Comparative Analysis of

Software Size, Efforts Estimation Methods

-Anurag Johari
(Engagement Manager - Fujitsu Consulting India Ltd.)

1
 Table of Content
1. Introduction ............................................................................................................... 3
2. Executive Summary................................................................................................... 3
3. Data Collection and Methods Used ............................................................................ 3
3.1 Activities not covered under actual efforts (approved and billable) ........................ 4
3.1.1 Project Management activities......................................................................... 4
3.1.2 Other Activities (Generally project specific) ................................................... 5
3.1.3 Fixed Activities ............................................................................................... 5
4. Phase wise Effort Distribution ................................................................................... 5
4.1 Observations .......................................................................................................... 6
5. Size and Schedule ...................................................................................................... 6
5.1 Observations .......................................................................................................... 8
6. Cost Considerations ................................................................................................... 8
6. Recommendations ....................................................................................................... 9

2
1. Introduction
“The future can be best predicted based on past” is the best suited dictum for custom
application development projects. Generally project managers do struggle in determining the
size, efforts and schedule of the project even before start. Very often the estimates vary
significantly and mostly exceeding the upper side of the estimates.

There are many other challenges, like evolving requirements, delays due to dependencies,
team cohesion and individual’s capabilities etc. further impact directly on productivity,
efforts and schedule and bring the project managers on their toe. They do struggle internally
and externally in explaining the reasons for significant variation in efforts estimated and
actual. The seasoned project managers do use more than one estimation methods and refine
the estimate based on their experience, customer history and the team’s capability.

This document presents an analytical comparison of various methods for software size
determination and efforts estimation and their relationship with the schedule. The sample size
is very small, but the sample data is taken from the projects, which were executed by the
team under the author, who had monitored the same from RFP to delivery.

2. Executive Summary
The customer had engaged the company on long term basis, for rendering software services.
Under the engagement, billing rates are fixed and RFP / requirements are estimated and
client approval precedes the development. The delivery team estimates the efforts and
schedule with multiple methods and submit the final estimate for customer’s approval.
During SDLC the efforts are phase wise distributed as per plan and tracked in client specified
format.

This paper is based on the study and analysis of four successfully delivered custom
development projects. The scope of this report is limited to efforts and schedule and does not
cover quality and team composition etc.

There are interesting observations and important learnings, which can help the project
managers to understand the various aspects of estimation, for better estimation, costing &
pricing, scheduling and other aspects of engagement.

3. Data Collection and Methods Used

For the customer engagement, actual efforts are being tracked under standard phases. For the
engagement, these efforts are being captured in customer’s Timesheet, which also has
corresponding phases for each project (using SDLC).

3
Regarding efforts estimation and scheduling methodology, The customer prefers WBS based
estimation due to its simplicity. However the delivery team uses other methods as well to
validate the estimate.
In this study we used following projects (Oracle and coldfusion technology) for study and
analysis of size, efforts and schedule.

1. Project-1
2. Project-2
3. Project-3
4. Project-3
The reason for sample selection was that these projects were successfully delivered using
SDLC and for these projects the accuracy of efforts data for tasks / phase was closely
monitored throughout SDLC.

The delivered application sizes were estimated using WBS, UCP, FP (IFPUG) and for given
size, following methods were used to estimate the efforts and schedule.

1. WBS - Total Estimated Efforts = ∑ Task wise efforts

2. UCP - Estimated Efforts = Size in UCP * Productivity ( Hrs / UCP)

3. FP Linear Relationship –Estimated Efforts = Size in FP * Productivity (Hrs / FP)

4. CoCoMo– Estimated Efforts = k*(Size in FP) ˄ b

5. Schedule Computing – We used a parametric equation derived (using regression

analysis) from the much larger sample data of Cognizant Technology Solutions
(CTS), and publicly available on internet.

3.1 Activities not covered under actual efforts (approved and

billable)
For reporting purpose, following activities are not covered under the head of
actual efforts.

3.1.1 Project Management activities

Following activities were covered under separate head namely “Onsite-Offshore
Coordination”, and could not be tracked on specific project basis.

1.1. Project & Resource planning

1.2. Communication (internal and External)
1.3. Inter Group Coordination
1.4. Team Management
1.5.Client Reporting

4
 3.1.2 Other Activities (Generally project specific)
We also performed following activities, which are neither captured in customer
timesheets, nor quantified at project level.

1.1. Team’s communication with onsite

1.2. Configuration Management
1.3. Team Meetings
1.4. Idle Time (in the need of receivable or client approval)
1.5. Presale activities (Not project specific)
1.6. CR Management ( Review, Impact Analysis, feasibility and R&D, estimation)

3.1.3 Fixed Activities

1.1. Server Management

1.2. Timesheet Management
1.3. Issue Tracker management
1.4. WSR
1.5.SQA /IQA/PSM
1.6.PMP, QMP,CMP Management (applicable to WebMAID II, III and CC-III)
1.7.PPAT Management
1.8. Process lite Management
1.9. Causal Analysis Meeting

4. Phase wise Effort Distribution

Following is the phase wise efforts distribution (Table-1) for given project. However it does
not include efforts as mentioned under section 2.1.The table contains actual efforts spent
across the SDLC phases and support. Here we have not included the % distribution for each
project as that is not relevant to the context.

Project-3 Project-4 ** Project-2 Consolidated

Phases * Efforts
% Distribution
(1160 Hrs.)# (1016Hrs.) # (1401Hrs.) #
Requirements 99 103 90 292 8%
Planning 115 20 93 228 6%
Environment 22 16 38 1%
UI
Prototyping 190 90 110 390 11%
Design 161 210 205 576 16%
Coding 356.5 414 607 1377.5 39%
Quality 128.5 108 250 486.5 14%
Release 28 49 23 100 3%
Support 10 12 40 62 2%
Mgmt
Total 1110 1006 1434 3550

5
 Table – 1

* Though task wise efforts were accurate, but there were discrepancy in phase wise effort
distribution for Project 1, thus not included
in this sheet.
** RBM includes nominal (14 Hrs) efforts under Project Management
# Approved Efforts

Given below is the graphical presentation of phase wise effort distribution.

Phasewise Effort distribution

3% 2% 8%
14% 6%
1% Requirements
Planning
11% Environment
UI Prototyping
Design
Coding
Quality
16%
Release
39%
Support

Fig -1

4.1 Observations
 Development efforts (Coding + Testing) are 53 %, which is slightly higher than
industry wide accepted norm of 50%.

 UI prototyping and Design efforts are 27 % against the industry average is

between 20-30% depending on project. However in present case, considering the
design complexity and, templates in use the efforts are on higher side.

5. Size and Schedule

The size, efforts and schedule data (actual and estimated) is given below. The applications
delivered, were sized as per WBS, UCP, FP and following estimation methods were applied
to compute the planned efforts and schedule.

6
 1. WBS Estimation - ∑ ( Tasks efforts)
2. UCP estimation – (Efforts = UCP* Productivity)
3. Linear Effort Estimation – (Efforts = Size in FP * Productivity)
4. CoCoMo Efforts estimation – As per CoCoMo guidelines
5. CoCoMo Schedule – (Schedule in months = 2.5 (CoCoMo Efforts in Adjusted
Person Month) ^ 0.35).

Project Size and Schedule (Computed and Actual)

Project WBS Size UCP Size(FP) Actual Linear CoCoMo Actual CoCoMo Actual
Efforts (UCP) Efforts Efforts FP Efforts Schedule Schedule Efforts
Effort *** as % of
CoCoMo
Efforts

Project-1 510 30 840 101 570 606 968 3.4 4.7 59%
Project-3 1050 44 1232 172 1110 1032 1787 5.3 5.8 62%

Project-4
* 990 53 1484 187 1006 1122 1686 3.6 5.6 60%

Project-2
** 1450 67 1876 218 1407 1308 2347 3.9 6.4 60%
Table – 2
Note: All efforts are in person- hrs. and schedules are in months

* We did apply the best practices for requirement management.

** One module was based on .Net and Windows CE. But considering its small size its
is considered at par.

*** Schedules are computed using the equation derived by CTS using its own data.

Following is the Size vs. efforts graph.

Fig-2

7
 5.1 Observations

5.1 Most striking feature is consistent relationship between estimated

COCOMO Efforts and Actual Efforts. Actual Efforts are almost
60% of COCOMO Efforts.

5.2 Linear Effort estimates are very close to actual efforts.

5.3 There is considerable variation between COCOMO schedule and Actual

schedule.

5.4 Actual Efforts are nearly same as WBS based.

6. Cost Considerations
Activities under Section 2.1.2 and 2.1.3 are non billable as client does not feel getting any
tangible output of these activities. But these activities are substantial portion of Cost
Component.
Fixed activities need to be performed irrespective to number of ongoing tasks / project for the
engagement.
Following are the fixed activities along with respective efforts (approx.) required to perform.

Fixed Activities (Shared across all projects)

Activity Description Monthly Efforts
in P. Hrs.
(Approx.)
Server Management 12
Timesheet Management PN Form, adding resource, creating tasks, 48
assigning task, extending dates, generating
reports, entering timesheet data (for The
customer timesheet), deactivating the tasks,
timesheet approval, Issue resolution for
invoicing
Issue Tracker management 2 Hrs / week 8
WSR 3 Hrs (2 Hrs for internal and 1 hr for client WSR) 12
/ Week
SQA /IQA, Preparation, Audit, follow up and issues closure 8

PSM 6
Project Matrix Updates 2

Causal Analysis Meeting 2 Meetings (Internal & with CAT) per month with 6
3 team members participating. Each meeting for
one hour.
Total 100
Table-3

8
This scenario is important especially when most of the QMS activities are brought under
single umbrella for the engagement.

Other activities as mentioned under section 3.1.2 are generally project specific and
proportional to volume of work.

Following graph (Fig-3) presents the Non billable Efforts (cost component) and its
relationship with volume of work.

Activities as per section 3.1.2

Efforts

Fixed Activities

Volume of Work

Fig – 3

6. Recommendations
1. In general actual efforts are 60% of CoCoMo efforts. But these actual efforts do not
include certain activities as mentioned under section 3.1.2 and section 3.1.3. Thus in
totality CoCoMo is a good measure of over all cost of the project.

2. In the absence of past data, estimation and effective planning of activities under
section 3.1.2 is difficult. Thus we need to track these activities in consistent and
structured manner.

3. Activities under sections 2.1.2 and 2.1.3 do fit well in rest of 40% part
(CoCoMo efforts – actual efforts), but client may not accept CoCoMo based

9
 estimation. Thus we can use productivity factor (60%) for WBS bases estimation.
As on now we are just including 10% contingency buffer, which is altogether a
different aspect. Alternatively a good pricing can take care of hidden cost.

4. CoCoMo effort based schedule calculation (using parametric equation) is inconsistent

with actual schedule data. Thus we need to derive our own equation (based on The
company data).We will need 30 projects data, to normalize and derive the equation.

5. There are considerable efforts spent on fixed activities as per section 3.1.3. We may
need to work out the strategy either to minimize the same in lean phase or some kind
of minimum billability for these activities can be negotiated with the client.

-----Concluded-----

Note:- For any discussion related to estimation methods or this study, the author can be
contacted at [email protected]

10