Home Services FAQs Online Proposal Famous Buttons Success Stories Available Presentations Contact Us
  
    
If you cannot find your answer, please contact:
Bill Hufschmidt
(262) 789-9190
Bill.Hufschmidt@FunctionPoints.com


Here are some start-up questions we typically hear:
How do I get started?
How do I convince my management?
How can I guarantee success?
What do I need to prepare for a Function Point Count?

Here are some more advanced questions (not for casual browsers):

How can I baseline 700 Systems?
How do I maintain my Function Point Data?
What are the latest developments in measurements?
So, what's in it for me?
What's your TOP INDEX?
Will there be Knowledge Transfer?
What is Leveraging?
What is a Measurement GRO (Get Requirements Once)?
Is there a simplified RFI Process for Measurement Start-Up?
How do you handle Algorithms?
What are CEO Bragging Rights?
What is Business Management Advocacy?
What are some of the issues of Chargeback?
How many points can a Function Point counter count?
Crawling, Walking, Running?
What about Estimating?
What are Internally Generated Transactions? Do they get counted?
What are some of the issues of Outsourcing and Service Level Agreements?
What is a Productivity Risk Analysis?
What is the 10 Dimensional Metrics Model?
What is a Software Balanced Scorecard?
Does Time to Market apply to Software?

We hope these FAQs help you with your planning. Please contact Bill Hufschmidt when you are ready for a success.

How do I get started?

You have already made the most important move. All of the successful organizations have utilized outside resources to get started. This is our mission and we do it well! The free proposal is specifically designed to address the most commonly asked questions, before they get asked.

How do I convince my management?

As the Products and Services show, the most successful way to convince management is with the Management Overview. Part of this is the challenge to management to commit resources. If they will not commit resources, it is not in your own best interest to pursue measurements at this time. After all, if you do a world class job on something that no one cares about, how will it help your career?

How can I guarantee success?

DSC offers a Measurement GRO (Get Requirements Once)! This gathers managers and principal users together to introduce what measures are; what measures are available; how easy it is to get them and how to prove their value. The GRO session then gets the group to define their requirements and set priorities. In short it allows you to manage their expectations.

What do I need to prepare for a Function Point Count?

Using our Fastpath approach, preparation for Function Point Counting usually requires less than one hour. There are two activities that can speed the counting process:

Assembling Application Information about Files, Reports and Screens.
Reviewing the 14 General System Characteristics
File List
The file list includes all permanent files created, maintained or accessed by the application. These are databases, permanent non database files (sequential files, index files...), tables, objects,.... Temporary, or work files are not included.
Report and Document Lists
The Reports List includes any reports and/or transactions sent outside the boundaries. The Document List includes all documents received and/or processed from outside the boundaries. Either of these could also include batch transactions sent or received.
User Interaction List
The User Interaction List includes screens, windows and dialogs, ... used in the application. Samples are helpful if they are readily available. An alternative to samples is to access the live application (test or production) to review the user functionally during the counting session.
14 General System Charteristics
Review the 14 GSC's in relation to the application to be counted.

How can I baseline 700 Systems?

We were recently asked how to baseline 700 systems if there wasn't enough budget to count them all. The fastest way to Baseline is to do it once! The best way is to follow all of the published IFPUG CPM 4.2 rules using our exclusive FASTPATH method. This speeds the process, increases consistency for comparability and most important insures staff and management support of results.

However, we realize there may be a point of diminishing return, where clients think of alternative tools and/or techniques. Let's review the most commonly considered alternatives.

Backfiring with tools. It has been my dream, since I first learned of function points, to automate the counting process so we could move right into using the data for business decisions. The original work on automating counting took 31 basic mainframe COBOL projects, vintage over 15 years ago, counted them, divided by the Lines of Code (KLOC) and found an average of 105 COBOL LOC per function point. Of course, the days of a single architecture are gone and no one since has ever seen or reviewed the original data. Yet this infamous number, that needs to be adjusted for subsequent rule changes, has been used to justify, calculate and prove every imaginable claim and assumption, including the secret of life.

Over the years more and more tools have tried to automate the process. I have even had tool vendors attend our public classes. Some like IEF should be able to do it, but TI never got it fully working and has never updated it for changes in the model. One client told me they found Viasoft's tool to be 300%-400% too high. Our own comparisons found a range of 40% to 250+% of the original number (GE, PacBell, others). On the upper end, clients did not care to pursue points beyond the range. Nevertheless, the concept has value, if it is validated for your systems. See recommendations below.

The problem with tools we have seen is the counting standard itself. The rules call for counting logical files and logical functions, not physical. You do not count temporary or work files. You do not count backup files for restart or recovery. Tools have trouble distinguishing these. What about advanced issues, like algorithms or alternative media? What about multiple logical functions on a single screen or multiple screens to accomplish a single function? When is a purely navigational menu an inquiry? How do you count mapping systems, Lotus Notes or other advanced technologies (BellSouth, AT&T, Ameritech, others)? Although the impact of these may be insignificant on the final count, they can undermine staff and Management "buy in." We once saw a single physical screen that had over 2500 fp (the threshold for a major system).

The primary alternatives are Functional Estimates, File Estimates and Function Point Averaging. Functional Estimates count the number of files and multiply by 31 fp. The assumptions are that everything is average and every file has add, change, delete and query functions associated with them. File Estimates count the number of files and assumes they represent 36% of the total functionality. Function Point Averaging also assumes everything is average but still requires the time to collect all the data.

There are three problems of the alternatives. First, they tend to break on major systems. Second, the assumption of everything being average is false. Third, Management and staff don't believe the numbers and their perceptions are right 80% of the time.

From our experience, 0-1000 fp is a small system; 1000-2500 fp is a medium size system and 2500+ fp is a major system. Major systems have major business impact because of major amounts of data that require major amounts of processing. They are often delivered in on-line, batch and other modes at multiple sites where volume considerations generate additional requirements. The assumptions of everything being average just don't stand up. How can a 3270 screen with 93 DET's (TIRKS) be average? How can a GUI referencing 17 files be average? If the Management and staff don't believe it, using any alternative is wasting time and money.

Recommendation:

  1. Assume a Management 80/20 approach.
  2. Assume limited resources.
  3. Assume seven Directors each having responsibility for 100 systems.
  4. Assume credibility is required or the entire measurement program will be scrapped including the measurers.
  5. Get an inventory of systems by Director.
  6. Interview each Director individually. Determine their priorities and needs for measures.
  7. Review their inventory with them. Have them pick their "Top 20" systems. These will include all of their major systems plus a few of their personal favorites. Then pick at least 2-3 additional systems for each "typical" platform. Be sure to include any systems or projects that are currently "hot items" for them (high visibility or risk, questionable estimates, new technologies,….).
  8. Count their Top 20! This will insure Management and staff support for their data. It will also validate assumptions that can apply to the remaining 80% of their systems. Choose an alternative method. We believe KLOC without a tool will suffice.
  9. Contact the "support" person, who prepared the inventory, for a KLOC list of all of the Director's systems.
  10. Validate the KLOC/FP for the systems counted. Be sure to consider H/S/P/R (Hardware, Software, People, Requirements) factors of every system, project and platform.
  11. Establish a KLOC/FP ratio for each Director. Be able to explain the differences. Anticipate skepticism and fear. Remember Function Points are still new and Directors abhor Risk.
  12. Extrapolate "averages" for the remaining 80%.
  13. Repeat steps 5 - 12 six more times, once for each director!(see recommendation 3.)
  14. Cross check the results.
  15. Report the results. Report to Directors, individually, first.
  16. Report all of the results together. Be aware of people who say the information should be kept confidential. It is unprofessional to NOT publish but it is prudent to be ready to explain differences.
  17. Budget resources for follow-up. Some Directors will ask for additional data. Some will question results and may require rechecking or auditing the counts. Remember there is a built-in American mentality that "more is better" so there are potentially six unhappy Directors. Also remember that function points is only one dimensional. The Directors will consider the review meeting a big win if you can address their concerns. However, they will also want to know more information about other dimensions in the Ten Dimensional Metrics Model. This is information they can use in business decisions, but you will need resources to collect and analyze it.
  18. Item 17 is critical. We have seen many, many programs and players stall and fail when the "goal line" was considered collecting function points. Function Points is just one means to the end
  19. Be prepared to offer appropriate training to staff, middle managers and senior managers. One client is working towards CMM Level 2. They have had two areas independently assessed. Both assessments identified their Measurement Program at the beginnings of CMM Level 4. They have trained approximately 20% of their staff and every manager has at least seen a Function Point Management Overview.
  20. Be prepared to integrate function points and measures into changes in your Process. You are about to open a door and see a whole new panorama with all of its opportunities and pitfalls.
  21. Use common sense, which we feel is our strong suit.

That is our "top 20" steps for baselining 700 systems.

How do I maintain my Function Point Data?

Maintaining Function Point Data (Counts)

"Work FP vs. Base FP"

Function Points are always counted following the same published rules. When maintaining or updating function point data, there is a difference between Work vs. Base. Work is everything we do encompassed in the FP model. Base is the net total functionality available to "users" at any point in time.

In a brand new development 1000 fp of Work should deliver 1000 fp of Base functionality to the user. However, some development yields no on-going functionality, e.g. a conversion. If the above project requires 100 fp of conversion programs, the estimate of Work is 1100 fp, but the Base remains 1000 fp.

In an enhancement project, Work vs. Base can fluctuate between ratios of:
          1 fp of Work yields 1 fp of Base
   to    1 fp of Work yields 0 fp of Base.

For example, if a new "average" rated report is requested, the result is 5:5, 5 fp of Work yielding 5 new fp of Base functionality. If instead, a new field is added to an existing average report and the report remains average, the result is 5:0. That is, 5 fp of Work with no increase in the total Base functionality.

User Work

These are activities that the user is responsible for, but that we perform for business reasons. For example, balancing is an on-going user responsibility (requirement). If this is performed by people in DP Operations on third shift, it is still User Work. This is very important in staffing justification. Another example is security. Creating security functionality is an Information Systems' (IS) responsibility. Adding 2000 users to the security table is a user responsibility, even if it is performed by someone in IS.

Time spent adding rows to tables, like changing rates in rate tables, does not add user functionality and therefore is not counted as Work FP's. The time is User Work and it currently impacts your overall productivity and quality numbers in a negative fashion.

We recommend staffing and recording this effort separately so it can be understood and explained to management. Functionality created to reduce this effort, e.g. a new screen to update a rate table instead of recompiling and retesting multiple programs is counted.

Suppose 25 tables (ILF's) with 25 maintenance screens are created to reduce future NPA splits (splitting area codes in Telecommunications) by 2000 hours/split. Total functionality increases by 360 fp. (Assume 75 EI's (A,C,D) and 25 EQ's, all LOW and the VAF=1.20). Bottom line productivity (Function Points Supported per Person) is doubly improved by increased user functionality and reduced User Work.

Supplemental Metrics

Supplemental Metrics are additional metrics that help "tell the story" of issues outside the function point model. These might be the number of users, number of accounts, sales impact or other objective number that helps explain what we are doing and why. For instance, an NPA split may require 5000 hours of Work and result in no increase in Base FP's for users. A supplemental metric might be that as a result of the split, our rule set data base now has 10,000 rules versus the previous 8,000, a 25% increase.

One system processes 6 Billion records per month. What was it 2 years ago? Has the arrival of cellular phones increased the number of calls people make; or has voice mail reduced it? Both of these business features have probably caused more work than the increase in FP might show. Their impact can be shown with supplemental metrics.

Supplemental metrics recognize that the Function Point model is not a "silver bullet." Like any other tool, if management only uses one number, we guarantee they will shoot themselves in the foot. It becomes the Metrics Coordinator's job to assist all levels of managers with understanding and conveying their full story.

Counting Enhancements and Releases

To count a release, you need to analyze the files, screens and reports. This information is usually contained in a Design document, Time & Cost document and/or Bulletins.

Design documents: Check the input section for new on-line and/or batch transactions. These will update one of the files (ILF's) identified in either the original count or a subsequent release. Some transactions just start a process, like an input parameter. Be careful of the word "input." In the FP model this may be an External Interface File.

Design document: Check the processing section for new or changing files. If a new file (ILF) is identified a logical question is how is it maintained. If it is maintained outside the system, it will be an EIF. If it is maintained inside our boundaries there must be on-line or batch input (EI transactions). If these were not identified in the above input section, contact the subject matter expert (SME) to find out why.

Updating user data is a user function even if it performed by someone outside of the "user" area, e.g. system administrator or user "coordinator" in Information Technology. Be sure to count these user functions.

Design document: Check the output section for reports. Don't forget EO's on multiple media get counted once per media actually used. Therefore a report viewed on-line, on paper or on fiche is three EO's. Note: A "file" in data processing jargon may be any set of data on tape or disk. If the data is sent outside our boundaries it is an External Output to the FP model..

Design document: Check the miscellaneous section for any screens, files or reports not previously identified.

Time & Cost document: This may also identify screens, files or reports to count. Read the summary carefully, as a free format may hide user functionality, either Work or Base, that should be counted.

Bulletins: Check each bulletin associated with a Change Request (CR). This may also be free format so watch for all function point components (EI's, EO's, EQ's, ILF's and EIF's). Functionality in Bulletins associated with Maintenance Requests (MR's) should have already been counted.

Situations and scenarios

Maintenance functionality, (generally but not exclusively MR's) has already been claimed and counted. Therefore do not count either Work or Base function points. New functionality, created under an MR can be counted, but this may be a quality issue.

Scrubs and Scans, for new user requirements, count as Work fp.

When analyzing Work, be aware that staff members may not use FP jargon when documenting MR's and CR's. Contact the SME if necessary.

Recording

Using the FP Tools, record each of the function point components and identify whether they were ADDed (New), CHANGEd or DELETEd. Use the appropriate input or output matrix to determine Low, Average or High ratings. For Change or Delete, if the rating is not obvious, check the original count for its rating. Original ratings were derived with the SME's. This provides additional cross checking, while improving consistency and quality.

What are the latest developments in measurements?

Latest Developments in Measurements

There are new developments in Measurement that you should be aware of. First, we continue to collect data very quickly. We just completed collecting function point data for a major utility, representing 1000 developers, in two months.

Second, our Productivity Risk Assessment is being used regularly as a mandatory requirement for all major projects. The process begins with a function point count and check on estimated Productivity. It then reviews all of the hardware, software, people and requirement Risk factors of the project. It identifies all of the issues that are helping and/or hindering the project team and reports back to management in just two days. Management then reacts to eliminate any hindering factors.

Although this sounds like normal project management, Management gets an assessment from an entirely different perspective. We have now performed several of these and Management has altered the direction and scope of every project. The difference and added value of the process is the documented quantification of expectations.

Third, the quantification of expectations allows us to measure corporate goals and map results to prove your value and effectiveness.

Fourth, we've found a way to save millions on the Request For Proposal process by using function points. It's very simple but does require some management training.

I hope these help you in planning.

So, what's in it for me?

Career Paths in Metrics

Is there a career path in Metrics? Yes! However it is under construction. For the latest information, contact Bill Hufschmidt (262) 789-9190.

What's your TOP INDEX?

TOP (Team Overall Performance) Index

Is your team's performance based upon one or many factors? If you succeed at one while others slide will you be considered successful? Does the importance of each factor change with time? Which factors are most important and by how much? Would you like to be able to prove your success and get credit for what you accomplish? Is your team judged by facts or opinions?

If these questions can affect you, you need a TOP Index contract for performance.

A TOP index is a way of summarizing varied and changing performance criteria that are influenced by varied and changing priorities unique to your environment. A TOP index accounts for issues such as:

Productivity, Usage, Volume or Penetration,

Estimating, Customer Satisfaction,

Quality, Mission Criticality,

Cost, (Doing Things Right).

Staffing,

SEI CMM, ISO, Baldrige, (Doing The Right Things).

In the world of measurements, the concept of a dashboard has become popular. But dashboards don't affect the business. Gauges don't help me if my flight is 4 hours late or is diverted to the wrong city. You can be ISO certified making concrete life preservers.

Until Measurements help with business issues, they won't be taken seriously!

Development Support Center's 10 Dimensional Metrics Model provides an integrated measurement framework for managing your software environment and proving business value. It is a documented and quantified contract to guarantee and prove success.

What will your TOP index be and when will you install it?

Will there be Knowledge Transfer?

Knowledge Transfer Process

Our process starts by establishing an overall Metrics Coordinator. This person will be dedicated nearly full-time during implementation, but he/she will not need to be an addition to staff.

The Metrics Coordinator will assist with identifying who and where key personnel are; how they fit in the internal organization; particular management strategies and/or tactical issues; and other important concerns that outsiders might waste time learning. He/She will transfer that knowledge to us while we transfer metric knowledge to them. When we are done, this individual will attain Expertise Level 7 or 8. At one client, four individuals achieved Level 8. This accounts for why they are one of the best measurement programs in the country and why their measurement program was independently assessed, twice, at the beginning of CMM level 4.

The Metrics Coordinator should have more management focus than technical focus, but the latter is helpful in gaining staff and middle management respect and support. Project management experience is a plus. During the start-up phase, he/she should be relieved of as many other responsibilities as possible to ensure maximum knowledge transfer.

The Metrics Coordinator assists with:

After Function Point implementation, our experience is that this person will be too valuable to return to their old responsibilities. Their new role becomes a permanent position. They will be swamped with management requests that today are impossible to fulfill, or even conceive. For example, one Metrics Coordinator saved their company 1.4 million dollars in a single afternoon. Another evaluated a 70 work year project in just two days. The stories go on and on.

No other employees will need to be dedicated to metrics, although several may get normal management requests for information to solve business needs. As stated above, on-going measurement becomes part of normal project management.

What is Leveraging?

Leveraging: Proving Your Business Value

Bragging Rights are the stories and anecdotes, of common sense experiences, that acknowledge and validate goals and objectives as well as the results of team efforts to accomplish them. Bragging rights are a natural source of pride for direct team members and any associated individuals. CEO Bragging Rights are the ultimate corporate Bragging Rights.

Do you have results that can be considered CEO Bragging Rights? Most Information Technology professionals overlook, or do not know how to report the bragging rights they provide. Leveraging is a natural way of explaining their contribution.

Leveraging combines a system's basic functionality with its usage to explain its business impact, growth, penetration and/or change. Leveraging can prove you are working on the right things.

Let's review two systems.

Two systems, measured with the consistent, objective and auditable Function Point model, are both 1000 Function Points. System A is used by 5 people and system B is used by 100 people. It is a true statement that system B is leveraged 20 times (20x) more than A.

Leverage is a way of measuring a particular system's impact, not importance! The usage factor (in this case, users) will not always apply to every system. Whatever the usage factor, it must make sense from a business perspective and be an independently verifiable number (e.g. number of stores for a shoe company, number of branches for a bank or number of total transactions).

The usage factor may also represent "penetration" of a system. For example, a new agent system will be rolled out to 5000 agents over 18 months. After one month, 250 agents are using the system. After one year, 3500 agents are using the system but business has expanded and there are now 5500 agents. What if the system was enhanced by 20% during the same period? How can I measure the full impact?

Initial impact (after one month):
1000 FP x 250 agents (usage factor) = 250,000 Leveraged Function Points
Penetration: 250 agents/5000 agents = 5%

First Year Projected Impact: 1000 FP x 3500 agents = 3,500,000 LFP
Full Projected Impact: 1000 FP x 5000 agents = 5,000,000 LFP

Actual In Progress Impact (after one year):
Enhanced System Functionality: 1000 FP to 1200 FP = 20%
Penetration: 3500 agents/5500 agents = 63.6%
Note: Business has grown. Total agents is now 5500, up 10%

Full Projected Impact: 1200 FP x 5500 agents = 6,600,000 LFP
This is a 32% greater impact than original expectations.

Actual Final Impact (at 18 months after full implementation):
1300 FP (includes 100 extra Function Points during final 6 months)
x 5600 agents (100% Penetration but now 5600 agents)
7,280,000 LFP

The change in impact = (actual final impact)/(original projected impact)
= 7,280,000 LFP/5,000,000 LFP
= 1.456

System Impact's Bottom Line = 46% greater impact than ever expected!

The above impact may not be readily apparent to management but does reflect increased contribution of the system. The impact may not always be apparent on the corporate bottom line. For example, sales may only be the same as last year. But, is that the fault of the system? Can IS professionals point to a good job? We think so. Look at the SABRE airline reservation system. It is a CEO Bragging Right even during a sea of red ink.

Sometimes a usage factor may change. A bank service bureau serviced 300 banks with $6B in assets. The service bureau usage factor was the number of banks. Each year the service bureau grew by approximately 30 banks or roughly 10%. One year the service bureau added only 1 extra bank. But that bank was as big as the other 300 banks combined - $6B in assets. To understand and explain the business impact, the service bureau added assets as a usage or leverage factor.

The selection of the usage factor is important. If management doesn't "buy it," it won't be very useful. And of course, the usage for one system may not make sense for another. Be careful not to mix apples and oranges.

We cannot always prove immediate direct total value. However, we can prove the numbers and they may provide a fresh, easily overlooked perspective. After all, there are only a few reasons for doing business projects: Mandatory requirements (legal, regulatory, corporate, political) Increased sales or market share, Increased efficiency or reduced costs, Increased customer satisfaction, Investment for future improvements in the above.

If the measures do not eventually reflect these, then we are simply gathering data and wasting time.

What is a Measurement GRO (Get Requirements Once)?

Measurement GRO (Get Requirements Once)

The fastest way to define success in measurements or any other business problem is to do it once!

This is possible using Development Support Centers' Get Requirements Once, GRO, Assessment Workshop. A Measurement GRO can guarantee success in you Metrics program by getting management to define their needs and requirements.

This on site workshop brings together your management team to define their objectives and assure their support. It allows you to focus on the critical measures needed within your organization right now. (Remember, management's focus is a moving target and you will need to do this again in the future. However, the known possibilities are defined in the 10 Dimensional Metrics Model.)

GRO is a highly structured requirements definition and detailed design process. The process brings management, systems personnel and users together in interactive "brainstorming" under the leadership of a trained GRO facilitator.

Some of the benefits you receive are:

  1. Reduction of up to 50% in the time required to define management and/or user expectations.
  2. Reduction in design costs and time of 20-50%.
  3. Improved management and user satisfaction, involvement and commitment.
  4. Improved communication among diverse and dispersed units.
  5. Rapid identification and resolution of problems and issues.
  6. Measurable improvements in quality that reduces maintenance and enhancement costs.
  7. Reduced misunderstandings and related conflicts.

Avoid false starts and wasted time on unnecessary or unrealistic expectations. Avoid failure and arouse management's desire to succeed by helping you succeed.

Is there a simplified RFI Process for Measurement Start-Up?

Measurement Start Up-A Simplified RFI Process

The most critical success factor in measurement start up is choosing the right partner. In your RFI process begin with a half page description of your shop: size, environment, reasons for starting a measurement program. Then ask the key questions:

What is the experience and expertise that distinguishes their company?
What are the first steps and deliverables in starting function point measurements?
How long is a management overview and what will we get?
Do you have case study training? How long is it? Can I use my own case study?
Do you have follow up or refresher classes for enhancements?
What about estimating? How early in the life cycle do you apply FPA?
What about knowledge transfer? What quantified level of expertise is achieved?
What is the anticipated counting rate of graduates?
What is your guaranteed counting rate? What is your guarantee?
What are your fees, terms and conditions?
What is your recommended approach: how long and how much?
What are the anticipated results?

The above questions will give you enough insight to choose a partner. Vendors should be instructed to limit responses to a half page or less per question. Supplemental materials can be welcomed. However, vendors should be told that these may not be included in your decision. After all, measurement needs to be low overhead and your intent is to pick a partner. It is not to wade through ten pounds of books and sales material for a term paper.

We recommend a pilot with multiple optional engagements. This maximizes your flexibility and minimizes your risk. One follow up engagement will consult on reporting and using the data. This supplies information to prove the impact of your other initiatives, like project management. It can also prove the effectiveness of strategies like purchasing packages. Other engagements can be used if you decide to outsource measurements, outsource everything, or move into the quality, cost, leverage, value and other arenas.

A note about software tools: Vendors are often overly enamored with their software. As small business owners, many have poured heart, soul, time, emotion and possibly their own money into it. This tends to over inflate its importance in their minds. Sometimes software becomes an ego extension, like a car. It is not unusual that they emphasize software instead of educating you.

Yes we also have software. But as you noticed above, we did not ask questions about software. Do you hire teachers because they drive a Ford, Chevy or Toyota? There will be plenty of time to make educated software decisions after you understand measurement capabilities and refine your goals and objectives.

How do you handle Algorithms?

Who has them?
Every company has algorithms or embedded systems. We have seen them in Airlines, Banks, Insurance, Telephone, Manufacturing, and a dozen other industries.

What are they?
Remember the tables in the back of your old Algebra and Geometry books? Do you remember how those tables were calculated and what the variables meant? Or do you remember you had to look to this column and then to this row to find the answer?

We view algorithms like these tables. If they had been prepared by the system and stored, they would be ILF's. Instead the calculations are done in real time. We view them as variable(s) coming in; optimizing of the variable(s); and variable(s) going out. Our definition is "optimizing multiple variables to obtain a predictable result."

The Problem is Effort.
The problem with algorithms is the unquestionable extra effort required to program them versus any other component of the Function Point Model. Caper Jones initially looked at these and gave them a 1-10 rating. On the low end were simple calculations like date conversions and fp calculations. On the high end were true algorithms like calculating monthly fees: for life insurance (age, sex, product, smoker, diabetic, actuarial tables…) or telephone billing plans (features, length of calls, time of calls, day of calls, distance of calls, volume, 800 service…).

Embedded or real time systems add in constantly changing factors that feedback data. For example, a brewing process that is also dependent upon changing temperature and/or density and/or presence of certain elements…, or a navigation process that includes speed, direction, altitude, wind, location, enemy aircraft vs friendly aircraft….

The Problem is Real to the Business.
The business problem of algorithms and embedded systems is understanding, that:

  • Algorithms' uniqueness may require special boundary considerations.
  • Algorithms will probably have lower delivery rates.
  • Experts concerns are often emotional, but their perception is their reality and their
  • pronouncements are very well respected.

Our experience is that the following do not represent algorithms: check digits, general date formatting, calculating totals, sorting….

The following would be considered algorithms: date parsing for an international publisher. Dates may come in many formats (April 1999, Apr 99, 4/99, 4/1999, or European format of dd/mm/yy, …). A chemistry related algorithm is Unique Table Generation (UTG). It determines the unique hexagonal numbering (ordering) for the nodes in a chemical structure. The total design and development required 1800 hours and three doctoral theses have been written about it.

The Problem is Emotional.
Due to the extra effort and potentially extra skill sets needed just to understand the problem, experts often feel function points do not accurately represent the work it takes them to develop and maintain algorithms. This reaction is understandable but misplaced because, function points don't measure work, they measure size. The analogy we use is "Which weighs more, a 1000 pound gold ball or 1000 pounds of chicken feathers?" It is a trick question. They both weigh the same. But to a FedEx agent, who must get both across the country overnight, they are two different business problems. The ball requires special packing and security; the chicken feathers for an oil spill may take a whole plane. But, in either case, the agent must know the weight of the load to calculate the fuel necessary for the flight.

The Threat to Any Measurement Program
Due to the business importance and need for any algorithm or embedded system, its development is assigned to the most talented and respected developers. These are the same people who must support measures or they will die.

The Secret
The secret to algorithms and embedded systems is to recognize them and the emotional issues they bring. Then put them in their own sub-boundary and acknowledge:

  • their uniqueness and contribution,
  • their special requirements and skill sets,
  • their "unusual" delivery and support rates.
Having their own sub-boundary means that they can easily be located and updated if the counting rules ever change or are further defined.

What is the process?
To count algorithms we recommend the following process:

  1. Create Inventory - Name, description, language(s), LOC, #variables, #rules…
  2. Isolate boundaries if necessary, document any assumptions made or verified. This is important if formal rules are passed in the future.
  3. Categorize using available objective criteria (#variables, #rules…)
  4. Review with the perception of the application experts as L,A,H.
  5. Do an application wide validation check. For example, if Algorithm 1 is perceived as twice as complex as Algorithm 2, does it have twice the variables, or twice the LOC….
  6. Do a cross system validation check, if possible. Here experts and managers from multiple applications rate identified Algorithms. See if these mirror the experts' evaluations. This is actually done now, on pure opinion, via the Budget process.
  7. Assign FP values using the ILF Matrix (L=7, A=10, H=15). Check this for reasonability via the number of variables using the middle row of the EI matrix. For example, if an algorithm has 8 variables, it would be average, if it had 16 variables it would be high.
  8. Review with management.
Remember, algorithms are not defined in IFPUG CPM 4.2. However, we once presented these ideas at an IFPUG conference and unbeknownst to us, Boeing presented the same ideas immediately afterwards. They called theirs 3-D function points. Capers called his ideas feature points, but he has since recommended just using function points.

More important than the name, (We are jealous we didn't think up a good catchy phrase for a button.) is that all three of us independently reached similar conclusions. We cannot speak for all companies, but we know of at least 250 companies, who do give algorithms a similar view.

Embedded Systems
To count embedded systems, use a similar process of inventorying, isolating boundaries, categorizing and validating. Then identify "What is Happening?" This will breakdown what the embedded system does, into EI's, EO's and possibly EQ's, and ILF's. The user in these cases may be another machine, but that is all right. Since these components are already defined, a simple note suffices to document how and why they are counted.

Related Issue
Some companies have a related issue of Internally Generated Transactions. These, like algorithms and embedded systems, are not specifically addressed in the counting rules, yet they provide user functionality based upon user requirements. For more information contact us.

What are CEO Bragging Rights?

Bragging Rights are provable facts, good enough to brag about. They are usually overlooked until a good measurement program is installed.

One client, with a systems staff of over 3000, was recently considering "Outseasing" a central table maintenance function. When I measured it I found the team had achieved a productivity delivery rate five times (5X) better than associated teams and a quality rate ten times (10X) better than the proposed overseas agent. Their story had become a "CEO Bragging Right" that was totally overlooked until we applied measurements.

Ten minutes into a half-hour review with the CIO, he stopped the meeting. "Would you like additional resources?" He asked. Considering all of the cutting and downsizing that had been going on, the manager and director were stunned. It was a contingency they had never imagined. Ten minutes later the CIO stopped the presentation again. "Are you sure you have enough resources. I don't mind throwing money at things like this."

Recently I ran into another client facing impending budget and performance reviews. We had set up their measures but he wasn't sure how to use them to his advantage. Together, we found his TOP index had improved by 87%. That was all he needed.

The future of managing IT, is being able to determine and act upon your Unit Cost for Software, (Development, Maintenance, Staffing, Rollout, Risk, Value…) the same as you would act upon any change in your Unit Cost of operation. Unit Cost, can be your first Bragging Right, valuable enough to become a CEO Bragging Right.

What is Business Management Advocacy?

Business Management Advocacy or Increasing Management Effectiveness By Replacing Opinions With Facts.

There are three measures that transcend every part of every business: time, money and functionality. Time and money have been around forever. However, without the Function Point model, Management loses 1/3 of the measures that can give them insight into their businesses. This is especially true in the E-economy, where business management has even less understanding of what is going on.

You Can't Manage What You Can't Measure!

If a financial audit discovered a flaw or lack of controls in cash processing, it is an auditor's responsibility to identify it and review it with top management. In the E-economy, IT budgets continue to grow, independent of Risk or Proof of Value requirements. Therefore, it seems to me that auditors and IT have the responsibility to identify this and offer an independent service to help manage it. I call it Business Management Advocacy.

Business Management Advocacy
Imagine the CEO in the Boardroom during the annual budget review. Marketing promises a 10% increase in Sales. The CEO says "That is not good enough, I expect 20%!" He/She can measure results (Facts!) to see if the goal is met. Meanwhile, Operations promises to meet the 20% increase with an X% change in this or that. The CEO can again measure results to see if the goal is met. IT says "Trust Me. We need a new system." CEO's lose control and are essentially held hostage by Opinions. They don't like it. They need Business Management Advocacy.

Business Management Advocates represent the CEO and Users of IT services. They have to prove their 20% improvement. IT does not.

Who are the users? They are every department within the organization, except IT. Take a conceptually simple concept of Payroll. There are more than 20 Users of Payroll information. They include, employees, supervisors, management, the Payroll, Human Resources and Pension departments, The Controller for cash flow, taxing authorities, the Social Security Administration, Banks for automatic deposits, the Legal department, … They all need Business Management Advocacy to prove their 20% improvements.

Remember the 10 Dimensional Integrated Measurement Model with the colored arrows. We discuss it from the IT view. However, everything about it is even more true for Users. As a Risk Management tool it has no competition. Unit Costing, Estimating packages, comparing vendors… all of the savings we have described, are now approachable from the User perspective, thereby increasing Management control and effectiveness.

What are some of the issues of Chargeback?

Chargeback systems should include function points, as well as other appropriate factors. The traditional CPU seconds, memory, disk space… may be meaningful as a starting point. However, as more organizations go to value added pricing, increased functionality becomes a better representation to a client of what they are getting. A system of 5000 fp has twice as much functionality as one of 2500 fp. Since the client can use all of the functionality for all of their accounts, they are receiving twice the value.

I see banks looking everywhere for "Fee for service" charges. In the US, some banks have tried charging people for using "Tellers." Others are now charging for each "cash" machine transaction. They are charging whatever they can get whenever they can get it. Usually these fees are waived when customers have off setting balances, typically $2500 or $5000 in savings or idle checking accounts.

We see "leveraged" function points as an easy way to sell fee structures. For example: suppose a bank service has 1000 function points (1000 unadjusted fp X 1.0 adjustment factor). Further suppose it is used by 100 customers in the main branch. Its leveraged business impact is 100,000 leveraged fp (1000 X 100=100,000).

Now suppose it is enhanced. The unadjusted fp goes up by 10% to 1100. The adjustment factor goes up by 10% to 1.1. The total function point count, available to the bank is 1210 fp, a 21% increase. Now suppose the enhancement allows the system to be used in all branches and the anticipated market is 500 customers. The leveraged business impact is 600,000 leveraged fp (1100 X 500=600,000). This is a six fold (6X) increase in the leveraged business impact. I would not charge 6X more, but is a lot easier to talk about a 50-100% increase.

This example has been dramatically constructed to demonstrate the concept. It works best in high growth situations. Large mature systems do not change so dramatically. Large systems may instead utilize "Cycle Time" to justify chargeback increases.

Cycle Time
Cycle time is the amount of time it would take to totally rewrite the system. For example, suppose a 12,000 fp system is experiencing 1000 fp worked per month. If this is strictly "changes" with no new functionality added, the cycle time is 12 months or one year. If the original development was $1M I can justify on-going charges of $1M per year. One assumption here is that users request all of the "changes." Since there should be no learning curve, and since test systems and data bases are in place, we should be experiencing increased productivity. We often see productivity of enhancements doubles, thereby providing a "profit" margin.

Supplemental Measures
There are many supplemental measures within a bank that can broaden a "user's" understanding of Information Systems value. We recommend using these in conjunction with function points to explain value. Supplemental measures are measures already collected by the company or bank so there is no additional overhead and no question of their validity. For example, the $100M total IT budget may be spread across 10M accounts or 500M checks and ATM transaction processed. Supplemental measures make our cost per transaction look very low.

How many points can a Function Point counter count?

Data collection and analysis: DSC has counted over 20,000 applications and projects. Platforms range from legacy mainframes to the most advanced client/server and even peopleless environments. Our experience shows that you can expect "counting rates" of:

300 Function Points per day (FP/day) by first-time counters with one day's training.
800-1,000 FP/day (IBM 's stated counting rate is 800 FP/day.) with our class, case study and one week of "FASTPATH" counting assistance.
1,000-1,500 FP/day with "FASTPATH" after two weeks of assisted counting.
We have observed that if 10 people have supported a system for 10 years, we can probably count it in one to two days with them. For every 12 developers, we can probably count their existing base in one (1) week.

Guarantee: We guarantee DSC personnel will count at least 2,100 Function Points or two systems per counting day with proper documentation and system expertise. If this is not achieved we will prorate our fees. This is seven times (7X) the rate of "first time counters".

To improve consistency and accuracy in counting, DSC's "Fastpath" process includes:
-Following all IFPUG 4.2 standards.
-Delivering "audit" quality documentation of the Value Adjustment Factor and each detailed functional component (ILF, EIF, EI, EO, EQ). Any assumptions or other relevant comments are appropriately documented to your company specific documentation standards.
-Paying particular attention to establishing clear understandable boundaries. This is often the most misunderstood aspect of measuring. It causes misperceptions, miscommunications and mismanaged expectations among staff, users and management.
-Exercising "due diligence" in working with your subject matter experts to insure accuracy while avoiding any impact on existing project schedules. We have participated in studies and audits that have proven our accuracy levels at 98.8%.
-Utilizing special communication skills that facilitate gathering information and gaining staff and management support in the process. Our goal is to expand the management 80/20 rule to 98+/10, with experts "buy-in" to the process and the results. This means we can get better than 98% accuracy in less than 10% of the time you will spend on your own.
-Employing a common sense "roll up our sleeves" approach resulting in the least cost, greatest value, with quality deliverables.
-Function Pointing on an ongoing basis will take less effort than time reporting and/or travel and expense reporting.

Crawling, Walking, Running?

Q. "We believe in metrics, however, one needs to crawl before walking and walk before running. Function point counting is running and we are just not there yet."
A. Crawling, walking, running may be true, but does it apply here? When a baby is born, the first thing we ask and check, are Metrics. 10 fingers, 10 toes? How many pounds? How many inches? When your baby starts to crawl, do you care how fast or do you care how old he/she is? If a baby is one and not crawling, I would be worried. If a child is two and not walking, I would be worried. Competition and comparison are basic to life.

Crawling, walking and running are milestones. We use different measures around these and other milestones. Functionality is the fundamental measure of Software. Along with Time and Money, they are the only three measures that cross every part of every business. They are fundamental properties like size, weight, temperature, state (solid, liquid, gas), …. They are fundamental before Quality. A fundamental property of crawling is age (or maturity of process?). A fundamental property of walking is distance (or maturity of process?). A fundamental property of running is speed, direction, attitude, conditioning, determination, …(or leadership?).

Fully 1/3 of all new CIO's are from outside of IT (and we all know the average life expectancy). They may be from another place in the business or from an entirely different business. How can these people get their hands around the problems they face? They need to know the fundamentals. Remembering that "Quality is a Journey not a Destination," they need to know where they are now.

Since this is often framed in terms of Risk, lets take the example of companies that are just starting the journey. A fundamental question would be, "What is my biggest system and is it my biggest RISK?" Most companies know this. However, when asked what is the second biggest system, there is never a clear consensus among management. They can agree it is either B, C, or D. Then comes the backbreaking question (that highlights Senior Management frustration and why the new CIO is from outside IT). "How much bigger is the biggest than the second biggest?" Nobody knows. So how can I compare the Quality of before and after, within and without, of Production to Distribution, of Financial to HRMS? How can I identify the Best of the Best? How can I make myself, and my Quality Program, look good?

How much better is a company after starting QA methods? 10%, 20%, 100+%? How much testing is enough? How many test scripts will we need, …? Functionality provides the "stake in the ground" necessary to measure progress. Function Points is standing still before and after running. It is part of the fundamental agreement about where we will run to and how long we expect it should take. It is the Unit Cost of our product. In my experience, Vendors that know their Unit Cost are more profitable. Customers that know the unit cost of what they are buying usually buy more.

Quality and Metrics are not competitive. They are both required for any successful endeavor. The real Quality in a QA program is to be able to bring insights that lead to better decisions.

"If I don't know where I am, why should I take the Quality train. And if you can't tell me where I will wind up, then how much respect have you earned?" Happy and Healthy computing.

What about Estimating?

Q. Is it possible to estimate any project, sight unseen, regardless of technology, in two minutes, quantifying and documenting all assumptions, with a guarantee? AND can you do it as early in the life cycle as "gleam in the eye" of the boss or customer? A. YES, if you are prepared!

As you might expect, we use Function Points to do this. Estimating is one of the very first uses and benefits of installing Fastpath Function Points and Metrics.

The purpose of the estimate is to CYA against specification creep. E.g. if the project is 1000 FP and if we determine your historical Delivery Rate is 25FP/Work Months, where 1 WM=130 applied hours, then we could expect the project to be 40 WM.
If during Design, the project grows to 2000 FP, everyone remembers the 40WM estimate. No one remembers any of the caveats that were made with the estimate. The 1000 FP estimate quantifies and documents every assumption made, regardless of where you are in your lifecycle.

Furthermore, the 100% increase in the project size (from 1000FP to 2000FP), with no increase in resources and/or timeframe, quantifies the Risk this organization and management is taking. What are the chances the team can increase its productivity by 100%? What are the chances the organization can make a profit if it must deliver 100% more product for the same price? Yet managers unknowingly step into this trap everyday.

If I can double my productivity, I should be able to leave Wednesday at noon for the rest of the week. Is this happening in your organization?

Q. Isn't everything based upon the historical Delivery Rate? Not all projects and/or applications have the same rate. How does this get factored in? A. You are correct. Different applications, different systems, different projects, different teams have different Delivery Rates. Delivery Rates are affected by four variables: Hardware, Software, People (including Staff, Users, Management) and Requirements.

We help you establish, validate and interpret your Delivery Rate for your situation.

Q. I need this. Jump to the bottom line. How do I do it and how long will it take? A. If you want us to do a single project, getting prepared will only take a couple of days. If you want to gain a full knowledge transfer, the process is: Q. I need this. Jump to the bottom line. How do I do it and how long will it take? A. If you want us to do a single project, getting prepared will only take a couple of days. If you want to gain a full knowledge transfer, the process is:

What are Internally Generated Transactions? Do they get counted?

Suppose a clerk receives a report with calculated totals (EO). If any item exceeds a certain value he/she goes to a terminal, updates an account (EI) and submits parameters (EI) for a letter with calculated totals (EO) to the customer. Assuming all of the above are average and VAF 1.00, the systems provides 18 fp.

Now suppose the above process in automated, via an existing batch process. The first report is eliminated; the update is automatic; and the submission of parameters for the second report is automatic. Does the functionality of the user (management) decrease?

At first, it appears functionality may decrease, but only by 5fp, the amount of the first report. (This report may change to an audit report of what the system is doing automatically. If so there would be no reduction in fp's.) The rest is simply being automated. In fact, it is these types of enhancements that reduce overhead. Suppose this enhancement saves $100,000. It can only occur by magic or automated (less expensive) functionality.

Yet the FP model might not pick it up. How to account for it? Several of our clients count these situations as Internally Generated Transactions. These are identifiable transactions that may occur in a batch or real time process. They meet all of the identification rules except "received from outside the application boundary." Actually they would meet this criteria if they were sent off to another system and then that system resubmitted them.

These transactions are technically outside the "published rules." They are only identifiable if you are wearing the "common sense" hat. They must be validated by the application expert and documented separately as a company specific interpretation.

In a related issue, suppose Application A sends a message (EO) to Application B (EI). (Application B does update an ILF.) If Application A and B are reorganized into a new combined Application C, does user functionality decrease?

This gets into physical versus logical. We always defer to the application expert and/or the user expert. However, we ask "What is really happening? What is the user view of it." It must be documented and validated by the application expert, not the FP counter.

What are some of the issues of Outsourcing and Service Level Agreements?

Function Points have become a standard of any Outsourcing Service Level Agreement (SLA). However the best agreements of the most successful companies that have outsourced, have some very specific items that have saved the companies millions. Outsourcing agents that fully understand their own capabilities, usually have no problem with these.

Productivity
If I am outsourcing any service, I need to understand my own productivity. If my team(s) can deliver 25 FP/WM, I would require that the outsourcing agent match it and agree to productivity gains in the future. For example, I expect a 20% gain a year from my own staff, so I would also expect it from my Outsourcing agent and write it into my SLA. If they don't make it, I would require lower fees.

Productivity includes project work, support and maintenance. I expect gains in all areas.

This requires baseline function point counts and project counts. One of our clients "bit the counting bullet," paid to have us count and saved more than 50 times (50X) our fees the first year.

One client established budgets, based upon the historical Function Points. For example, if a team or staff produced 1000 function points last year, the budget for this year is 1200. The user can now choose how and where to use their budget.

Another client established a staffing level for Support, about 15% of the staff. If you have 300 developers, this would be about 45 people. Even though these people were outsourced, they were dedicated to support and not available to be applied to projects. Therefore, the SLA called for 45 support people, plus 1200 additionally developed function points in Year 1 of the SAL. Year 2 called for 45 people plus 1440 additional function points.

Quality
Direct Quality measures, with Function Points, can be reflected in the above Productivity rates, or they could be separate measures. For example, if the Weighted Defects/FP ratio is .xxx, the I would expect a 20% improvement of my own staff or of my outsourcing agent.

Experience Index
I once worked for the largest insurance company in Wisconsin. It was one of the premier companies to work for. They always had middle management problems and after five years, I left and went to a Utility. 5 ½ years later, after I had advanced as far as I could, I left the Utility and went to the bank.

At the same time, the insurance company outsourced. Their staff was given 30 days to sign up with the outsourcing agent or they would be terminated. My first assignment at the bank was to increase the staff. I hired 50 people in three months. 15-20 of the people, including a dozen superstars, were from the insurance company.

This was a major, major error on the part of the insurance company because they did not include an Experience Index in their SLA. In interviewing the people, they seemed to have a common reason for leaving. As one superstar told me, the insurance company had been telling him for years that he was doing a great, great job. Then one day, the insurance company said they were unsatisfied and going to outsource. This destroyed any sense of loyalty with their staffs and many, many left. They were replaced by the Outsourcing agent with "rookies," thereby saving the Outsourcing agent a couple million dollars. One user manager that I had worked with, said he called the Outsourcing agent for some systems help. In walked a "kid" straight out of college. The kid pulled out a blank sheet of paper and asked what the manager wanted. What the manager wanted was a seasoned analyst who could understand and recommend alternative business solutions, not just a note taker. Today the company is still in business, but it is no longer the leader in its field. It is not even among the top 1/3. The only one who benefited from outsourcing was the Outsourcing agent.

Now think about it. Who will leave? The people with the most skills can easily obtain good positions in other companies. Who stays, the people with the least skills. The insurance company "shot itself in the foot."

The Experience Index, is a measure performed by you to determine the overall staff experience level for Data Processing Experience, Company Experience and Assignment Experience. Lets say a person has 10 years experience in DP, 5years at this company and 1 year in his/her current assignment. That becomes a "floor" or base experience level that the SLA should require be maintained with penalties for noncompliance.

The floor needs to be analyzed. For example, on small systems, 0-1000 FP, and medium sized systems, up to 2500 FP, experience may not be extra critical to the business. However, on major systems, for example Deposits, Loans, Credit Cards… in a bank, it often takes several years of experience to fully understand effects of some system changes. This is the knowledge and experience base that is core to the business that the SLA should address.

Measuring the staff and consulting on yearly changes could be a service you provide.

Outsourcing Oversight
One of the keys to successfully managing an Outsourcing agreement is to manage the SLA. There are many new tasks, requiring an active staff, about 2-3 percent of the people being outsourced. This staff also takes on many user business decisions that today, are often ignored, often with great loss of opportunity. So if a company is going to outsource 300 people, you can become part of the deal.

The new tasks and ignored decisions occur whenever an IT department is "below average." What does this mean? To understand this we need to work backwards. If every IT department were ranked by efficiency from 1 to n, how much profit could an outsourcing agent make by taking over number one. Probably none. Therefore they only look for the below average least efficient "shops."

In consulting with any company thinking of outsourcing, you should make them aware of some of the changes that they must make in their processes. Too often, companies view signing an SLA as a panacea or "cure all."

What is a Productivity Risk Analysis?

Productivity Risk Analysis For Estimating and Project Progress Assessing

When a project team, or external vendor, gives a company an estimate, how does the company or CIO/CEO know the estimate is correct? They should trust it, because the team or vendor did their best job? Is anyone ever engaged to do less than their best job? No! The question is how does the CIO/CEO know the estimate is correct? The same is true of a Project Progress Report.

"We can estimate any project, sight unseen, regardless of technology, in two minutes, quantifying and documenting all assumptions, with a guarantee!" AND we can do it as early in the life cycle as "gleam in the boss or client's eye" once we are "prepared."

Imagine the value if a CIO/CEO could validate an inside (or outside vendor) estimate. How about a "second opinion" for a high visibility, business critical project.

How do we "get prepared?" By collecting historical function point data; by establishing a Software Unit Cost, and by performing a Productivity Risk Analysis.

The Software Unit Cost is reflected in the Function Point per Work Month Delivery Rate. A Productivity Risk Analysis analyzes the Hardware, Software, People, and Requirement issues of a project, in comparison with similar projects within the company.

The result is an independent, fact-based analysis, allowing management to make informed decisions, or take corrective action, before it is too late. It reduces Risk.

The Productivity Risk Analysis does not eliminate management responsibility to apply managerial judgment. Management will continue to make business decisions for business reasons. However, better information usually leads to better decisions.

How much does it cost? A Productivity Risk Analysis costs less than zero dollars. We proved this with one of our saddest, but true, stories. Once we were called to a major manufacturer to check an outside estimate (bid) for a project. We did and saved the company $1.4M in one day. The sad part was that we were not working on commission.

Productivity Risk Analysis: You Can't Manage What You Can't Measure!

What is the 10 Dimensional Metrics Model?

The 10 Dimensional Model displayed on the this website Home page is the most memorable way to help executives, managers and staff understand and relate the value of measurements in key business issues.

Until the late 1980's, there were only two measures that transcended every aspect of every business: time and money. Now there is a third measure, the amount of functionality a person or process has to do a job. Functionality, in conjunction with other measures, provides information and insights never before possible. Do you know of anyone who isn't affected by computer software?

The 10 Dimensional Model starts with Function Points (FP) measuring functionality. Even if you have no knowledge of functionality, you can understand that 10,000 FP is five times (5X) more capability than 2,000 FP. Function Points are a relative measure like Dow Jones points. A 10,000 Dow is better than 2000 Dow.

When FPs are combined with time, they provide productivity information for development and maintenance (support) environments. Once established, productivity can be analyzed and projected forward for estimating based upon actual experience rather than "trust me" salesmanship.

Quality, Cost, Staffing, Process and other performance measures can also be normalized and compared. For example, if defects increase by 20%, is it good or bad? It sounds bad, but what if the size of the system increased by 100%? The defects per function point is actually a steep downward trend. Comparisons can be made across technology, across business functions and to a great degree across system design.

Company goals and objectives can be universally applied and results reported. Said one CIO, "I expect every unit (team, department, division….) to get 20% better and prove it."

All of the above address CIO/CEO questions and issues about "Are we doing things right?" Once these has been proven, we move onto the more important issues of "Are we doing the right things?"

These issues are factored in with Leverage, Customer Satisfaction, Skills (PeoplePoints) and Vision measures.

The model addresses some of the business questions your management faces (below) with little if any data to help in business decisions. BEWARE, this is a moving target. Today's issues and measures to help address them, will be out of date tomorrow. Hence the framework, rather than a fixed set of 100 reports. How many of the following Issues can you answer for your management?

Time - Productivity & Estimating (Development/Maintenance)

Defects (Quality Index)
  • Is your quality rate (defect index) improving? By how much? Can you prove it?
Cost (ROI & Risk Analysis)
  • Is your company cutting costs or cost conscious? Does it know where to cut? Can you prove it?
  • Is your company interested in improving its ROI? Would you ask for 5, 10, or $50 million without calculating some kind of ROI? What was the ROI on your Information Systems budget last year?
  • Can you add quantified data to risk analysis?
Staffing (Build, Buy, Outsource)
  • Are you understaffed, overstaffed, or properly staffed? Can you prove it?
  • Do you select software? Can you quickly quantify Build vs. Buy or Product-to-Product evaluation?
  • Is outsourcing a good or bad idea? Can you prove it? Will you lead and control the discussion or just be a walk-and-see victim? Process Do you use CMM, TQM, ISO, Baldrige or other Process measure? Can you normalize comparisons?
Process
  • Do you use CMM, TQM, ISO, Baldrige or other Process measure? Can you normalize comparisons?
Usage (Leverage)
  • Can you measure the corporate contribution of your systems accounting for changes in size, complexity, and usage?
Customer Satisfaction (Value)
  • What is the value of your IS investment? Can you prove it?
Mission Criticality (Importance)
  • Can you prove you are working on the right things?
  • What are the critical success factors in installing measures?

What is a Software Balanced Scorecard?

Software Balanced Scorecards - The Icing on the Cake

Today, in the world, there are three measures that transcend every part of every business. Most businesses only use the first two, time and money. The third is software functionality. It measures the tool set capability people have to do their jobs.

By measuring functionality, and combining it with other existing measures, IT can provide fresh insights to Management to address Business issues like:

Productivity and Estimating,
Quality,
Costs, Revenues, ROI,
Staffing,
Process,
Business Impact,
Business Alignment,
Customer Satisfaction and Loyalty,
Value,
Mission Criticality and other changing Business conditions.

IT can address Tactical and Strategic Goals and Objectives with multiple metrics that increase understanding and focus of Employees, Suppliers and even potential Customers. For example, recently a major bank acquired another major bank. It now has two of all of its systems. Which ones should it save and which ones should it discard? What is the productivity of each? What is the Quality level and or unit cost? …

A Software Balanced Scorecard can address all of these issues and more.

Overcoming Fear and Objections
Selling the value of measurements has always been tough. "After all we've lived this long without them, so we'll survive if we don't get to them until tomorrow." Probably, but the attitude causes a slow, barely perceptible, erosion of Senior Management confidence and control.

To prove this, ask yourself a simple question. "Would you invest $20,000 of your Pension money in my business?" I've never had a taker. Why? Because people don't understand what I do.

Now imagine your CEO wondering if he/she should approve a million dollar item within your budget. After all, he/she doesn't understand exactly what you do? He/she does understand the alternative, put it in the bank. At a minimum, expenses are reduced by $1M, revenues are increased by $50,000 and he/she can get a bonus. So why should he/she approve that $1M item?

Plus don't forget the Y2K scare, the perception that IT overspent all that money without enhancing competitive advantage. So what was the ROI of your Y2K budget or your dot.com subsidiary? What is the ROI of any budget item, of your whole budget?

Whenever I've told CEO's and CIO's "You Can't Manage What You Can't Measure." I have never had anything but complete agreement. Yet middle management invariably delays, losing credibility, when a simple Software Balanced Scorecard, can prove the immense value added of IT.

What is a Software Balanced Scorecard?
A Software Balanced Scorecard is an easily understood, select set of metrics that quickly convey Business information. Items reported should help management make better decisions by replacing opinions with facts. For example three departments claim they improved productivity by at least 10%. Which one did best? What were the key success factors? Will those factors transfer to other departments? Which department had the best Quality and by how much? …

A Software Balanced Scorecard, that shows Productivity increased by 25%, Quality improved by 32%, and/or Maintenance was reduced by 30%, can quickly prove the value of a staff, the success of Management policies or both. Even more valuable is when the Software Balanced Scorecard alerts Management to misperceptions or problems before it's too late to react.

As business conditions change, so should the Software Balanced Scorecard. One year the Business focus may be on Productivity. The next year the focus may be on Quality, the next on Value…. It is imperative that the Software Balanced Scorecard responds to these Management needs or it will become overhead, hence expendable.

Software Balanced Scorecards must also be meaningful and "current." For example, Operations departments have reported many statistics for many years. However, once a performance level becomes a "standard perception," it becomes nearly useless for further reporting, like reporting 99% "up time."

The Software Balanced Scorecard allows normalization for comparative purposes. The ability to normalize comparisons is one of the key features of Software Metrics.

Who should prepare the Software Balanced Scorecard and when? A Software Balanced Scorecard is a collaboration of Business Management, User Management and IT. Reports are prepared by IT after consulting with Business and User Management about their current information and insight needs.

In selecting the skills of the reporter, obviously, analytical and communications skills are essential. However, entrepreneurial, risk taking, inquisitiveness, patience and self-starter skills may be equally important. Remember, the most successful "Metrics Maniacs" are people looking for new business insights for business reasons, rather than "techi kudos."

When some reports are requested on a regular basis, it is important that a "reasonable" schedule of activity be established to avoid impacting project schedules. For example, if a bank wants information about Deposit, Loan and Trust Applications, monthly reporting will probably add overhead. In addition, how much can major systems change in one month and how much "change" can management undertake in one month? Therefore we suggest staggered scheduling. For example, for a bank, report on Deposits in January and July, Loans in February and August…

Other Important Measures for a Software Balanced Scorecard. The actual Software Balanced Scorecard below reflects the Needs of one Business at a particular point or span of time. Other measures are equally important, valuable and limited only by the circumstances and perspective of the requestor.

Since business needs are constantly changing, it is recommended that each report be followed with agreement on what the next report should contain. It is our experience that as soon as Management gets better information to make better decisions, they move on. "What's next?"

Scorecards are milestones, not destinations. A Software Business Scorecard addresses changing business needs versus a Shelfware Scorecard with a fixed set of reports that never seem to meet changing management questions and needs. Shelfware Scorecards have pre-determined solutions before determining requirements.

In selecting a Scorecard tool, companies may make a mistake that can be politely discussed based upon this new approach of requirements before solutions. I have never seen an IT Installation improve because of a Scorecard. Improvement only comes with serious quantified Management commitment, e.g. a goal that all units improve productivity or quality or their TOP Index by 20 %. The most successful and advanced IT Installations I know, use simple Software Business Scorecards, with minimum bureaucratic overhead. They quickly address specific management issues and provide insights into business issues such as staffing and resource allocation.

To discuss this, I recommend my clients supply three projects they consider "successful and typical," in terms of Hardware, Software, People (Staff, Users, Management) and Requirements. This is referred to as the H/S/P/R. The projects can be different applications and/or technology. We measure these projects for their productivity (FP/Work Month), where 1WM=130 applied hours. Let's say the three projects achieved 25 FP/WM, 20 FP/WM and 10 FP/WM. Different projects can and will have different productivity. All of these rates are OK based upon their H/S/P/R and management's view that they were successful.

Next we provide a one page Software Business Scorecard of what we have achieved at the Client. It should contain:
# of systems counted,
#FP, Counting Rate,
Staff Support Rate,
Unit Cost Rate,
The Three Project Delivery Rates and maybe a TOP Index.
Rack and Stack reports by: System, FP, VAF, Unit Support and Unit Cost.

The total expenditure should only be about one-two work days after you're done your baseline count. More importantly, we have now established the Client's own historic database of projects. Three projects they know and understand versus hundreds of tool set projects they know nothing about (and can't find out about, because they are blinded).

Conclusion: Think outside the box. Be responsive and PROVE IT with a Software Balanced Scorecard!

Does Time to Market apply to Software?

Time to Market: a Marketing Strategy
In order to improve Client Business productivity, e.g. Total Assets/Employee, they must analyze their H/S/P/R needs from the Business view, not just the Systems view. In doing this, organizations we know, usually require better systems with reduced Time to Market.

Before we can prove reduced Time to Market, we need to establish the Client's current Time to Market rates. This is done by analyzing the Client's new historic database of successful projects. Let's suppose, for a technology set, we find the average successful project is as follows:

3 WM/1000 FP for Requirements and Planning,
6 WM/1000 FP for Analysis,
9 WM/1000 FP for Design,
12 WM/1000 FP for Construction (Programming, Unit Testing, System Test),
3 WM/1000 FP for Regression Test and Implementation.

We can now analyze other projects, especially those being estimated and in progress, to see how they compare to the Client's own benchmark of successful projects. Variations are framed in terms of RISK. For example if a project estimate is 1 WM/1000 FP for Requirements and 16 WM/1000 FP for Construction, the ratio is 1:16 versus the benchmark ratio of 1:4. This project appears to have a 4X Risk factor. The natural management question is why. If the team can explain, there should be no problem, but if they can't, management may have uncovered a major problem, when there is still time to do something about it. We call this a Productivity Risk Analysis, see attached.

Likewise, once a phase of development is completed, we can confirm that a project is on schedule or having problems. For example if the estimate called for 3 WM/1000 FP but the actual time was 6 WM/1000 FP, the Function Points in conjunction with the Client's benchmark rates could determine quickly and early if a problem exits, even if the team is not reporting any problems.

In analyzing other projects, both new and completed, we can refine the Client database. We can even establish ranges of tolerance, so that management need only review projects outside the ranges.

Time to Market now becomes a Marketing Strategy. Measurements are needed to start and keep the Client's database up to date, and to control Risk. However, major management issues of productivity, quality, cost, staffing, process improvement and others, still need improvement. Your company in conjunction with selected partners can now offer services to improve and prove Time to Market.

These services could include Process Improvement, Quality Assurance, Test Preparation and Analysis, Cost Analysis, ROI and Benefit Assessment.
 
 | Home  |  Services  |  FAQs  |  Online Proposal  |  Famous Buttons  |  Success Stories  |  Available Presentations  |  Contact Us  | 
Copyright © 2001 Development Support Center. All rights reserved.Send comments to