A STUDY ON SOFTWARE PROJECT COACHING MODEL

USING TSP IN SAMSUNG

Taehee Gwak and Yoonjung Jang

SE Team, Software Laboratories, Samsung Electronics Co., Ltd

416, Maetan-3Dong, Yeoungtong-Gu, Suwon-City, Gyeonggi-Do, Korea 443-742

Keywords: Software Project Management, Team Software Process(TSP), Coaching Development Team.

Abstract: Reasonable planning for the project, monitoring

of the project status, and controlling of the project with

appropriate corrective actions are key factors for successful software project management. If the project

leader gets helps and supports from an expert who have the know-how of software project management

instead of depending on only their own discretion, these activities can be conducted much more effectively.

The TSP

(Team Software Process

) is a process framework developed to provide guidelines on software

development and management activities for teams. Samsung Electronics introduced the PSP/TSP

technology to meet visualization and high efficiency needs in software development, in 2003. In this paper,

we propose the software project coaching model based on the TSP/PSP experiences to support the project

leader and the team members efficiently and analyze the applying results and effects.

1 INTRODUCTION

Through experience as a software process and

quality engineer for many years, we’ve learned that

the key factors that make software projects

successful are effective project tracking and

controlling with reasonable project planning. These

in turn are governed by a right judgment of a project

leader on the project status and follow-up measures.

Currently, most project leaders in our organization

manage their projects in the way that they make a

project plan according to organizational standard

process and assign works to team members. And

then they review what have done, what to do next

week, and issues raised at the weekly team meeting.

However, there exist limits in that project planning

and updating are dependent on only the project

leader and so the level of project control is likely to

vary according to skill level and experiences of the

project leader. Therefore, it is needed to develop the

practical guideline for software project management

so that consistent and effective project management

can be applied, considering the development

environment of our organization.

The TSP

(Team Software Process

)

(Humphrey, W., 2005) is a process framework

developed for the purpose of providing the

guidelines on software development and

management activities for teams, while the SEI-

authorized TSP launch coach helps a project leader

and team members to apply the TSP. Within

Samsung Electronics, the PSP

(Personal Software

Process

) (Humphrey, W., 1995) training course

that was a prerequisite for using TSP have been

offered to 375 developers since 2003 and a few TSP

projects have been tried. In this paper, we propose

the software project coaching model based on

PSP/TSP experiences for years. This coaching

model is shooting for customized coaching for the

software development team on the TSP basic

framework.

This paper is organized as follows. In Section 2,

we l

ook into the TSP basic concepts and TSP

experiences in other organization. We explain the

software project coaching model in Section 3 and

analyze the effects by comparing results before and

after applying the coaching model in Section 4.

Finally, we conclude and describe future works in

Section 5.

431

Gwak T. and Jang Y. (2007).

A STUDY ON SOFTWARE PROJECT COACHING MODEL USING TSP IN SAMSUNG.

In Proceedings of the Second International Conference on Software and Data Technologies - SE, pages 431-436

DOI: 10.5220/0001327004310436

 SciTePress

2 RELATED STUDIES: TSP AND

TSP EXPERIENCES IN OTHER

ORGANIZATIONS

The TSP Framework contains Team Building

Process which is represented by the TSP launch

workshop and Team Working Process that addresses

engineering processes and practices used by the

team. All activities on the TSP framework are

conducted under technical support of TSP coach.

The TSP can be a process instance for team

which offers how to do each engineering or

management activity concretely, while the process

areas in project management category of

CMMI(Capability Maturity Model Integrated) (SEI,

2006) and PMBOK(Project Management Body Of

Knowledge) (PMI, 2004) focus on what activities

should be conducted for project management.

Therefore, in many organizations that adopted the

TSP, we can find easily a feature to apply the TSP in

conjunction with CMMI in order to improve their

process. The P-3C organization of NAVAIR(the

Naval Air Systems Command) is a good example

that has accomplished software process

improvement from level 1 to level 4 in just 27

months through using the TSP in conjunction with

the CMMI framework (Wall D. S., McHale J.,

Pomerroy-huff M., 2005). Other case is using the

TSP linking with Six Sigma (Dan S., Duine V.,

2006). The TSP philosophy is consistent with Six

Sigma’s in that it takes a quantitative approach

regarding software development, so there can be

synergy if two methodologies are integrated

properly. Many other valid approaches to adopt the

TSP can exist and, in this study, we focus on a

customized coaching, considering characteristics of

development team and organization.

3 SOFTWARE PROJECT

COACHING MODEL USING

TSP

When new methodology or technique is introduced

in some organization, in order to convey its intended

messages properly, it is helpful to keep the person

who supports and monitors it. The TSP coach has a

role to help a team leader and team members to use

the TSP framework as well as acts as a change agent

for successful introducing of the TSP/PSP

technology within the organization. In the initial

stage, our aim was to delivery the TSP basic

concepts accurately and to verify the value and the

usefulness. However, there existed a limit in

increase of effectiveness for the issues such as

redundancy of existing process and tools and a

burden of additional data gathering. In our view, this

was due to that we didn’t consider enough the

organizational development environment and the

characteristics of the development team, and finding

a solution for these problems becomes the goal of

this study.

The primary focus of the coaching model

proposed in this paper is on two purposes:

customizing TSP process in order that it is merged

with the organizational development environment

and customizing TSP coach’s role to support the

team in different ways according to the team’s

characteristics. Figure 1 illustrates our approach.

Figure 1: Approach for developing coaching model.

Figure 2 is the detailed coaching process of our

model. The thin solid line and the dotted line

represent the coach’s roles and the development

team’s roles respectively, and the thick solid line

means the cooperative activities of the coach and the

team. The activities at the top of figure are

conducted under the leadership of the coach,

whereas the activities enumerated in the bottom part

are works of which subject is the team. We also

marked improved area and added area in our

coaching model with symbols shown in a legend of

Figure 2.

Figure 2: Detailed coaching process.

ICSOFT 2007 - International Conference on Software and Data Technologies

432

3.1 Project Selection

This step is for selecting projects to apply the

coaching model. Basically, the projects that can

meet requirements needed to use the TSP should be

selected, for our coaching model is based on the TSP

framework. The TSP was devised for supporting the

team with 3~15 members and the prerequisites are as

follows;

[The prerequisites for using the TSP]

- Not only should the senior manager understand enough

the TSP central concepts and respect the plan made by

the team, but he or she should not use the team’s data

for any other purpose such as evaluating individuals.

- There should be an agreement regarding data

collection and submission from the team members.

- All team members have taken a PSP for engineer

course.

However, in this study, we couldn’t find the

team in which all team members had taken a PSP for

engineer course. Hence, we decided to offer the PSP

course to the team members who didn’t get the PSP

training just before the TSP launch, after selecting

projects only by the first and second requirements.

3.2 Team Evaluation

The roles and efforts of the coach vary with the

maturity level of the development team. Therefore,

we defined the four maturity levels as described

below and tired to conduct a customized consulting

and supporting according to the team’s level.

[Team Maturity Level]

- Unlearned: The team that doesn’t learn about the basic

concepts and techniques of software measurement and

quantitative project management.

- Learned: The team that has learned about basic

concepts and techniques, but inexperienced in the TSP.

- Skilled: The team that has knowledge and experience

of the TSP, and so gets skill in estimation and analysis

for quantitative project management.

- Self-directed: The team that is managed itself. The

team is skilled in teamwork, works aggressively to meet

its goals and deals with issues came up effectively. The

TSP achieves high performance by building self-

directed teams.

Table 1: Criteria for evaluation team level.

PSP Trained

TSP

Experience

Rarely

(~50%)

Partially

(50%~)

Fully

(100%)

0 year Unlearned Unlearned Learned

1 year Unlearned Learned Skilled

2 year Learned Skilled Self-

directed

The maturity level of the team depends on TSP

experience and the number of PSP trained members

and the criteria for evaluating team’s level are as

shown in Table 1.

3.3 Launch Preparation

The one of primary jobs of the coach that is

conducted in this step is integrating existing process

or tools used in project planning with the TSP

launch process. For example, if there is a checklist

for risk assessment, the coach prepares in advance to

use it at the TSP launch meeting 7 which is for

conducting risk assessment. It is very helpful to

make and use a mapping table, and the coach can

check special items that are different from those of

other teams after including general items in the

mapping table in advance. In addition, the coach

prepares to explain how to do the TSP launch or

what are the responsibilities of the TSP role for

managers in accordance with the status of the team.

3.4 Coaching Effort Estimation

It is important to estimate efforts needed for

coaching to make a reasonable coaching plan, for the

coach involves other works except coaching within

our organization. In this step, an overall coaching

effort is estimated based on three kinds of

information: team’s maturity level from step 3.2, the

number of team members, and the overall project

schedule estimated at the TSP launch. Table 2

illustrates a coefficient for calculating coaching

efforts. For example, if the unlearned team with

eight members estimates the overall project schedule

at six months, we estimate that it will take three

months for the coach to support that team. But this

estimate can be adjusted according to coach’s

experience with 27% VAF(Value Adjustment

Factor). This VAF was derived from analyzing

individual coaching time in the PSP training, for we

decided the VAF would be equivalent between

individual coaching and team coaching. For four

years, we’ve provided PSP training to 375

developers and found that there was a deviation of

about 27% on the instructor’s effort needed.

Table 2: Reference table for estimating coaching effort.

Team Size(persons)

Team Level

Small

(3~5)

Medium

(6~8)

Large

(8~15)

Unlearned 1/3 1/2 1/2

Learned 1/4 1/3 1/2

Skilled 1/5 1/4 1/3

Self-directed 1/5 1/5 1/4

A STUDY ON SOFTWARE PROJECT COACHING MODEL USING TSP IN SAMSUNG

433

3.5 Making a Coaching Plan

We divided activities of the coach into four types;

developer training, supporting the TSP

launch/relaunch, weekly monitoring and consulting,

and interim or final performance analysis. Because

developer training among these is finished before

project starting, the coaching plan focuses on three

other activities. The key difference from original

TSP framework is to have a different coaching plan

according to team’s maturity level. For example, in

case of the self-directed team, the coach makes a

plan to guide the team based on team’s requests,

contrary to the unlearned team that a conscientious

coaching plan is needed. The way of coaching also

differs from team to team in that the on-site

coaching period for the unlearned team is much

longer than that of the skilled team or the self-

directed team.

3.6 Customized Coaching

In this step, customized coaching is conducted

according to the coaching plan made in step 3.5.

Once the team members are familiar with the TSP

process and gather reliable data in two to four

weeks, the main job of the coach moves into helping

the team leader to cope with various problems from

the project in a timely manner. Inevitably, a project

is likely to show deviations from planned in its

actual progress. A significant schedule deviation

may mean that the project may be heading for failure

and this situation demands that the project leader

understand the reasons for the variation and take

corrective actions. For the purpose of supporting the

project leader more effectively, we developed the

guideline checklist for causal analysis and control

actions by the project status. After getting at current

status of the project by actual versus estimated

analysis of schedule, effort, and defect, the project

leader finds the reference number using the project

status indicator table. And then he or she checks

reasons and actions to consider in the guideline

checklist and applies corrective actions, if necessary.

However, if the estimates were too aggressive, the

project leader might revise the estimates. This

mechanism is illustrated in Figure 3.

3.7 Coaching Assessment

This is the step for assessment and improvement of

coaching model. After completion of the project, we

provided evaluation forms to the team members to

get information regarding team satisfaction and ask

for feedback. The questionnaire contains five

categories as drivers of team member satisfaction as

described below.

[The Team Satisfaction Survey]

- Fidelity: The questions to ask how faithful the member

prepares and works in conformity with the process

provided.

- Method/Process: The questions to ask whether the

method and process provided are practical and helpful.

- Materials/Tools: The questions to ask whether the

materials and tools provided are practical and helpful.

- Benefits: The questions to ask about real benefits that

the member feels due to applying new method.

- Other Improvement Proposals: The questions to ask

about improvement proposals for new method and the

coach.

We made three or four questions per each

category and scored one to five points per question.

When we conducted this team satisfaction survey to

the team members in the base projects of this study,

we got high scores and positive comments (see

Figure 4).

4 RESULTS AND EFFECTS

In this section, we analyze the results of the

coaching model proposed. Our analysis points of

view are the benefits of the coaching model and the

team satisfaction. Customized coaching by the team

characteristic, such as team size and maturity level,

is the first and most important benefit. Without the

formalized coaching model, it is difficult to apply

customized coaching, for there are not objective

criteria and strategy for this. Second, the mapping

table was developed so that the TSP process could

be linked with existing organizational process and

tool systematically, in turn avoiding redundant work

due to applying the TSP. Third and fourth, it is

possible to provide consistent coaching through

standardization of coach’s role and easy to improve

a coaching skill through sharing and transfer of

coaching know-how or experiences. Fifth, readiness

for spread of the TSP framework throughout our

organization and sixth, flexibility in introducing new

technology such as cost estimation techniques are

primary benefits that are feasible only when using

the coaching model proposed. And all these benefits

were obtained without an injury to the TSP basic

concepts and effects.

ICSOFT 2007 - International Conference on Software and Data Technologies

434

Figure 3: Reason and follow-up actions by project status.

Table 3: Performance results before and after coaching model.

With TSP

Key Performance

Without

TSP

Before using

Coaching Model

(for initial plan)

After using

Coaching Model

(for initial plan)

Deviation

TSP Results by

SEI (Range)

Schedule Deviation (%) - 3.5 (25) 0.65 (19.15)

5.4 : 1

(1.3 : 1)

6 (-20~+27)

Effort Deviation (%) - 20.97 15.25 1.4 : 1 26(5~65)

System Test Defect

(Defs/KLOC)

4.09 1.86

0.88 2.1 : 1 0.4(0~0.9)

System Test Duration

(Days/KLOC)

3.04 2.67

1.81 1.5 : 1 0.5(0.2~0.8)

System Test Effort

(% of total)

33.8 19.6

6.1 3.2 : 1 4(2~7)

# of projects 1 2 4 - -

Average Size (LOC) 16,866 53,797.5 51,443.3 - -

Figure 4: Team satisfaction evaluation.

A STUDY ON SOFTWARE PROJECT COACHING MODEL USING TSP IN SAMSUNG

435

Table 3 shows the results of project performance

analysis before and after applying our coaching

model. To compare more objectively, we mentioned

both results in company with the TSP results

reported by SEI. As shown in Table 3, we’ve got

better results than that of non-TSP projects, but also

the results as good as SEI reported, irrespective of

applying our coaching model. However, we can find

the fact that the performance after using our

coaching model is much better than that before using

the coaching model. In other words, our coaching

model is beneficial to project performance, too. The

impact of increasing productivity of the teams or

Hawthorne effect on this performance improvement

can be negligible because we don’t focus on

increasing in productivity of team in our

methodology. But, the fact that we can’t statistical

analysis, such as significant level because base

projects are very few is a limit of our paper.

Figure 4 shows the results of the team

satisfaction survey. The questionnaire mentioned in

section 3.7 was used and especially, the answers of

the detailed questions in benefit area were analyzed.

For all four items, namely a reasonable project plan

and schedule, a feeling of belonging and motivation,

practical and helpful guidance, and timely measure,

more than 80% of the team members answered

positively. That means the team feels a high level of

satisfaction.

5 CONCLUSIONS

In the PSP/TSP introducing strategy of SEI, the roles

of agents who lead a change are strongly

emphasized. Therefore, the change agent, such as

PSP/TSP instructor and TSP coach, is core part, not

just role in the PSP/TSP process. After introducing

the TSP technology for quantitative and effective

software project management, we felt need for

developing the coaching model which was designed

and formalized for considering our development

environments and various characteristics of the

development teams. Thus, the goal of our coaching

model was offering customized coaching and we

started with improving TSP coach’s role. We strove

to minimize overload and a redundant work due to

applying the TSP through integrating organizational

process with the TSP process and provide the

optimized and consistent coaching to the team.

Our coaching model has three important

elements; the method for team evaluation, the

reference table for estimating coaching effort, and

the guideline checklist which would be used for

conducting causal analysis and finding corrective

actions by project status. In addition to provide the

customized coaching, this research makes several

contributions. The first is the fact that it is easy to

share and transfer know-how for strengthening of

coaching skill. Second, estimating coaching effort

helps the allotment and control of coach’s workload

so that can accelerate improvement of coaching

capability. Another contribution lies in that the

coaching model is structured to be flexible enough

to add the new technology, process or tools for

software development. Lastly, this study has

additional contribution in that the coaching model

also has a beneficial effect on improving of project

performance.

In order to estimate accurate coaching effort,

more project data and coaching experiences should

be accumulated, for this study is based on the data

collected in only several projects in limited domains.

Also, through integration other methods used in

project management such as software estimation

techniques and software reliability model, the more

improved our coaching model is, the more effective

coaching we offer, leading to successful software

project management.

REFERENCES

PMI, 2004. A Guide to the Project Management Body of

Knowledge (PMBOK® Guide) - Third Edition, Project

Management Institute, Newtown Square, PA.

SEI, (2006). CMMI Model Version 1.2, Retrieved March

2, 2007, from Carnegie Mellon University, Software

Engineering Institute Web site:

http://www.sei.cmu.edu/cmmi/models/models.html

Humphrey, W., 1995. A Discipline for Software

Engineering, Addison Wesley, Reading, MA.

Humphrey, W., 2005. TSP: Leading a Development Team,

Addison Wesley, Boston, MA.

Dan S., Duine V. (2006). Experiences Integrating PSP

and TSP with Six Sigma, Retrieved March 2, 2007,

from Carnegie Mellon University, Software

Engineering Institute Web site:

http://www.sei.cmu.edu/tsp/sym2006-

presentations/integratesix.pdf

Wall D. S., McHale J., Pomeroy-Huff M. (2005). Case

Study: Accelerating Process Improvement by

Integrating the TSP and CMMI, Retrieved March 2,

2007, from Carnegie Mellon University, Software

Engineering Institute Web site:

http://www.sei.cmu.edu/publications/documents/05.re

ports/05sr012.html

ICSOFT 2007 - International Conference on Software and Data Technologies

436