Towards Reducing the Costs of Benchmarking – Part 2 : Simple Function Points

Introduction

Part 1 of this article set out some of the issues associated with software sizing which contribute to high cost of benchmarking.  Part 1 also discussed Automated Function Functions as one potential solution.

This second part focuses on new research which seeks to reduce benchmarking costs in another way –  by simplifying the IFPUG method for sizing the software product. The outcome of this research is discussed in this article.

Simple Function Points – The Impetus

Why try to simplify the Function Point sizing method?

The use of Function Point Analysis has waxed and waned over the years, for a variety of reasons.

Anecdotally, those organisations who have implemented the method often find it just too hard to build and retain the knowledge needed for the successful implementation of the technique and even where they do, find the cost of sizing more than they wish to spend.

In more formal research, the Forrester Group released a report in 2009 entitled Function Points: A Critical Analysis Of The Pros And Cons Of Adoption. Whilst this report is a few years old now, in my opinion its content is still true today.  One of its findings pointed to the 600 pages of rules for the IFPUG sizing method rules as a significant barrier to take up.

Organisations want easy, fast and agile measurement methods while still achieving reliable results.

There are publicly available as well as proprietary methods  which have sought to address these issues.   An example is the David Consulting Group’s FP Lite ™ method (see  http://www.davidconsultinggroup.com/insights/publications/fp-lite-an-alternative-approach-to-sizing/). Typically, these alternative methods adapt or simplify existing approaches and measure their success by assessing the size produced against the size from the more detailed traditional method.

The research behind Simple Function Points takes a different approach.

Simple Function Points – The Research

Data Processing Organisation (DPO)  is a long time established company in Italy specializing in software measurement and related services and innovative products such as the Early and Quick  Function Points for IFPUG Function Point Analysis. Roberto Meli is DPO’s CEO.

In late2010, DPO  initiated a research project with the objective of simplifying the sizing process. It specifically sought to:

“Define a new functional  measurement consistent with the framework of the ISO 14143 family of standards, totally  compatible with the IFPUG (method) when applied on the same object of measurement, but ..

1. Easier to apply
2. Easier to learn
3. Less susceptible to different interpretations
4. Less susceptible to “manipulation” of measurements
5. Designed to allow an easier update of existing measurement assets
6.  Designed to allow an immediate conversion of existing assets counted with the IFUG method”

The first two points are very important in addressing the issue of cost. Complicated rules take time to learn and are so very easy to misinterpret or to completely forget.

The last point, that of compatibility with the IFPUG method, ensures that existing organisation and industry assets in the form of benchmarking databases are preserved and can continue to be used.  An issue with newer sizing methods, function points or otherwise, is that these database assets are essentially lost as there is no compatibility or conversion between sizing methods and collection of benchmark data must start anew.

For those who may not be familiar with the ISO certified functional size measurement methods, there are two principal steps in the sizing.

• The first step analyses the software product and  breaks it down into  the functionality delivered. These functions are formally referred to as the Base Functional Components (BFC) and are more or less equivalent to functions as users of the software would see them.
• The second step assigns a weight or score to each function where the score attempts  to express the complexity of the function. Both steps are governed by the rules for the specific sizing method.  The scores for each function are then totalled to give the overall size.

The starting point in DPO’s research was to question whether the second step, that is, assigning the complexity weighting, was actually making the resultant size measure any ‘better’ for its primary intended purposes of benchmarking and estimation.

DPO’s initial research was conducted using a sample of about 800 project s from the International Software Benchmarking Standards Group (ISBSG) database.  This study showed that:

“The accuracy of a model of correlation between actual effort and the software functional size does not decrease when considering only the number of BFC.”

In other words, the extra precision of the further classification and detailed sizing of complexity was not delivering a better correlation of size to project effort.  The effort of detailed sizing with its attendant cost  was not increasing the usefulness of the resultant size obtained.

However, the size as simply a count of numbers of functions  does not allow continued use of  benchmarking data based on function points.

Thus the next part of DPO’s  research was to find a structured way of converting the simple count of functions for a software product  to the Function Point size as would be obtained using the detailed IFPUG method.  This ‘same’ size is in a statistical sense, of course.

The result is a conversion method which identifies two generic function types equivalent to the Transactional Function Type class and the Data Function Type class of the IFPUG method. Each generic function type is then assigned  a constant single generic weighting, 4.6 for Transactional Function Type and 7.0 for Data Function Types.  This simple assignment of weights is in marked contrast to the other tedious and lengthy processes required under the most prominent sizing methods. As an observation, these values are very close to the IFPUG Average weightings.

Simple Function Points – The Outcome

The method has been named Simple Function Point.

The research findings were first presented by Roberto Meli to the United Kingdom Software Metrics Conference (UKSMA) in 2010. A copy of this presentation can be found at http://www.uksma.co.uk/conferences/conference2011/presentations/07RobertoMeliSimpleFunctionPointDescriptionV2.pdf.  The method continues to be well received.

In June 2011, the Simple Function Point Association (SiFPA) was formed with Robert Meli as President. SiFPA (www.sifpa.org) now maintains the rules and is currently developing a Measurement Manual. Since the technique is compliant with the ISO 14143 framework, it is their intention, in time, to be ISO certified.

Additional research by DPO in a small number of their client organisations has confirmed the same findings.

Simple Function Points – The Benefits

The key attraction of this method is the simplicity.

• The method can be easily learned in a day, rather than the 2-3 days of full IFPUG training. This immediately is a cost saving. Where there is less to learn, there is less to forget.
• The method is easy to apply. It  uses the same rules for identification of functions as the IFPUG method but allows the complicated, often arcane, rules for complexity to be simply ignored.
• Sizing activity is 2 – 5 times faster than doing a complete detailed IFPUG count. This represents a significant cost reduction for benchmarking.

Importantly,  it supports  continued use of IFPUG benchmarking data so these assets are preserved.

My own organisation has long held the view that the additional effort of the detailed count was not delivering additional value and so we welcome this research supporting our observation and experience. The simple discipline of identifying all the functionality in a piece of software delivers immense value to an organisation or project, whether for benchmarking, estimating or managing project scope.

Of course, there is likely to be push-back from some Software Metrics professionals who will see their skills as being devalued.  The IFPUG Certified Function Point Specialist exam tests knowledge of the full 600 plus pages of rules.

However, failure to listen to what the market actually wants usually ends in tears.  In this regard, the Simple Function Point solution is worthy of serious consideration.

About the Author:

Robyn Lawrie is a director and principal consultant for CHARISMATEK Software Metrics (www.charismatek.com) .  She is the Vice Chair – Metrics for QESP and Vice Chair – International Function Point Users Group (IFPUG) Membership Committee.

Robyn has more than 40 years of IT Industry experience. A major focus of her career has been  the improvement of the software process in general and, in particular, the use of metrics in Software Requirements Management, Scope Management and Estimation.