KE1222

July 14, 2022

M O H A N B I R S AW H N E Y, VA R U N P O D D A R , A N D C H A R L O T T E S N Y D E R

Tapping into a Digital Brain:

AI-Powered Talent Management at Infosys

In the autumn of 2017, Rahul Gupta glanced out of the plate-glass window in a conference
room on the Infosys campus in Bengaluru, India, to see a large flock of starlings lifting off from
nearby trees and soaring into the sky. As the birds swooped left and right in a stunning display of
aerial acrobatics, he marveled at the mystery of nature that allowed so many birds to sense and
respond as one. It was just this sort of agile synchronization that Gupta, the head of data and
analytics at Infosys IT, sought to enable within his own organization.

Infosys was a global information technology consulting and software services provider
that operated in more than 40 countries, with a market capitalization of about US$33 billion
and revenues of US$10.5 billion in 2017.1 A large employer, Infosys had more than 200,000
employees,2 including software specialists and technical consultants. These employees traditionally
had been assigned to projects based on a detailed manual assessment of their skill sets and
experience. Two years earlier, however, delivery leaders at Infosys had recognized the need for
the organization to optimize the talent allocation process based on real-time client project needs,
available employees, employee skill sets, and upcoming technology trends. What had begun with

1 “Infosys Market Cap,” Macrotrends, accessed September 17, 2021, https://www.macrotrends.net/

stocks/charts/INFY/infosys/market-cap; “Infosys Revenue,” Macrotrends, accessed September 17, 2021,
https://www.macrotrends.net/stocks/charts/INFY/infosys/revenue.
2 “Infosys: Number of Employees,” Macrotrends, accessed September 17, 2021, https://www.macrotrends.net/
stocks/charts/INFY/infosys/number-of-employees.

© 2022 by the Kellogg School of Management at Northwestern University. This case was prepared by Professor Mohanbir
Sawhney, Varun Poddar, and Charlotte Snyder based on conversations with Infosys executives, to whom the authors
are grateful. Cases are developed solely as the basis for class discussion and are not intended to serve as endorsements,
sources of primary data, or illustrations of effective or ineffective management. For pedagogical purposes, the authors
have fictionalized individuals, conversations, strategies, assessments, and other details to protect the identity of the
people and institutions involved. To order copies or to request permission to reproduce materials, call 800-545-7685 (or
617-783-7600 outside the United States or Canada) or email custserv@hbsp.harvard.edu. No part of this publication
may be reproduced, stored in a retrieval system, used in a spreadsheet, or transmitted in any form or by any means—
electronic, mechanical, photocopying, recording, or otherwise—without the permission of Kellogg Case Publishing.

This document is authorized for use only in Prof. Shubhi Gupta's Managing Talent and Career in Organizations_T6 at FORE School of Management from Jan 2023 to Jul 2023.
 A I - P ow e r e d T a l e n t M a n ag e m e n t at Infosys KE1222

a one-page list of skills gaps in the workforce evolved into a broader effort to better understand
each employee’s proficiencies and career aspirations, ultimately blossoming into a company-wide
artificial intelligence and machine learning initiative to optimize the talent management function.
The vision was to create an agile, dynamic organizational structure powered by a “digital brain.”

As part of this vision, Gupta was helping to construct a centralized data foundation for an
AI-based talent management system that would provide a 360-degree view of open positions and
available employees and make recommendations to optimize value, skills, networks, billing, and
collaboration across the organization.

As the murmuration of starlings concluded their twilight performance and settled into the
trees to roost, Gupta’s thoughts returned to the table, where he and two colleagues were working
to make the talent management product vision a reality. Seema Krishnan led the execution team as
the artificial intelligence and machine learning expert charged with building the pipeline and the
models, and Praveen Pandey served as the domain expert who ensured synchronization with the
talent management group.

The three had gathered to study the feedback they had collected during interviews with
ground-level users of the current manual (off-line) process for allocating talent to client projects.
Given the recommendations and needs of these various stakeholders, the project team now had
to make several data, design, and deployment decisions before they could begin building the new
solution. First, they had to consider what data variables should be used to define “supply” (i.e.,
employees available for assignments) and “demand” (i.e., staffing needs for client engagements).
They also had to establish a system for ensuring that this data was accurate and kept up to date.
For automated talent allocation to function effectively, the employees, managers, and talent
management group at Infosys would need to comply with data collection and use the system in
a consistent way. This would not be possible if these stakeholders failed to trust in the system’s
recommendations. The team members wondered how they could drive adoption and build trust
throughout the organization. Last but perhaps most critical, the team had to establish a plan for
maintaining the system—even at this early stage. Determining how success would be measured
and who would be responsible for tracking these key performance indicators would be the only
way to enable continuous improvement.

Talent Allocation at Infosys

In 2017, Nandan Nilekani, the CEO and a co-founder of Infosys, challenged the company to
model itself after a living organism that could sense, respond, and adapt to the changing needs and
conditions. His vision involved using data, artificial intelligence, and machine learning to remove
latency, drive automation, conduct instant simulation. It also included gathering micro-feedbacks,
with the idea that Infosys could transform itself into a sentient “live enterprise”: an organization
capable of continually reimagining and reengineering processes to accelerate the testing and
implementing of new ideas.3 This networked, integrated ecosystem would be designed to observe

3 Jeff Kavanaugh and Rafee Tarafdar, The Live Enterprise: Create a Continuously Evolving and Learning Organization
(New York: McGraw-Hill, 2021), 13.

2 Kellogg School of Management

This document is authorized for use only in Prof. Shubhi Gupta's Managing Talent and Career in Organizations_T6 at FORE School of Management from Jan 2023 to Jul 2023.
 KE1222 A I - P ow e r e d T a l e n t M a n ag e m e n t at Infosys

and capture data continuously, enabling intuitive decisions with minimal human intervention. The
goals were to free people from daily, mundane tasks that technology could handle more efficiently
and to encourage constant experimentation to improve, simplify, or eliminate processes to yield
more value.4

Allocating employees (“talent”) to client projects was an ideal starting place for the
transformation that fit Nilekani’s vision. Even minor misalignments between talent and clients
could increase the risks of client dissatisfaction or employee churn and have serious repercussions
for the company’s bottom line. Infosys’s leadership believed that a superior employee experience
would build resilience through longer employee tenure and stronger collaboration, ultimately
enabling the company to outperform its competitors in innovation, client satisfaction, and
profitability. Thus, ensuring that matches between people and projects brought out the best
in individual strengths and group collaboration across the enterprise, which was crucial to the
company’s overall success and impact.
Talent allocation at Infosys required significant time and resources—more than one million
hours of effort per year.5 The tasks seemed simple enough: determine open positions for clients’
projects; evaluate the skills and experience they required; identify employees who were both available
and qualified; and assign those who best matched the requirements of the projects. Historically,
these tasks had been performed manually. Developing a concise list of prospective candidates for
new projects took days or even weeks. Hiring managers followed a three-step process to allocate
employees to projects; they identified an inclusion set based on who was available, an exclusion set
based on who would not be eligible due to factors such as competitive constraints from previous
projects, and finally, a list of required skills based on client needs to determine an appropriate fit.

As staffing decision-makers conducted each of these repetitive, high-volume assessments, they

encountered several challenges. Infosys consisted of a large collection of mini organizations, each
of which managed hiring and project allocation decisions independently. At any given moment,
an employee on one delivery team who was unutilized might be a perfect fit for another delivery
team’s needs, but a central team managing demand and supply across the organization was
needed to recognize the opportunity. Information about which employees were available for new
assignments needed to be visible across the organization, and the reliability of data had to be
improved. And whereas Infosys had a predefined taxonomy of more than 2,000 skills that could be
attributed to an employee (e.g., expertise in the Java programming language or the .NET software
platform), the company recognized the need to have a more consistent process for attributing
skills to employees. Some delivery managers chose to update lists of skill sets manually for their
employees; others did not.

To alleviate these challenges, Infosys formed a Talent Task Force (TTF) consisting of a panel of
experts with three main goals: streamline personnel recommendations; address career aspirations
of employees; and determine areas in which the company should recruit new talent. This task
force was responsible for providing recommendations to staffing decision-makers to facilitate more

4 Kavanaugh and Tarafdar, The Live Enterprise, 14.

5 Sunil Thakur, “Digital Brain for Talent Management,” white paper, Infosys Ltd., June 2, 2020.

Kellogg School of Management 3

This document is authorized for use only in Prof. Shubhi Gupta's Managing Talent and Career in Organizations_T6 at FORE School of Management from Jan 2023 to Jul 2023.
 A I - P ow e r e d T a l e n t M a n ag e m e n t at Infosys KE1222

efficient pairings between available employees and open projects, in effect, to fine-tune the process.
However, the value of these recommendations relied on the true insight of demand, supply, and
employee skill sets across the organization. The task force members did not make final decisions
on talent placement; they provided recommendations to staffing decision-makers to facilitate
more efficient pairings between available employees and open projects. Hiring managers might
decide that the recommended employees did not have the required skill sets, or the employees
themselves might refuse an assignment if they did not feel that it was a good fit. Consequently,
TTF’s recommendations experienced significant rejection rates.

The task force further recognized that reducing these rejection rates would require improving
the current methods for identifying demand, supply, and skill proficiency.

Identifying Demand for Talent

Since the introduction of the task force, Infosys had a central, online system for managing
staffing demands and identifying available employees; however, the information was not always
accurate. Additionally, several open positions listed in the demand system were unreliable; some
were obsolete, some were duplicates, and others had abstract descriptions that did not clearly
convey the required skills. Some delivery managers did not use the company’s demand system
unless they had a need that none of their internal talent could fulfill. Instead, as these delivery
managers became aware of new demands in their sphere of operations, to ensure client satisfaction,
they frequently staffed the new assignments with their own employees who were rolling off projects
without ever entering their team’s needs into the central system.

Identifying Supply of Talent

To allocate people quickly and accurately to projects, the task force and staffing decision-
makers needed accurate information regarding who was available for reassignment, as well as
forecasting capabilities to anticipate who would be available in the short term.
The company’s fundamental sources of talent supply were employees about to roll off projects
and those “on the bench” (i.e., not currently assigned to projects but still on the payroll); however,
consistent, reliable information about employees’ work status was not readily available across the
organization. Although the task force could theoretically tell when employees were eligible for
new assignments based on the reported end date of their current projects, this data was not always
reliable because managers could update it as the project progressed. As a result, the TPD had to
engage in time-consuming, manual processes to confirm the availability of each person for every
staffing recommendation.

Historically, all delivery managers had run their own talent pool of about 200 people. When
employees finished a project, their profiles were supposed to be accessible in a central pool of available
employees (“the bench”) for other delivery teams in the organization to recruit from. However, in
the interest of retaining talent supply and ensuring client satisfaction, delivery managers tended to
reassign their employees to new opportunities as projects wrapped up without officially releasing

4 Kellogg School of Management

This document is authorized for use only in Prof. Shubhi Gupta's Managing Talent and Career in Organizations_T6 at FORE School of Management from Jan 2023 to Jul 2023.
 KE1222 A I - P ow e r e d T a l e n t M a n ag e m e n t at Infosys

the employees to the bench. If appropriate opportunities did not immediately present themselves,
some managers would place a “buffer” code on the employee’s profile, indicating that the employee
was being held for a manager and was therefore inaccessible to other managers. These practices
made true visibility of available talent challenging.

Identifying Employee Skill Proficiency

Matching employee skill proficiency with project requirements was crucial to ensuring
maximum productivity and successful outcomes for both employees and projects. Hiring managers
needed to be able to identify clearly both the skills required in each project and those each employee
possessed to determine who would be the best fit for a particular assignment. However, limited
data existed regarding how much hands-on experience Infosys employees had in working in a
particular skill area.

The task force estimated an employee’s skill proficiency by looking at two sets of variables: (1)
how long he or she had worked on a particular project, and (2) which skills were required for that
project. This method was a good proxy of proficiency, but it was not perfect because not everyone
on a project worked on everything. Hiring managers commonly rejected recommendations on the
basis that the recommended employee turned out to lack sufficient expertise in a required skill.

In addition, the current age of digital transformation and constant technological evolution
meant Infosys’s employees were always training to develop their existing skills and learn new
ones, which added another layer of uncertainty when the task force was looking for skill matches
among employees.

Designing the New Talent Management Solution

The goal of the new talent management system was to enable managers to feed their demand
requirements into the system, which would then instantly provide them the top 10 qualified
recommendation profiles (see Exhibit 1). The new system would support the TTF, in their work
of matching employees to client needs, eliminating routine tasks and freeing them up to focus on
higher-order functions.

Before that could happen, though, the project team first had to design a data strategy that
addressed the following concerns: what data variables should be used to define supply (e.g.,
employees’ personal details, skill proficiencies) and demand (e.g., client details, project descriptions);
how to ensure that the data was accurate and up-to-date; how to incentivize adoption of the new
system; and how to measure success and enable continuous improvement (see Exhibit 2).

The project team hoped to expand its data foundation to include unstructured data from within
the company’s systems, as well as from publicly available sources. By incorporating information
from sources that included human resource systems and social media, data could become richer
and potentially more accurate in predicting successful outcomes for both people and projects. The
team members considered integrating employee LinkedIn and Facebook accounts to enrich the

Kellogg School of Management 5

This document is authorized for use only in Prof. Shubhi Gupta's Managing Talent and Career in Organizations_T6 at FORE School of Management from Jan 2023 to Jul 2023.
 A I - P ow e r e d T a l e n t M a n ag e m e n t at Infosys KE1222

employee dataset; however, they knew they would need to tread cautiously to ensure that they did
not introduce sources of bias into the data. Optical character recognition (OCR) could be used
to mine data sets such as project descriptions, project rules, development strategies, and personnel
lists to bring visibility to similarities in employee experiences and networks—though OCR did
have its shortfalls, as it wasn’t always as accurate with nonstandard documents, such as those with
background images. All of the data could potentially be modeled as an employee knowledge graph
(EKG), linking each employee with his or her skills, projects, networks, interactions, and devices.
Ideally, the EKG could drive contextualization and search, expedite staffing, improve matching,
enhance the employee experience, and increase client satisfaction.

As the project team turned its attention to determining data variables that would accurately
profile employee skill sets and client engagement needs, the potential impact of the new system
on the lives of each employee in the company weighed heavily on data and analytics head Gupta.
“We must recognize that the machine learning initiative we are building is not just about
our company’s operations or even simply about our customers,” Gupta told the team. “It’s going
to profile all of our colleagues and play a significant role in defining their career paths by either
recommending them or filtering them out as job opportunities arise. You don’t get any more
sensitive than this. Everyone needs to be able to trust this system.”

Execution team leader Krishnan agreed. “This is why we must put as much data as possible into
it so that the algorithms can form the most thorough and accurate picture,” she said. “Computers
will not discriminate, as people might. Hiring managers might think they know who would be
good for an upcoming project, but the system itself will be better equipped to find the most
qualified people across the entire organization for each position.”

“Perhaps, but that was not the case for women who applied to Amazon for software engineer
positions at one point,” Gupta countered. “Amazon’s AI-based recruitment tool was supposed to
assist in finding the most qualified candidates, but the data fed into the system was biased; it was
based on the company’s predominantly male work force. Because software development roles are
filled mostly by men, the algorithm deduced that they must perform better than women, so it
ended up discriminating against female applicants. Although it’s true that computers should be
able to make unbiased assessments, it’s also possible that our data could be biased in ways that we
aren’t even aware of. Our company prides itself on being employee-centric, and we must make sure
that we are looking out for our people’s best interests.”

“With access to literally hundreds of data variables across our internal systems, I’m sure that we
can integrate rich sources of information into the new recommendation engine without causing
bias,” replied Krishnan, looking over some brainstormed lists of variables the team was considering
for inclusion in the solution (see Exhibits 3 and 4).

“But what variables seem most critical to include in our new data foundation, and what
variables would be too sensitive to include in the system? Should we, for instance, exclude
gender as a variable in profiling employees?” Gupta asked. “At a minimum, we need to maintain
comprehensive, granular data on where our talent is deployed, what each individual is doing, what
they are good at, and how relevant their skills might be on other client engagements.”

6 Kellogg School of Management

This document is authorized for use only in Prof. Shubhi Gupta's Managing Talent and Career in Organizations_T6 at FORE School of Management from Jan 2023 to Jul 2023.
 KE1222 A I - P ow e r e d T a l e n t M a n ag e m e n t at Infosys

“Our hiring managers have told us that the biggest reason that they reject candidates is due to a
lack of skill proficiency,” domain expert Pandey replied. “We need to improve our skill proficiency
assessments. We know hiring managers also want to know what potential candidates have been
doing in the most recent two years in terms of skills and training. In addition, they have requested
that we highlight staffing or project connections in the recommendation profiles. As you know,
they believe that teams composed of people who have previously worked on similar accounts or
projects are predisposed for success. They also want individuals with outstanding performance
ratings to receive a higher match percentage.”

“Wait a minute—increasing the match percentage for individuals with outstanding performance
ratings doesn’t seem right. By doing that, we might we inadvertently create a prediction system in
which only the best employees feel valued, and the average performers get demoralized. We need
to tread carefully here,” Krishnan cautioned.
“Understood,” Gupta said. “Switching gears for a moment, however, we certainly have lots of
data, but we don’t have the luxury of sensors automatically sending terabytes of information on
each individual every two minutes. How can we make sure that our data is consistently available,
reliable, and up-to-date?”

“The most important thing we can do to ensure strong data hygiene and to improve the
system over time is to require everyone to use it,” Krishnan said. “Employees should be required to
update their skills regularly—maybe even as part of their performance reviews. Managers should
update all their employees’ skills as well, so that we aren’t just relying on self-assessments. When
hiring managers want to staff positions, they should have to fill out a demand in the system.
And the person they hire has to be recommended by the system. If hiring managers don’t find
whom they want through the system, they need to keep adjusting their demand requirements until
their perfect fit shows up. We can use telemetry to track engagement with the system. And once
someone is hired, the system can reflect that new position automatically on both the supply and
demand sides in employee profiles, client profiles, and project descriptions. The system will be able
to sense, respond, and adapt—and it’ll just keep getting smarter.”
“Some of this sounds a bit draconian,” Pandey said. “I worry that all of these requirements
could spark a backlash from hiring managers.”

“If the system works well, then everyone will appreciate it,” Krishnan replied.

“Maybe we can find other ways to ensure that the talent management system recommendations
will be implemented,” Pandey said. “I would rather have everyone want to use the system. What
about word of mouth? If we have some early success stories, we can publicize them widely to
help everyone see the benefits of using the system. We can make the user interface super intuitive
and highlight how fast and reliable the system is. And the new talent management system’s
recommendations are accepted at higher and higher rates, so we can promote the metrics and show
how great the results are to drive further adoption.”

“Results speak for themselves,” Krishnan commented. “With accurate data that provides true
visibility into our supply, demand, and skill proficiency, the system’s recommendation engine will

Kellogg School of Management 7

This document is authorized for use only in Prof. Shubhi Gupta's Managing Talent and Career in Organizations_T6 at FORE School of Management from Jan 2023 to Jul 2023.
 A I - P ow e r e d T a l e n t M a n ag e m e n t at Infosys KE1222

be able to shift from micro-optimization within specific delivery teams to macro-optimization

across the entire organization. As soon as delivery managers start benefiting from new people who
have the skills they requested, they’ll recognize the system’s value and speed—and they’ll never
look back.”

“It would be great if people would want to use it voluntarily,” Gupta said. “But what if delivery
managers don’t choose to use it and try to hoard their talent? We have encountered this problem in
the past. Giving managers an option to use the system can be ineffective. It allows for a wide range
of acceptable usage and creates ambiguity about what needs to be accomplished.”

Pandey interjected. “Speaking of accomplishment, we need to define key metrics so we can

measure success, develop a baseline, and continue to drive adoption,” she said. “We could measure
employee satisfaction and manager satisfaction by conducting a survey four weeks after a match
recommendation is executed. It would help us assess how effective our predictions are. We could
work with HR to incorporate performance appraisals so we can assess if better matches were
correlated to better job performance. We would need to identify a good control group though.
Maybe we should even bring clients into the feedback loop.”

“We have some obvious choices for quantitative metrics here,” Gupta said. “Lowering the
rejection rates for recommendations seems like a fairly direct measure of whether the system is
working as expected. Ideally, we should also be able to reduce the amount of time employees
are idling on the bench and get more of them actively moving from one project to the next
without downtime. Turnaround time to an accepted recommendation seems to be another
important metric.”

“I was thinking about the possible effect on employee attrition,” Krishnan responded. “Better
talent mobility and matching should reduce our attrition rates. High attrition impacts company
culture, hurts morale, and leads to the added cost of hiring and retraining new employees. I truly
believe this system can have a transformational impact across our entire organization.”

“And who will have access to this all-encompassing system?” Gupta asked. “Hiring managers
certainly need to, but what about clients, engagement leads, and individual employees who are
asked now and then to update data? We’ll need to be very careful about privacy and consent issues.”

Conclusion
As Gupta walked across the Infosys campus later that evening on his way home, he knew he
and his team still had much to consider.

Logic suggested that the more data the system had, the better its recommendations should
be. The project team had access to hundreds of variables, each of which could contribute a fuller
picture of the employee and client project, resulting in a higher rate of match acceptances. Yet
some of the data was likely not usable—some variables might introduce unintentional bias into the
algorithm or might be too sensitive to include. The way Gupta saw it, three options were available.
Option A would be to use as many variables as possible and let the models select the statistically

8 Kellogg School of Management

This document is authorized for use only in Prof. Shubhi Gupta's Managing Talent and Career in Organizations_T6 at FORE School of Management from Jan 2023 to Jul 2023.
 KE1222 A I - P ow e r e d T a l e n t M a n ag e m e n t at Infosys

relevant ones. Option B would be to be conservative at the start, and only select the current job
situation and recent engagement variables, setting aside all individual preferences and historical
engagement variables. And Option C would be to spend some time pre-processing the data, which
would entail combining, converting, or eliminating variables so the project team could identify the
most influential and relevant variables.

Whatever data variables were chosen, the project team had to ensure the data itself was reliable,
accurate, and consistent. The old talent allocation system, under which managers could hoard
talent by reserving certain employees for future projects or refusing to rotate them off existing
projects, had to be erased from institutional memory and replaced with the new system—but
how to do this? One option would be for the project team to issue a series of mandates: that both
employees and managers update skills at appointed times; that managers must use the system
for each staffing request; and that if managers rejected a recommendation, they would have to
resubmit a revised staffing request until the system provided a recommendation they accepted. It
might be difficult, however, to achieve compliance simply by mandating it.

Perhaps the team should focus instead on making users want to use the new system, creating
positive word of mouth by publicizing early wins and making the interface as easy to use. High
match rates from the start would seem necessary for this option. A third possibility might be
striking a balance between the voluntary, free-will approach and the draconian mandate—perhaps
starting with a proof of concept with one or two teams. Instead of a general commitment to “use
the system,” these teams could be armed with the steps required to make decisions based on the
predictions, and the project team could support them with specifically designed objectives.

Gupta knew the success of an AI initiative was rarely about the AI; it usually came down to
the people and the politics of an organization. The project team’s choice of data strategy had to
consider long-term adoption, compliance, and maintenance. Yet he wanted to ensure that his
team did everything possible to build a talent allocation system that would be embraced by clients,
staffing decision-makers, and employees alike.

Kellogg School of Management 9

This document is authorized for use only in Prof. Shubhi Gupta's Managing Talent and Career in Organizations_T6 at FORE School of Management from Jan 2023 to Jul 2023.
 A I - P ow e r e d T a l e n t M a n ag e m e n t at Infosys KE1222

Exhibit 1: Talent Management Solution-Product Vision

Ta le n t - M at c hing O u t pu t

Ta le n t De t ail s

Source: Infosys internal documents.

10 Kellogg School of Management

This document is authorized for use only in Prof. Shubhi Gupta's Managing Talent and Career in Organizations_T6 at FORE School of Management from Jan 2023 to Jul 2023.
 KE1222 A I - P ow e r e d T a l e n t M a n ag e m e n t at Infosys

Exhibit 2: Process Diagram for Creating and Maintaining the Talent

Management System

Collecting the Data

• What data is available?
• How is the data
reported/collected?

Measuring Success and Assessing Data Quality

Continuously Improving the • How recent is the data?
System
• How accessible/available
• How will success be is the data?
measured? • How accurate is the data?
• Who will be responsible • How consistent is the
for maintaining the system data?
after launch?
• How relevant is the data?
• How will the system be
updated/retrained for
continuous improvement?

Adoption and Compliance Defining the Variables

• How to ensure buy-in across • Which variables should be
the organization? included?
• Which variables should be
excluded?

Source: Created by the authors.

Kellogg School of Management 11

This document is authorized for use only in Prof. Shubhi Gupta's Managing Talent and Career in Organizations_T6 at FORE School of Management from Jan 2023 to Jul 2023.
 A I - P ow e r e d T a l e n t M a n ag e m e n t at Infosys KE1222

Exhibit 3: Sample of Supply Variables under Consideration

Current Job Situation Individual Preferences Historical Engagement Recent Engagement
• Current role • Ethnicity/nationality • Length of service • Time in current position
• Salary • Language(s) spoken • Amount of leave taken • Time in current grade
in past year
• Projected end date of • Gender • Time since last leave
current project • Amount of sickness in
• Age • Time since last salary
current/previous years
• Current department change
• Disability
• Average time in each
• Current location • Time until contract
• Home city/country salary level expiration
• Current appraising • Preferred locations • Average time with each
manager • Time until next review
• Willingness to travel/ manager cycle
• Current manager relocate • Number of manager
performance rating • Time until work permit
• Educational background changes in past 5 years or visa expiration
• Previous manager
• Marital status • Training courses
performance rating
completed in past 2
• Self-tagged skills • Normal working hours years
• Manager-tagged skills • Positions applied for in • Certifications completed
past 2 years in past 2 years
• Positions rejected • Background check,
• Career/skill aspirations including confirmation
that employee has not
worked for a client
competitor

Source: Created by the authors.

12 Kellogg School of Management

This document is authorized for use only in Prof. Shubhi Gupta's Managing Talent and Career in Organizations_T6 at FORE School of Management from Jan 2023 to Jul 2023.
 KE1222 A I - P ow e r e d T a l e n t M a n ag e m e n t at Infosys

Exhibit 4: Sample of Demand Variables under Consideration

Client Context Current Engagement Historical Engagement
• Client name • Current project with • Number of past projects
client with client
• Number of years as
client • Company project • Number of current
engagement lead projects with client
• Client location
• Project manager • Client satisfaction rate
• Client size
• Employees on project • Amount of budget
• Client market cap
variance from initial
• Project scope
• Client revenue projections
• Required skills
• Client industry • Number of projects with
• Staffing start date extended staffing end
• Language(s) required
• Staffing end date dates
• Project budget • Number of projects that
required additional staff
• Billing rate during their span
• Outside vendors
involved in project

Source: Created by the authors.

Kellogg School of Management 13

This document is authorized for use only in Prof. Shubhi Gupta's Managing Talent and Career in Organizations_T6 at FORE School of Management from Jan 2023 to Jul 2023.