MEASURING I.T. PERFORMANCE
Summary of previous module:
- Last time, we discussed the importance of aligning an IT strategy with the corporate strategy.
- We reviewed methods, structures, processes, and people issues that can facilitate IT alignment.
- We also discussed that accounting for IT as an overhead function, making it difficult for senior managers to assess IT’s contribution.
MEASURING I.T. PERFORMANCE
In this module, we explore measurement systems that help answer the question:
“Is my IT function performing well?”
What are some general guidelines for strategic measurement systems?
- Balanced Scorecard Approach
What are some valid ways to assess IT performance?
What measures can be used to assess the productivity and quality of software development and support--the most expensive and most difficult IT activity to assess?
BALANCED SCORE CARD APPROACH
by Robert Kaplan and David Norton, HBR, 1992, 1996
Problem the articles address:
Exclusive use of Financial Measures such as NPV, Payback, and IRR can lead to
poor decisions focused on the short run.
Example: Financial assessment of total outsourcing:
- Vendors buy IT assets (up to $300 million)--why won’t a bank even give you a loan on IT assets?
- Vendors postpone payments until latter part of the contract
- Vendors may purchase company stock
- All this looks great from a NPV standpoint--what is missing from this picture?
“Best objectives can be badly achieved”
Reduce time to market: Improve management of new product innovations or
Release only new products with minor differences from old products
Reduce production
costs: Better set-up times or
Producing more standardized, low margin products
Faster customer
service on phone: Hang up quicker
Develop an IT system that pulls all relevant customer documents on
line
BALANCED SCORE CARD APPROACH
No single measure can provide a clear performance target.
Balanced Scorecard is a set of measures that gives managers a fast but comprehensive view of the business. Four categories of measures:
Customer perspective: How do we look to out customers?
- Lead time: customer order to delivery times
- Quality: defect rates
- Performance: customer satisfaction surveys
Internal perspective: What must be excel at?
- Productivity: output/input
- Delivery: projected verses actuals
- Employee Turnover
Innovation and learning perspective: Can we continue to improve and create value?
- New Product delivery times
- Adoption of new technologies
- Employee Education
Financial perspective: how do we look to our stockholders?
- NPV, ROI, Return on Capital, Stock prices
The how:
1. Translating the Vision
How do you translate a mission statement like “provide superior service to targeted customers”?
Need to develop scorecard measures for the specific products and services for the specified and articulated target segments.
2. Communicating and Linking
- Communicating & educating
- Setting Goals--Ask SBUs, then employees to set own objectives in accordance with corporate scorecard
- Linking Rewards to Performance Measures--but what can be the unintended consequences of this?
Example: UK Heath Care system rates hospitals on fastest patient care
3. Business Planning
-- need to integrate strategic planning with budgeting & resource allocation.
In Earl paper, we saw the approaches to IT strategy tended to focus on one or the other
- Method approach rated high on strategy vision, low on implementation
- Aministrative approach rated low on strategy vision, too focused on budgets & resources
- Organizational approach rated high on both.
4. Feedback and Learning
- Adapting to changes
- Adapting to learning, like unintended consequences
BENCHMARKING AS A STRATEGY FOR MANAGING CONFLICTING STAKEHOLDER PERCEPTIONS OF INFORMATION SYSTEMS
by Lacity & Hirschheim
How can senior managers assess IT performance?
- They want to be convinced they are getting the best value for their money.
- It is very difficult to develop absolute, objective, measures of IT performance.
- Benchmarking is a way of assessing relative performance.
Benchmarking: Measuring internal IT performance against other companies’ IT performance
FOUR SOURCES OF BENCHMARKS
1. Informal Peer Comparisons
Pro: IT managers trust peers
Con: Other stakeholders don’t perceive information as objective because it’s too easy
to filter out “bad” news.
2. Informal Outsourcing Queries
Pro: Quick and generates a number
Con: Other stakeholders don’t perceive information as objective because it’s too easy
to filter out “bad” news; Vendors don’t put much effort into the informal bid.
3. Formal Outsourcing Evaluations:
Pro: Most rigorous and valid way to assess performance
Con: expensive, disruptive
4. Benchmarking Services
Benchmarking services collect data, normalize it, create reports which plot the client’s performance against a selected reference group of the client database.
Pro:
- Exhaustive assessment against hundreds of companies
- Perceived as objective by other stakeholders
- Less disruptive
Con: Some IT managers select benchmarking services to obtain a good report card
“If you want a bunch of graphs that say you are pretty good you can buy those”
WAYS TO SABOTAGE/BIAS BENCHMARKS
- Select a reference group of marginal performers
- Select technical benchmarks which have no meaning to other stakeholders
- Select benchmarks which focus on wrong IT values: Cost vs. Service
- Undermining the Data Integrity of the Benchmarks
- Designing Reports that Misrepresent the Data
OBTAINING VALID BENCHMARKS
BENCHMARKS FOR VARIOUS I.T. FUNCTIONS
A CLOSER LOOK AT FUNCTION POINTS
Application Software Development and Support is the hardest area to manage--estimates range as high as 80% of all systems fail because they are late, over-budget, and lack functionality.
There is a need for valid measures of software productivity and quality
“If you can’t measure it, you can’t manage it”
We need software measures that:
1. Estimate project completion time
2. Estimate project cost
3. Determine the value of the project to user
4. Calculate analyst & programmer productivity
5. Estimate system quality
6. Independent of technology
7. provide a benchmark against other companies
PRODUCTIVITY & QUALITY MEASURES IN GENERAL
Productivity = Output/Input
Quality: Number of defects per unit of output
In systems, inputs are easier to measure: total man-hours, total costs, etc.
What do we use for output?
Traditionally, we have used lines of code (LOC)
LINES OF CODE
Example: on average it takes 10 man-hours to develop one line of code
What does this mean? How do we interpret this number?
The averages vary with the programming language
It motivates non-efficient coding (i.e. the more the better)
It isn’t calculated until later stages of SDLC, thus a poor indicator of time & budget
Users cannot understand it very well
FUNCTION POINTS
Developed by Albrect of IBM in 1979
Function points estimate software size based on the functions delivered to users in the form of outputs (reports), inputs (data entry screens & forms), queries, files (such as a customer database), and interfaces to other systems.
Function points becomes our measure of output
Productivity will be function points/man-hours
Quality will be number of defects per 1,000 function points.
We will see that function points meets our requirements for sound software metrics:
1. Estimate project completion time
2. Estimate project cost
3. Determine the value of the project to user
4. Calculate analyst & programmer productivity
5. Estimate system quality
6. Independent of technology
7. provide a benchmark against other companies
Additional benefits:
overhead rate associated with counting function points is low:
It takes about 5 minutes to count the function points associated with
what it took you a month to program.