OPERATIONS MANAGEMENT

Case Analysis : INFOSYS BPO

Submitted by :
a) Dheeraj Subramanian (1012019)
b) Milana Jagadeesh (1012029)
c) Balachander Vasudevan (1012015)
d) Sachin Vasudeo Bapat (1012052)
e) Nirav Kumar (2009037)

1- A ) Dedicated Resources for each of the six queues:

Assumptions:

We have assumed that resources will work for 8 hours a day and there is no
provision for overtime work. We have considered effect of overtime in part 2)

Turn around time of 1 day for Q1, Q3 and Q5 is strictly enforcable and hence
the number of resources needs to be selected based on this condition.

Since the volume of requests is pretty high, we have used Average Handling
Time to do the analysis.

Exhibit 11 provides sample arrival pattern for the widgets. It shows that
volume of requests on one particular day can be much higher than others. We
have used this sample pattern for our analysis.

Calculation for simple widgets ( Q1, Q3 and Q5)

Number of resources = (Number of requests * AHT ) /
(60*60*8)

Number of man-days required to handle sample pattern

volume:
REVIEW
Q1
1
180
2
20.88
4.29
2.29
2.84
4.93
21.00

TAT
AHT
Priority
Day1
Day2
Day3
Day4
Day5
Resources

PROCESS
Q3
1
135
1
7.35
1.68
1.73
2.23
0.02
8.00

Resource Utilization with this scheme: Since we are planning for highest volume in
sample pattern, resource utilization on non-peak days is very low.

TAT

REVIEW
Q1
1

PROCESS
Q3
1

SETTLE
Q5
1

AHT
Priority
Day1
Day2
Day3
Day4
Day5

180
2
0.99
0.20
0.11
0.14
0.23

135
1
0.92
0.21
0.22
0.28
0.00

180
1
0.96
0.12
0.29
0.36
0.28

Calculation for complex widgets ( Q2, Q4 and Q6)

Turn around time for complex widgets is 4 days. Hence even if volume is
higher on a particular day, we can work on those requests for the next 4 days.
Following policy is used for calculation:
-

Resources should work with FIFO policy. i.e. on any particular day, they should
work with oldest backlog first.

If they complete all the backlog on a particular day, they should start
processing the requests received on the same day.

The policy will work as long as every request is completed within allowed 4
days. This means that if backlog is continuously growing, we will end up with
TAT of more than 4 days.

Formule used for complex widget resolution:

Available capacity = Maximum Capacity Backlog
Current Backlog = Todays work requests - Available capacity

Example : For Q2 , If we use only 1 resource, backlog will slowly start building as
shown in the table on the left. With 2 resources, backlog will be zero, and hence we
will be able to meet TAT requirements.

Resource

s
Today's
Avail Cap
work
1.77
1.03
1
0.93
0.8
1.77
1.03
1
0.93
0.8
1.77
1.03
1
0.93
0.8
1.77
1.03
1
0.93
0.8
1.77
1.03
1
0.93
0.8
1.77

Backlog
1
0.23
0.2
0.2
0.27
0.47
-0.3
-0.33
-0.33
-0.26
-0.06
-0.83
-0.86
-0.86
-0.79
-0.59
-1.36
-1.39
-1.39
-1.32
-1.12
-1.89
-1.92
-1.92
-1.85
-1.65

( Backlog in continuously increasing)

To
0.77
0.8
0.8
0.73
0.53
1.3
1.33
1.33
1.26
1.06
1.83
1.86
1.86
1.79
1.59
2.36
2.39
2.39
2.32
2.12
2.89
2.92
2.92
2.85
2.65
3.42
( No issues with backlog)

Similarly we can calculate for Q4 and Q6 as below:

( Q4 )
Resourc
es
Today's
Avail Cap
work
4.46
3.82
6.50
8.36
6.24
4.46
3.82
6.50
8.36
6.24
4.46
3.82
6.50
8.36
6.24
4.46
3.82
6.50
8.36
6.24
4.46
3.82
6.50
8.36
6.24
4.46

Q6 )

Re

Backlog
6
6
6
5.50
3.14
2.90
4.44
6.00
5.50
3.14
2.90
4.44
6.00
5.50
3.14
2.90
4.44
6.00
5.50
3.14
2.90
4.44
6.00
5.50
3.14
2.90

Tod
0
0
0.50
2.86
3.10
1.56
0.00
0.50
2.86
3.10
1.56
0.00
0.50
2.86
3.10
1.56
0.00
0.50
2.86
3.10
1.56
0.00
0.50
2.86
3.10
1.56

1- B) Common pool of resources for each of the function:

In this case,
-

All the simple requests received for this process should be completed in one
day. Hence number of resources should be enough to handle peak simple
widgets.

Same resources can work on complex widgets when they have free
bandwidth.

Policy should be
o

Complete all the simple widgets received on that day.

Once simple requests are complete, work on backlog for complex

requests in FIFO order.

If all the backlog for complex requests is finished, start working on

complex requests received on that day.

This can be put in tabular format as shown below.

Resources

21

Q1 work

Q2 work
20.88
4.29
2.29
2.84
4.93
20.88
4.29
2.29
2.84
4.93
20.88
4.29
2.29
2.84
4.93
20.88
4.29
2.29
2.84
4.93
20.88

Avail cap
1.77
1.03
1.00
0.93
0.80
1.77
1.03
1.00
0.93
0.80
1.77
1.03
1.00
0.93
0.80
1.77
1.03
1.00
0.93
0.80
1.77

Backlog
0.13
15.07
18.71
18.16
16.08
0.13
15.07
18.71
18.16
16.08
0.13
15.07
18.71
18.16
16.08
0.13
15.07
18.71
18.16
16.08
0.13

1.64
0.00
0.00
0.00
0.00
1.64
0.00
0.00
0.00
0.00
1.64
0.00
0.00
0.00
0.00
1.64
0.00
0.00
0.00
0.00
1.64

4.29
2.29
2.84
4.93

1.03
1.00
0.93
0.80

15.07
18.71
18.16
16.08

0.00
0.00
0.00
0.00

1- C) Common pool of resources for each category:

For simple requests, the number of resources should be enough to handle peak
demand on a particular day.
Q1
work
20.88
4.29
2.29
2.84
4.93
20.88
4.29
2.29
2.84
4.93
20.88
4.29
2.29
2.84
4.93
20.88
4.29
2.29
2.84
4.93
20.88
4.29
2.29
2.84
4.93

Resources
Q3 work

37
Q5 work
7.35
1.68
1.73
2.23
0.02
7.35
1.68
1.73
2.23
0.02
7.35
1.68
1.73
2.23
0.02
7.35
1.68
1.73
2.23
0.02
7.35
1.68
1.73
2.23
0.02

Q1+Q3+Q5
8.66
1.07
2.57
3.23
2.55
8.66
1.07
2.57
3.23
2.55
8.66
1.07
2.57
3.23
2.55
8.66
1.07
2.57
3.23
2.55
8.66
1.07
2.57
3.23
2.55

For complex requests, resources need to ensure 4 day TAT for all the
requests.

36.89
7.03
6.59
8.30
7.49
36.89
7.03
6.59
8.30
7.49
36.89
7.03
6.59
8.30
7.49
36.89
7.03
6.59
8.30
7.49
36.89
7.03
6.59
8.30
7.49

Resources

21

Q2
Q4 work
work
1.77
1.03
1.00
0.93
0.80
1.77
1.03
1.00
0.93
0.80
1.77
1.03
1.00
0.93
0.80
1.77
1.03
1.00
0.93
0.80
1.77
1.03
1.00
0.93
0.80

Q6 work
4.46
3.82
6.50
8.36
6.24
4.46
3.82
6.50
8.36
6.24
4.46
3.82
6.50
8.36
6.24
4.46
3.82
6.50
8.36
6.24
4.46
3.82
6.50
8.36
6.24

Q2+Q4+Q6
12.39
12.84
20.37
8.39
11.60
12.39
12.84
20.37
8.39
11.60
12.39
12.84
20.37
8.39
11.60
12.39
12.84
20.37
8.39
11.60
12.39
12.84
20.37
8.39
11.60

1- D) Common pool of resources for all six queues:

Avail C
18.61
17.70
27.88
17.67
18.64
18.61
17.70
27.88
17.67
18.64
18.61
17.70
27.88
17.67
18.64
18.61
17.70
27.88
17.67
18.64
18.61
17.70
27.88
17.67
18.64

This is the most flexible system since load can be shared across all the resources.
This will result in much higher utilization as shown below. But all the resources need
to be trained on all the 6 processes.

Resourc
es
Q1+Q3+ Q2+Q4+Q6
Q5
36.89
7.03
6.59
8.30
7.49
36.89
7.03
6.59
8.30
7.49
36.89
7.03
6.59
8.30
7.49
36.89
7.03
6.59
8.30
7.49
36.89
7.03
6.59
8.30
7.49

37

Avail cap
18.61
17.70
27.88
17.67
18.64
18.61
17.70
27.88
17.67
18.64
18.61
17.70
27.88
17.67
18.64
18.61
17.70
27.88
17.67
18.64
18.61
17.70
27.88
17.67
18.64

2- A) Benefit of overtime:

Backlog
0.11
11.47
24.18
25.00
29.51
0.11
11.47
24.18
25.00
29.51
0.11
11.47
24.18
25.00
29.51
0.11
11.47
24.18
25.00
29.51
0.11
11.47
24.18
25.00
29.51

18.
6.
3.
0.
0.
18.
6.
3.
0.
0.
18.
6.
3.
0.
0.
18.
6.
3.
0.
0.
18.
6.
3.
0.
0.

The sample pattern suggests that the demand can be particularly high on few days.
If we allocate resources by considering those days, resource utilization will be poor
on other days. This issue can be addressed if employees can work overtime on days
with high number of requests.
Following table shows modified resource utilization from 1-D. We have 32 regular
employees who can work 25% overtime on a day if number of requests are high.
This will improve resource utilization on non-peak days, but company had to incur
overtime cost.

Resourc
es
Q1+Q3+ Q2+Q4+Q6
Q5
36.89
7.03
6.59
8.30
7.49
36.89
7.03
6.59
8.30
7.49
36.89
7.03
6.59
8.30
7.49
36.89
7.03
6.59
8.30
7.49
36.89
7.03
6.59
8.30
7.49

32

Avail cap
18.61
17.70
27.88
17.67
18.64
18.61
17.70
27.88
17.67
18.64
18.61
17.70
27.88
17.67
18.64
18.61
17.70
27.88
17.67
18.64
18.61
17.70
27.88
17.67
18.64

Backlog
3.11
9.47
17.18
13.00
19.84
3.11
9.47
17.18
13.00
19.84
3.11
9.47
17.18
13.00
19.84
3.11
9.47
17.18
13.00
19.84
3.11
9.47
17.18
13.00
19.84

15.
8.
10.
4.
0.
15.
8.
10.
4.
0.
15.
8.
10.
4.
0.
15.
8.
10.
4.
0.
15.
8.
10.
4.
0.

Cost Analysis:
Following table compares employee cost in scheme 1-D against the one with
overtime. There is cost saving even after including overtime pay on 5 days a
month.
Resources
Cost
Cost

Normal
Normal
Overtime
Total

Days/month
37
32
40

20
15
5

2- B) Benefit of modifying the SLA:

The main issue for resource planning is highly uneven volume of simple
requests on few days. The SLA forces to complete all the simple requests in one day.
The company will benefit greatly if it can renegotiate TAT for simple requests. One
proposal could be to increase TAT to 2 days if number of simple requests exceed the
predefined limit.
Following table shows the above effect. If total number of simple requests
exceed 30000, then we are ensuring 1 day TAT for first 30000 requests and 2 day
TAT for remaining ones. Total number of resources required in 34 only and resource
utilization is pretty high.

Resourc
es
Q1+Q3+ Q2+Q4+Q6
Q5
30.00
13.92
6.59
8.30
7.49
30.00
13.92
6.59

34

Avail cap
18.61
17.70
27.88
17.67
18.64
18.61
17.70
27.88

Backlog
4.00
5.47
15.18
13.01
21.84
4.00
5.47
15.18

14.
12.
12.
4.
0.
14.
12.
12.

8.30
7.49
30.00
13.92
6.59
8.30
7.49
30.00
13.92
6.59
8.30
7.49
30.00
13.92
6.59
8.30
7.49

17.67
18.64
18.61
17.70
27.88
17.67
18.64
18.61
17.70
27.88
17.67
18.64
18.61
17.70
27.88
17.67
18.64

13.01
21.84
4.00
5.47
15.18
13.01
21.84
4.00
5.47
15.18
13.01
21.84
4.00
5.47
15.18
13.01
21.84

4.
0.
14.
12.
12.
4.
0.
14.
12.
12.
4.
0.
14.
12.
12.
4.
0.

Employee Cost :
Resources
Cost

Normal

Days/month
34

20

Summary
Scheme

Number of Resources

Dedicated for
each Queue

60

Dedicated stage
wise

47

Dedicated
category wise

58

Common for all

queues

37

Common for all

queues +

32 + overtime

Overtime
Common for all
queues+

34

Modified SLA

Pricing Model
FTE based billingThis model does not create incentives for the service provider because irrespective
of the efficiency of the resources, client pays it based on the number of resources
employed. This is extremely beneficial when transaction volumes are difficult to
forecast. Generally under this billing model the service provider would like to charge
the client for dedicated resources for each queue. Thus there would not be much
efficiency built into the resources. The resources will be trained for a single queue
and would continue to excel in the process related to that queue. Hence in our case
Infosys would like to charge client for 60 resources per month. This helps Infosys to
discharge all the requests as per the agreed SLA and take care of daily variances in
the transaction volumes. Thus operations would be designed queue wise and
trained/dedicated resources for the processes in that queue would be assigned.

Transaction based billingThis model creates incentives for the service provider to build efficiency in the
system because the billing is based on number of the transactions processed. Under
this model the service provider would ensure that resources are trained across all
the processes and have the capability to contribute to any process as per the need.
This benefits the service provider as he is able to deploy optimum number of
resources and achieve the agreed SLA for a specified transaction volume band. In
our case Infosys would like to deploy 37 resources for the client CreditPlus. If

overtime can be managed by Infosys then it would like to deploy 32 resources for
the client CreditPlus. Thus operation would be designed wherein the resources are
knowledgeable on all the processes within all the queues. Based on the demand of
transaction and agreed SLA, the resource will be assigned to the critical queue and
then would be shifted to another queue once the transaction is complete.

Outcome based billingThis model creates incentives for the service provider to build efficiency in the
system because the billing is based on the business results achieved by the client.
The business results could be increased profits or reduced operating expenses. By
taking up this task at an offshore location in India, Infosys can leverage cost
arbitrage .It can also train employees across processes and check quality at regular
intervals. This can help reduce the operating expense as employees trained across
process can take up any activity on queue which is restricted by SLA. Under this
model the service provider would like to reduce the turn -around time. This could
be achieved by running 24-hour at an offshore location and building priority in the
system. This would obviously results in profits for clients as they(clients) would be
ahead of competition in confirming back with their customer with regards to the
outcome of transaction. . In our case Infosys would like to deploy 37 resources for
the client CreditPlus. If overtime can be managed by Infosys then it would like to
deploy 32 resources for the client CreditPlus. Thus operation would be designed
wherein the resources are knowledgeable on all the processes within all the queues.
Based on the demand of transaction, agreed SLA ,and the TAT expected, the
resources would be assigned to the critical queue and then would be shifted to
another. The priority of the queue would play a significant part in deciding criticality
of queue in a process .Thus assignment of resources to the corresponding queue will
happen based on priority so that the resolution time is reduced and benefit of faster
resolution is achieved. This would also help to reduce the average case resolution
time. This would result in increased profits for the client.

Productivity Improvement
The suggested productivity improvement over the lifecycle of the project can be
achieved through following the below approach:
Infosys can employ educated workforce, at an offshore location, as soon as they
pass out of class 12.Such workforce will come cheaper and provide higher cost
arbitrage. This workforce will also have higher capability to undergo overtime. This
cost arbitrage can provide benefit to the tune of 20%-35%.Secondly such a
workforce would have higher capability to learn. They can be trained across
processes (ie Review,Process,Settle) ,leading to increased productivity, as they can
be assigned to any queue based on the priority of queue, SLA expected and TAT
planned. The productivity improvement can be very well visualized through the
adaption of training for employees. As we have calculated earlier that if the

resources are proficient in handling just one queue then based on the volume of
transaction, the number of resources required is 60.On the other hand if the
resources are trained to work across processes then the number of resources
required is 37.Thus the productivity improvement with training is to the tune of (6037)/60=38.33%. In a similar way if the resources are trained stage wise then the
productivity improvement is to the tune of (60-47)/60=21.6%.On similar lines if the
resources are trained category wise then the productivity improvement is to the
tune of (60-58)/60 =3.3%.Hence it can be observed that maximum
productivity improvement happens if the resources are trained across all
processes. Thus use of multi-skill resources provides considerable resource
efficiency. This coupled along-with workflow management system and a command
center can define the tasks the workforce should undertake on any given day. The
workflow management can be designed in such a way that it combines the priority
of the queue, the SLA expected, TAT expected, to decide the task for the workforce.
As an example as per Exhibit 11 ,if there are tasks on queue 1 and queue 3 and
both have TAT of 1 day then the decision should be based on priority. Hence
workforce should be first assigned to complete transactions on Queue3.Thus the
workflow management system should be designed to take care of all such criteria
and define the tasks for workforce. This would also enable accurate forecast.
Turnaround time can also be reduced through Standardization. Variance in tasks are
handled by separating them into different queues .Clubbing of activities into queue
is already done and 6 separate queues are designed. Automation of activities
through development of a common platform for workflow helps achieve further
improvement in productivity.
All-together productivity improvement to the tune of 40-50% can be achieved
through training of resources across processes(use of multi-skill resources),
workflow management, standardization and Automation of activities.