You are on page 1of 9

XLRI Jamshedpur

Lean principles, learning, and


knowledge work: Evidence from a
software services provider
ORM2 Assignment Submission

Submitted by:
GROUP No. 2 Sec- C
Amitabh Vajpayee (B13131)
Anand Odedra (B13132)
Anirban Chakraborty (B13134)
Anup Joshi (B13136)
Archit Singh (B13137)
Anupam Maity (B13183)


SECTION 1

Which topic/topics covered in ORM2B13-3 is discussed in the article? What is the primary
purpose for this research?

The topics discussed in the article which were covered in ORM2B13-3 are Lean Manufacturing &
Toyota Production System (TPS).
Lean manufacturing, or lean production, often simply, "lean", is a production practice that
considers the expenditure of resources for any goal other than the creation of value for the end
customer to be wasteful, and thus a target for elimination. Working from the perspective of the
customer who consumes a product or service, "value" is defined as any action or process that a
customer would be willing to pay for.



The primary purpose for this research is to test the applicability of the Lean Production System
to any Indian software services firm, through quantitative and qualitative research. The
quantitative analysis is primarily hypothesis-driven: problem definition, generation of candidate
solutions and evaluation/selection of a solution. It tries to identify as to what extent tasks be
specified within the context of knowledge work. The paper also tries to identify the significant
challenges to using ideas from lean production in a knowledge-based industry and to examine
how the challenges identified above are to be overcome.
Through empirical analysis, the comparison between lean and non-lean projects is made to
understand as to the degree of waste reduction induced in the software services firm by using
lean manufacturing system. Specific aspects of knowledge work- task uncertainty, process
invisibility, and architectural ambiguity- are discussed to question the relevance of lean
production in this setting.
SECTION 2

What was the research methodology used to carry out the research? You may explain
tool/technique/method/optimization model along with the results.
The series of methodologies used in carrying out the research are:
1. SELECTION OF THE TARGET ORGANIZATION
The selection of the organization on which the research was to be conducted, was done on the
basis of 4 characteristics.
The organization must compete in a knowledge work industry. This condition is
important as this introduces the challenge that much of the work being completed must
be invisible to observers.
The industry must have high task uncertainty, arising from number of sources, including
environmental change or customer involvement in production.
The architecture of the work should not be set.
The lean initiative progress should occur along with the research study. This way the
researchers will be able to view the process taking place, as opposed to just the
outcome.
Based on the above characteristics, Wipro was selected as the target organization.
2. DATA COLLECTION & ANALYSIS
Data was collected using multiple methods, including interviews, meeting observation,
inspection of internal documents, and analysis of archival project data. A semi-structured
interview format was followed to elicit consistent information across respondents. The data
collected was then categorised based on the four basic principles identified in the lean
manufacturing.
Iteration between fieldwork and data analysis permitted the researchers to empirically ground
their observations and identify anomalies to advance the process. Also, quantitative project
data for both lean and non-lean projects was analysed. The database of projects containing
detailed information, permitted evaluation of performance of lean projects with respect to
contemporaneous non-lean projects.
3. PERFORMANCE VARIABLES
In the research methodology, a number of performance variables have been used to quantify
and compare the performance between lean projects and non-lean projects.
Schedule Deviation Before the beginning of a project, both the schedule and effort are
estimated and agreed by the customer. The revised estimates for both schedule and
effort deviation are used in calculations, as these most accurately reflect a projects final
objectives.
Schedule Deviation = (Date Delivered Scheduled Date Due)(Scheduled Date Due
Start Date)

Effort Deviation It captures total number of hours expended by a team on its project.
Effort Deviation = (Actual Effort Estimated Effort)/Estimated Effort
The measure is then normalized as effort deviation can be larger for large projects.

Quality The number of defects in Customer Acceptance Testing (CAT) divided by KLOC,
has been used as the metric for quality measurement. Every project does not complete
CAT, so the sample data is reduced to 50 lean projects.

Lean Project An indicator variable is used, coded as 1 for lean projects and 0 for non-
lean projects.

Total effort, duration, and complexity Increasing effort, duration and complexity
decreases the operational performance, which are controlled for using estimated
number of project hours, calendar days required for the project, and actual kilolines of
manual code written for the project.

4. PERFORMANCE EVALUATION

For evaluating the performance of projects, three samples are examined.
Comparison of lean to all non-lean projects- Two matched samples are decided for both lean
and non-lean projects, by matching on the basis of three traits: SBU, contract type and end year.
Then the Euclidean distance model is used, which requires minimizing the Euclidean distance
using the formula containing the variables, total effort, duration and KLOC.

A second sample is constructed by excluding KLOC and then following the same process as
above. This enables evaluation of all lean development projects among the data.
Also, since some projects might not have gone through acceptance testing, not all matches
would have had quality values. So, restricting the pool of data so that all matches have quality
values will be the way to go.

5. TESTING HYPOTHESIS

Non-parametric tests have been used to test the hypotheses, as there is no reason to assume
normality, and the sample size being small for some tests. Wilcoxon rank-sum tests the
hypothesis with respect to schedule deviation to test the average values between lean and non-
lean projects. For the matched smaples, Wilcoxon matched-pairs signed-ranks test is used to
test significance of the sample without KLOC.
Also, Levenes test for equality of variance tests the difference in variances of all the three
samples. It rejects the null hypothesis that the two groups have equal variances in all three
samples, indicating the variance in case of lean projects to be much lower than that of non-lean
projects.

6. RESULTS

The data presented provides support that the lean initiative has positively impacted operational
performance at Wipro. The lean projects have shown better schedule and effort performance
than non-lean projects. However, there is no significant difference in quality. In terms of quality,
the mean defect rate of lean projects is lower compared to all non-lean projects, however, the
differences are not statistically significant.



SECTION 3
Main findings of the research and its implication on the society/industry:
The research has attempted to analyse the applicability of the classical lean production
methodology and its associated practices in the context of unstructured knowledge work.
The major findings of the research are as follows:
1) Lean production requires that both behaviours and outcomes are specified. The lean
method as exemplified by Toyota says that the architecture/structure should be
specified first, followed by the connections, and finally the process itself. The research
finds that Wipros implementation of lean technique shows progress with respect to
the first two types of behavioural specification but little in the final point i.e. the work
itself. Prior work outside of lean production notes that the appropriate investment in
task specificity depends on how often the task is repeatedly done. In the case of
software services, tasks are unique and tend not to be repetitive. When the structure of
work is unknown and flexible, the task specificity reduces significantly.
2) Problem identification is often the most critical step and they should be identified as
frequently as possible, and as early as possible. Identification of problems in the
context of knowledge work is often a social exercise. Problems are socially complex and
not easily identifiable.
Taiichi Ohno, the man behind the Toyota Production System, noted, To get rid of waste,
train your eyes to nd waste and then think about how to get rid of the waste youve
found. Do this over and over again, always, everywhere, relentlessly and unremittingly.
Identification of the problem is often the most difficult and critical part of solving the
problem, and more so when the work is unstructured, invisible and highly varying from
project to project. Over time, Toyota has built a system that can identify problems
relatively automatically, but this took many years of experience making incremental
innovations (Fujimoto, 1999).
To apply lean production in a sufciently novel setting, one would expect similarly
signicant effort to be necessary to learn how to identify problems in the new context.
The costs of not being able to identify a problem early can be huge. As in the case of
manufacturing, rectifying a problem becomes more difficult and costlier as the point of
identification moves down the value stream. Similarly, in the case of knowledge work,
the costs of rework and scrap increase exponentially as we proceed in the project
timeline.
3) Problems and solutions form a strict pair with respect to time, space, and person. The
Toyota production system advocates for continuous identification and resolution of
problems. Toyota has structured the problem solving process to minimize chances of
ambiguity and variability and maximize the chances of success. Lean production system
highlights the importance of using the scientific method of problem identification and
solving.
This article finds that in order to make the problem solving exercise successful, it is
essential to keep problems and solutions together in person, time and space. This is an
extension of the common logic that knowledge about a task comes from actually doing
and completing the task; therefore pairing of information and action at the same time
enables more effective learning and improvement. In the case of software code reviews
at Wipro, the article has found that keeping the problem and solution bound together in
a single person can also prevent an engineer from repeating the mistake.
Along with the person dimension, space i.e. location also plays an important role in
problem solving, due to the contextual and essential knowledge that is usually
embedded there. For example, some problems cannot be solved unless the
development engineer actually travels to the clients site.
Finally, the temporal dimension is also equally important. By creating opportunities for
individuals to solve problems immediately after the occurrence, the likelihood of
successful resolution increases. This draws analogy from the use of Andon cords at a
Toyota assembly line, where the worker stops the line to correct the problem as soon as
it is detected. If there is a time lag between identification and solving, some of the
information relating to the problem might be lost or modified, which will affect the
solvability of the problem.



IMPLICATIONS: The implications of the research findings can potentially be huge. The
software services industry is huge. According to market researcher Data Monitor, the
size of the worldwide software industry in 2008 was US$303.8 billion, an increase of
6.5% compared to 2007. So, if successfully implemented on a large scale, the savings and
efficiency increase due to implementation of lean systems in the software industry can
be huge. This case presents a classic example where the best practices in a particular
industry (automotive) can be adapted and furthered by another (software services).
Also, the findings of the merits of lean systems in knowledge work can be extended to
outside the software services industry eg. Medical services, art productions etc.

The above described findings of the research article are a new learning because it is for
the first time that the postulates and processes of the Toyota Production system has
been studied and analysed in the context of knowledge work. The TPS originated in a
routine manufacturing context, where the work is characterized by high analysability
and low task variability. Knowledge work, in contrast, is characterized by high variability
and low analysability. Hence, the applicability of lean production methods in a
completely different concept is worth studying. The research found both similarities and
differences between the implementation of a lean system in manufacturing industry and
software services industry. The major findings have been listed and described above.


SECTION 4
What are the limitations of this research as per the article?
The article says that one should be cautious in applying its results. This study examines the
applicability and implementation of lean production in knowledge work, by investigating the
experience at Wipro. It is possible that we cannot generalize the findings to other settings. It
also says that while Wipro has reaped the benefits of lean production, it still has a long way to
go to go realize the complete benefits of lean. This limitation is a necessary but unattractive
consequence of the detail and lack of recall bias that the real time nature of the study permits.
Finally, the basic premise that what Wipro has done is truly lean is questionable. Lean
production derives from a manufacturing context and analogies to knowledge work do not exist.
Since no definition of lean in software is accepted, it has been assumed that Wipro is
consciously trying to create a system for software services that imitates some of the unique
ways of working at Toyota.

What are the limitations of this research beyond what is already mentioned in the article?
In addition to the limitations listed by the authors, we feel that there are a few more that can be
noted.
1) This article only considers KLOC as the only indicator of project complexity. However
we feel that isnt completely accurate. KLOC isnt the single indicator for complexity.
Other indicators can be how frequently and to what extent the customers demands are
changing i.e. dynamicity of customer requirements.
2) To evaluate lean projects performance, the authors have examined three samples. First,
they have compared lean to all non-lean projects. Because such a comparison does not
take into account differences in project traits, they have constructed two matched
samples. In the rst sample, we identify control projects based on six factors (traits):
SBU, contract type, end year, total effort, duration, and KLOC. They have matched
exactly on the rst three traits, then selected the matched project by minimizing the
Euclidean distance using the formula:



We however feel that the performance evaluation metric can be a weighted mean of
the modulus instead of a RMS (root mean square) value. Here, equal weights have been
assigned to the three different factors. A better way of doing it would be to find
appropriate weights and then taking a weighted average of the modulus of the
differences between effort, duration and quality of lean and non-lean projects.

3) Since all projects do not complete acceptance testing, not all matches have a quality
value. The authors have repeated the tests, restricting the pool so all matches have
quality values, and obtain similar results. This could have noise effects on the accuracy
of the findings.
4) There could be a possibility of Hawthorne Effect. The Hawthorne effect (also referred to
as the observer effect) refers to a phenomenon whereby workers improve or modify an
aspect of their behaviour in response to the fact of change in their environment, rather
than in response to the nature of the change itself. The article says that this does not
seem too likely, given Wipros continuous focus on process improvement. Both
interviews and observation conrmed that lean projects at Wipro did not receive
signicantly more attention or resources than did non-lean projects. The authors have
defended their argument by saying that as the number of projects increases, the
likelihood of a Hawthorne Effect decreases.
However we feel that the likelihood of Hawthorne Effect is significant. Possible
explanations for the Hawthorne effect include the impact of feedback and motivation
towards the experimenter. Receiving feedback on their performance may improve their
skills when an experiment provides this feedback for the first time. Research on the
demand effect also suggests that people may be motivated to please the experimenter,
at least if it does not conflict with any other motive. They may also be suspicious of the
purpose of the experimenter. Therefore, Hawthorne effect may only occur when there is
usable feedback or a change in motivation.
5) We feel that in addition to the factors considered by the author, other factors could
have been considered and factored into the calculations while assessing the effects of
lean production on performance.
Here, the authors have drawn highly upon the findings of Spear and Bowen as published
in the article Decoding the DNA of the Toyota Production System. However, in
addition to the four major characteristics of a lean system, there can be others like
degree of system flexibility, and responsiveness to change.
The applicability of concepts like Kanban and the mapping of the wastes into the
seven categories as pointed out in Taiichi Ohno, within a software services context could
have also been explored.

You might also like