purpose software that can sit on top of existing accounting records. Having demonstrated that
IT systems. Capturing the remaining two-thirds natural-language tools could be successfully
deployed to produce report commentary, the

EXHIBIT 2 Many sub-functions in finance can be automated using current technologies. . .

Lorem ipsum

Potential for automation using proven technologies

Capturable using current technologies1 Technically automatable but difficult to capture technologies2 Not automatable using current technologies

Proportion of tasks within the sub-function, %

General accounting operations 35 61 4

Revenue management 31 62 7
Cash disbursement 31 63 6
Financial planning and analysis 23 56 21
Financial controlling and external 25 59 16
reporting
Audit 19 50 31
Treasury 23 53 24
Tax 26 55 19
External relations 18 40 42
Business development 15 41 44
Risk management 21 52 27
Finance as a whole 27 57 16

1
Taking into account the relative complexity and expense of different types of automation technology: robotic process automation, machine
learning, smart workflows, cognitive agents, and natural-language processing
2
Because of investment requirements and technological complexity

3 McKinsey & Company

 company has redesigned processes to enable this Optimizing workforce deployment
technology to be introduced later. Overall, the (Human Resources)
company expects to see cost savings of 35 percent As expectations evolve, HR needs a tech-enabled
over the next two years from implementing its transformation of its own. The possibilities are
automation road map. legion (Exhibit 3). Bots can act as a “third arm” for
the HR organization by supporting transactional
As the finance function becomes the hub for activities such as time collection, payroll, and
enterprise data, automation efforts need not be record keeping. Activities such as talent sourcing
limited to finance processes alone. One agricultural offer huge scope for algorithm-based technologies.
equipment manufacturer successfully automated Meanwhile, conversational AI platforms such as
its sales and operations planning process by turning chatbots and cognitive agents are beginning to
a handful of data scientists loose on financial and manage inquiries previously handled by HR service
operational data managed within the finance centers including benefits administration and
function. By introducing machine algorithms record-keeping activities. Such platforms provide
into the process, the company not only improved 24/7 coverage and operate alongside the human
efficiency but also enhanced its ability to react to workforce.
natural business cycles.

EXHIBIT 3 . . . as can many sub-functions in HR. . .

Potential for automation using proven technologies

Capturable using current technologies1 Technically automatable but difficult to capture2 Not automatable using current technologies

Proportion of tasks within the sub-function, %

Strategy, planning, and policy 16 46 38

Compensation planning and 17 52 31
performance management
Organizational development 18 44 38
Talent sourcing and recruiting 22 56 22
administration
Learning and development 20 57 23
Employee, community, and
services 22 55 23
Benefits administration 26 68 6
Time collection, attendance, 31 67 2
and payroll
Record keeping and reporting 29 67 4
Safety and health 23 67 10
Labor relations 17 43 40
HR as a whole 22 55 23

1
Taking into account the relative complexity and expense of different types of automation technology: robotic process automation, machine
learning, smart workflows, cognitive agents, and natural-language processing
2
Because of investment requirements and technological complexity

How bots, algorithms, and artificial intelligence are reshaping the future of corporate support functions 4
 Finally, predictive analytics can be used to improve Building a scalable technology
hiring, retention, and succession planning. One backbone
company undergoing a restructuring was trying In addition to supporting the deployment of
to identify promising employees to lead its new automation technologies in other functions, IT
organization, but found that HR and company data can take advantage of bots and algorithms in its
was scattered across the enterprise. Using machine- own operations (Exhibit 4). Our analysis shows, for
learning capabilities, the company aggregated example, that 40 to 80 percent of the basic activities
demographic, performance, and organizational data required to resolve service desk tickets can be
to identify the key drivers of employee performance, automated through RPA and related technologies.
identify the individuals with the greatest potential,
and find roles in which they would succeed. When one company analyzed incident tickets, for
Leaders then transformed the recruiting process instance, it found that between 25 and 35 percent of
to focus on early markers of success and redeploy them were requests for “password reset” or “access.”
talent in new roles. These measures enabled the To resolve these tickets, it introduced RPA bots
company to achieve improvements of 80 percent that connect with multiple applications via the user
in the conversion of new recruits, 26 percent in interface or application programming interfaces. By
productivity, and 14 percent in net income. adopting automated ticket resolution, the company

EXHIBIT 4 . . . and many sub-functions in IT

Potential for automation using proven technologies

Capturable using current technologies1 Technically automatable but difficult to capture2 Not automatable using current technologies

Proportion of tasks within the sub-function, %

Application development 20 55 25

Application maintenance 26 66 8

Application hosting 25 62 13

Network services 18 50 32

End-user services 25 62 13

Management support and

systems 19 43 38

IT as a whole 22 55 23

1
Taking into account the relative complexity and expense of different types of automation technology: robotics process automation, machine
learning, smart workflows, cognitive agents, and natural-language processing
2
Because of investment requirements and technological complexity

5 McKinsey & Company

instantly freed up employee capacity and reduced Decentralize governance. Traditional
outsourcing contract costs for helpdesk support, as transformation efforts tend to follow centralized
well as reducing resolution times and improving models, but technology-enablement programs favor
performance. Alternatively, service desk automation decentralized options. In our survey, respondents
tools exist that support automation of repeatable from successful organizations were more likely than
IT operations workflows such as user provisioning, peers to say their functions or business units were
password resets, and event log monitoring. accountable for delivering automation efforts, with
or without support from a central team. Conversely,
Similar use cases exist in areas such as application less successful organizations were more than twice
testing, data migration, and network administration. as likely as successful ones to say a central team had
Automating transactional activities like these can sole responsibility for delivering automation.
enable IT to free up capital and resources to focus
on strategic activities such as modernizing ERP Ensure IT is involved. Automation programs stand
platforms, migrating to the cloud, and developing or fall by the engagement of the IT function. The IT
new digital solutions for the business. teams at successful organizations are more likely to
have automated their own processes and taken part
Lessons learned in capturing value in initial discussions and planning for automation
Even the most successful companies face projects prior to the pilot stage. Among advanced
challenges in capturing value from tech-enabled industrial companies, 69 percent of successful
transformations. We have identified a few common organizations involved IT early in the automation
keys to success from automation leaders’ responses planning process.
to our recent survey:
Internalize costs and benefits. Leaders of successful
Make automation a strategic priority. Organizations efforts had a deep understanding of the total cost
whose automation efforts prove successful are more of ownership for automation projects. Across all
likely than others to have designated automation programs, the most common benefit cited was
as a strategic priority.2 Among advanced industrial reduced costs.
companies, about three-quarters of successful
automation programs had been prioritized as part of Prioritize workforce management. Many large
the strategic-planning process. organizations predict their companies will face
automation-related skill gaps in the future;
Deploy automation technologies systematically. successful organizations make addressing this gap
Whether companies achieve success through one of their top five priorities. They also agree that
traditional top-down deployment or flexible agile acquiring employees with the right skills is their
methods, following a systematic rather than ad hoc biggest automation-related challenge in the next
approach is vital. Our survey found robotic process three years.
automation to be the most commonly adopted
automation technology. In addition, successful How to get started
companies were more likely than others to cite A tech-enabled G&A transformation journey
the use of advanced technologies such as machine typically involves three phases: start-up, launch,
learning, cognitive agents, and natural-language and scale.
processing to supplement RPA.

How bots, algorithms, and artificial intelligence are reshaping the future of corporate support functions 6
 Start-up preparation and design through to build, test,
In this first phase, a company typically tackles: and refine.

Assessment and roadmap. To decide which sub- IT support. Even when sprints are led by other
functions, processes, and locations will benefit functions, involving IT early is critical to securing
most from tech-enabled transformation, start with the right infrastructure and environment and
a clear understanding of your organization and standardizing processes for deployment and
the activities it performs. Assess the potential for maintenance. Successful leaders establish clear lines
automation by combining top-down analysis with of accountability between functions, automation
task-by-task validation, then use your findings to resources, and IT support groups to avoid confusion.
inform decisions about which technologies to invest
in and where to deploy resources. Finally, translate Center of excellence (CoE). Most companies
all this into a roadmap to guide your program. choose to set up a tech-enablement CoE to provide
governance, build capabilities, and maintain assets.
Proof of concept. To demonstrate feasibility and This will typically follow a centralized model
potential for impact, build a practical application initially with some development capacity embedded
such as a simple bot or algorithm in weeks, not in functions, before moving to a federated model as
months. This gives you early experience with the transformation matures.
technology and a chance to create presentations,
videos, and other communications to generate Scale
excitement for your broader program. In the last phase, transformations typically
complete:
Vendor selection. Selecting the right technologies
to support your transformation is a balancing act Additional sprints. Once you have conducted a few
between maintaining a simple architecture and sprints, it’s time to scale up systematically and
maximizing impact. Most companies start with an rapidly deploy technologies in further sprints. As
RPA platform and add complementary technologies each new process is deployed, maintenance and
such as business-process management or optical support teams can resolve issues and manage
character recognition within the first three to six changes while continuing to refine their support
months. More complex automation tools, such as model.
natural-language processing, are typically added
CoE scale-up. The speed at which you scale up
after about a year. Emerging technologies, such
your CoE depends on the number of opportunities
as cognitive agents, are usually confined to pilots
in your pipeline. As your program scales, the CoE’s
during the early stages of a transformation.
interaction model with other teams will evolve to
Launch shift more responsibilities to the business, and in
Areas of focus in the launch phase usually include: turn the business will start to undergo a culture
shift with employees seeing technology as a source
Domain sprints. Companies typically build of support, not competition. Ongoing capability-
solutions through multiple rapid, intense working building and change- management efforts will help
sessions or sprints. A sprint usually consists of five to build support for the new way of working.
or six use cases relating to a specific “domain”: a
sub-function, process, or location. Sprints employ
agile methods and follow standard IT phases from

7 McKinsey & Company

Fueled by the promise of productivity gains, journey, such as banking, while capitalizing on
technology-enabled transformations are beginning internal capabilities already embedded in the
to reshape the future of work in support functions. organization, such as lean.
Bots and algorithms are already at work alongside
humans, but adapting to the disruption can be
1
For details of the analysis, see Frank Plaschke, Ishaan Seth,
challenging even for an industry familiar with
and Rob Whiteman, “Bots, algorithms, and the future of the
physical automation. Even so, advanced industries finance function,” McKinsey & Company, January 2018.
are well positioned to capitalize on lessons from 2
“The automation imperative,” McKinsey & Company,
other industries that are further ahead in the September 2018.

Alexander Edlich is a senior partner in McKinsey’s New York office, Fanny Ip is an expert associate partner
in the Southern California office, and Rob Whiteman is a partner in the Chicago office.

Copyright © 2018 McKinsey & Company. All rights reserved.

How bots, algorithms, and artificial intelligence are reshaping the future of corporate support functions 8