2. 2Experience Predictability in Software Project Delivery
Agenda
• Section 1: Introduction to Estimation Predictability
– The need
– Challenges
• Section 2: Estimation Approach
– Overall approach
– Estimation Framework
– Model Selection
– Continuous Improvement
• Section 3: Case Study
– Scenario, Gaps, TCS Approach, Decision Matrix, Solution, Results
3. 3
If collective estimation accuracy can be increased even by a minimal percentage, it
will translate to savings of multi-billion dollars
Experience Predictability in Software Project Delivery
The Need for Predictable Estimates
Quality
Cost
Schedule
Profit
Budget
Productivity
$3.6 Trillion
66%
50%
Binding Force of Estimation along with the parameters
– The overall Software Spend (Source Gartner)
– IT projects fail in US geography (Source Forrester)
– IT projects Rolled Back (Source Gartner)
4. 4
• Limited reuse of past organizational experience in estimates
Experience Predictability in Software Project Delivery
Common Challenges and Gaps
• Unavailability of standardized rules or guidelines defined for estimation
• Unavailability of varied estimation techniques for different project types
• Absence of defined guidelines to estimate the impact of different
project specific characteristics
• Practice of non repeatable methods even for the same technology or line of
business
• Inability to compare performance with respect to industry standards
• Limited knowledge of estimation techniques and models
• Absence of governance around estimation
5. 5Experience Predictability in Software Project Delivery
Agenda
• Section 1: Introduction to Estimation Predictability
– The need
– Challenges
• Section 2: Estimation Approach
– Overall approach
– Estimation Framework
– Model Selection
– Continuous Improvement
• Section 3: Case Study
– Scenario, Gaps, TCS Approach, Decision Matrix, Solution, Results
6. 6Experience Predictability in Software Project Delivery
Estimation Framework
Sizing Techniques
Utilization
• Apply Framework suggested
models
• Define Metrics for
measurement and bench mark
• Collect feedback and lessons
learnt
Measurement and
continuous feedback driving
Framework improvement
Schedule
Techniques
Cost
Techniques
Effort Techniques
AM Models
(Support
Model, CR
Model etc.)
AD
Models
Assurance
Models
Package
Models (Oracle
Apps, SAP etc.)
Estimation Approach
Standardized Model Selection
7. 7
An estimation framework is a collection of well defined components based
on best practices ensuring consistent outputs
• Experience Predictability in Software Project Delivery
Estimation Framework – Driving Standardization
• Size Estimator: Quantifies “work volume” of a given scope
• Effort Estimator: Derives the person-hours for scope implementation
• Schedule Calculator: Develops project schedule based on estimated effort
• Phase-wise Distributor: Apportions overall efforts and schedule across
phases based on SDLC type
• FTE Calculator: Computes Full Time Equivalents based on effort &
schedule
• Cost Calculator: Derives the overall project cost based on staffing and logistics
• Governance Umbrella: Ensures estimates are reviewed & vetted
• Feedback Adaptor: Captures actuals and lessons learnt to refine framework
8. 8Experience Predictability in Software Project Delivery
• The TCS estimation framework is accessorized by a “Multi Dimensional Decision
Matrix” which enables “FIRST TIME RIGHT” model selection.
Model Selection - Driving Accuracy
• “Decision Matrix” enabler consists of the following four dimensions:
- Estimation Stage
- Technology area and platform
- Project Type
- Software Life Cycle Used
• Based on the model, framework selects organizational baseline productivity
• Based on the decision matrix, the framework performs the following:
- Determines the applicable components of the framework
- Determines the specific methodology/ technique that would be applicable to
each chosen framework component
- Suggests the best fit model based on the organizational history
9. 9Experience Predictability in Software Project Delivery
• Benchmark Productivity
with Industry standards
• Scale effectiveness of
estimation models
• Perform Causal Analysis for
outliers
• Identify levers for
productivity improvement
• Cross-pollination of best
practices
• Refine Estimation models
• Implement Causal analysis
findings
Compute
• Productivity for various
tech-stack/platforms
• Estimation Variance of
different estimation models
• Other related delivery
metrics
Plan process for
• Collection of Actual Data
from closed projects at
regular cycles
• Feedback from Users on
estimation challenges
faced, best practices
involved
Plan Do
CheckAct
Continuous Feedback - Driving Improvement
10. 10Experience Predictability in Software Project Delivery
Agenda
• Section 1: Introduction to Estimation Predictability
– The need
– Challenges
• Section 2: Estimation Approach
– Overall approach
– Estimation Framework
– Model Selection
– Continuous Improvement
• Section 3: Case Study
– Scenario, Gaps, TCS Approach, Decision Matrix, Solution, Results
11. 11Experience Predictability in Software Project Delivery
• Most of the projects incurred regular cost and effort overrun (~150%-200%)
• Increased project management efforts (>40%) due to poor estimates/re-
estimates
• Lack of delivery predictability resulting in scrapping of projects amounting to
millions of dollars of recurring losses
• Huge expenditure due to induction of resources at higher rates at later stages of
the projects to complete them on time
The Scenario
Existing Challenges at a Large US Financial Corporation
• Poor Return On Investments (ROI)
• Dissatisfied clients
• No vendor performance comparison to augment outsourcing
• Difficult decision-making for the right investment opportunities
• No scope of validation of the estimates prepared by project teams
The Consequences…
12. 12
Applied the proven four phased approach
for process improvement
Experience Predictability in Software Project Delivery
TCS Solution Approach
1. Determine
2. Design &
Develop
3. Deploy
4. Deliver
Identify the gaps and
plan accordingly
Tailor, pilot and setup
an Estimation
Framework to
establish processes
and estimation
techniques aligned to
the needs
Integrate solution with
existing organizational
processes
Demonstrate
estimation
effectiveness through
KPIs
14. 14
• Improved predictability of project costs and schedules
• Measured and base-lined productivity levels
• Reduced cost of estimation/re-estimation, idle time, unplanned induction of staff,
project scraps and so on
• Created repository of historical estimation data
• Established estimation traceability to business requirements
• Improved quantitative risk analysis resulting in higher estimation confidence
• Provisioned for fact based inputs aiding vendor bid negotiations
• Measured scope creep at different stages of projects
Experience Predictability in Software Project Delivery
Deploy & Deliver
• Built solution awareness within the practitioner community
• Handheld projects for effective change management
Solution Deployment
Results Delivery
15. 15
Y-o-Y Improvement in productivityImprovement in Scrap Value Reduction
• Reduced cost/function
point (by 41%) for web
based projects
• Reduced cost/function
point (by 15%) for
mainframe projects
Experience Predictability in Software Project Delivery
556
592
541 523
218
142
0
100
200
300
400
500
600
700
Scrapvalue
(millionUSD)
0.041
0.045
0.061
0.065
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
Year 0 Year 1 Year 2 Year 3
CustomerProductivityin
FP/PH
54.70%
62.30%
82.90%
23.40%
36.50%
55.30%
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
70.00%
80.00%
90.00%
Year 1 Year 2 Year 3
20% band 10% band
Model Effectiveness Analysis
2 variance bands (+20% & +10%)
were defined for Model
Effectiveness Analysis
• Year 1: 3 Models were
used, 26% Coverage
• Year 2: 2 New Models were
introduced along with 3
existing, 55% Coverage
• Year 3: Coverage 80%
Stats
Tangible Benefits Realized
16. 16
The Key takeaways
Presentation Title
One of the critical parameters of bringing about certainty in uncertain times is
estimation predictability. This is possible by leveraging the robust, standard yet
flexible estimation framework which enables Project Managers to :
• Harness the estimation experience of executed projects to bring in the
desired predictability.
• Provide feedback for the improvements with further refinements
• Generate key metrics like variance, productivity, schedule & effort slippage
• Get the “best fit” estimation prescription applicable for different types of
projects based on parameter analysis
17. 17
Author profiles
Presentation Title
Pranabendu Bhattacharyya (CFPS,PMP) has more than 20 years of IT
experience and heading the TCS estimation Center of Excellence for last 8
years. He is an M-Tech (IIT KGP) and has been the chief consultant for
many estimation consulting engagements. He is one of the core members
of ITPC (IFPUG) guiding committee and presented paper in various
international colloquiums.
Sanghamitra GhoshBasu has 13 years of experience in software delivery
and project management. She has around 9 years of experience in
software estimation and has been instrumental in defining, developing and
deploying estimation models for multiple engagement types
Parag Saha has over 15 years of industry experience spanning multiple
domains including Transportation, Government, Insurance and Telecom-
RAFM. He is currently part of the Estimation Center of Excellence in TCS
and has been involved in defining and refining estimation models and in
deployment of these standardized models across multiple domains in TCS.