Home

European Competitive Measurement Symposium 2008
Paris, 5-6 February 2008

This Conference was initiated and conducted by the French Consultancy Spirula (www.spirula.fr) in the lovely Parisian ‘Bateau Mouche’ location on the river Seine. In all aspects it was a resounding success, not only because of the way it was conducted but more particularly because Spirula managed to have the participation of the major personalities in the software metrics world, i.e. Tom Gilb, Dan Galorath, Ton Dekkers, Roberto Meli, SMS, and so on. The purpose of this text is to emphasise Telmaco's and its partner SMS’s contribution to this event.
The river Seine and its barges Predict your project budget with R2S2
Grant Rule (Managing Director of SMS Ltd) gave an introduction to the now familiar ‘Rule’s Relative Size Scale’ (R2S2) (www.measuresw.com ). This is an observational way of classifying the software projects in terms of their sizes. In a superficial fashion, R2S2 leads to the classifying of the size of software projects as is done for pieces of clothing, such as ‘size M’ for medium, ‘size L’ for large, etc. We know well the benefits of this classification when shopping for clothes; hence we have an idea of what it can do for the programme manager when budgeting. The presentation took us to a deeper level of its impact on managerial decision-making processes. This classification considers only ‘successful’ projects, and demonstrates that around 90% fall into the small to large size range that is S, M1 and M2, to L ranging from about 100 to 3,000 CFP. The consequences of this observation do not stop there. This dimensioning is not a high accuracy measurement but it provides a trade-off curve usable around the very beginning of the life cycle, when accurate measurement is not yet feasible, in order to provide a very early estimate of the budgetary and schedule feasibility of a particular project. It can therefore save a lot of time, effort, and money for the organisation. Being aware of the powerful advantage thus given to its user, Telmaco has implemented R2S2 in its software size measurement tool, MeterIT-Cosmic (www.telmaco.co.uk ). This enables the measurer to capture very early the foreseeable size, budget, and duration of the project being considered and to carry out early money-saving management decisions.
The Critical Success Factors for a software metrics programme
Despite the availability of project predictability-assurance metrics-based methods and tools such as R2S2 seen above, metrics techniques do not seem to be put into use in the software industry as much as they could be. Charles Symons (Joint Project Leader, COSMIC ) talk aimed to present industry-wide evidence illustrating that proper use of good software metrics leads to good performance, whereas poor - or even no - use of metrics leads to poor industry performance. He first presented evidence that software metrics are still not mainstream. To understand why, he proposed a set of five Critical Success Factors (CSF) for a software metrics programme and observed that, in his experience, there are two CSF’s on which programmes most often fail. These are that the metrics must be credible to all stakeholders, and must be seen as complementary to the project processes.
A speaker in action He then examined the current state of measurement of five key performance parameters (productivity, speed of delivery, quality, and delivery to time and budget) and the corresponding industry performance. Quality, in terms of defect density, is relatively easy to measure, is generally well-measured and industry performance, at least in some sectors is amazingly good if one thinks of the reliability of the world’s communications, transportation and financial services systems. Conversely, productivity and speed of delivery are rarely or poorly measured (think of FPA or SLOC) and the evidence is that the software industry performs poorly on these parameters. Delivery to time and budget is also poor which is mostly due to poor estimating methods and scope control procedures. This in turn is related to inability to size new software requirements and to estimate reliably from past productivity.
The three keys to improvement
The memorable dinner gala From this analysis, Charles concluded that the key to overall improvement in software industry performance would be to adopt a combination of (a) a better software size measurement method (COSMIC, naturally, www.cosmicon.com ), (b) a good parametric estimating method (www.telmaco.co.uk ), and (c) a good project scope management process such as SouthernSCOPE (www.egov.vic.gov.au). Charles concluded by estimating the financial benefits that could be achieved resulting from improved productivity measurement and estimating, if the software industry adopted these improved approaches. With even the most conservative assumptions, the software industry could produce the same output at 25% less cost per annum. These benefits cannot be achieved without improved metrics.
Start your own benchmarking
While waiting for metrics to be recognised at organisational level and for results showing continuous software improvement from measurement, project benchmarking and estimating based on metrics, what can the programme manager do? This is where Bernard Londeix (Director, Telmaco Ltd) proposed Do-It-Yourself Benchmarking. The basic motivation for DIY Benchmarking is to increase one’s skill in managing his or her software project portfolio. For example, we know that our management decisions are good when they result in continuous and measurable reductions in cycle-time and/or productivity improvement. This is the best way of attracting more budgets for more projects. And this starts with the measurement of projects. But what is a project? Everyone has his or her own definition. Discussions continue during the breaks And is the definition measurable? This presentation illustrated just that: how measureable projects are identified, and how some specific data needs to be collected and analysed. Naturally, measuring a software project includes measuring the software produced by the project. Then, as above, we are back to the COSMIC method for sizing software (www.cosmicon.com ). The natural consequence is that a significant amount of data needs to be kept correctly for future analysis. Simple spreadsheets offers attractive facilities but require some preparation time, which is not always available to the Programme Manager. Fortunately, some ready-to-use tools exist (www.telmaco.com ) for carrying out local benchmarking. Hence, when the corporation is ready for benchmarking, the programme manager is already ready, in order to achieve the best result. See ECMS2008-Benchmarking.pdf
The Eiffel tower wasn't far away A landscape for software metrics
The Conference’s set of presentations managed to present the problem of the software project, not only as a manageable piece of work, but also as a potentially profitable endeavour for the enterprise. The quantitative aspect of the project was introduced in terms of effort, duration, and finance. Also, the very valuable contributions of other speakers were very worthy of consideration: they all together provided an almost complete landscape of the applications of software metrics to an organisation’s software capability. This was made possible thanks to the excellent conference http://www.europeancompetitivemeasurement.org/en/) set up by Spirula (www.spirula.com). For supplementary information into software metrics topics, readers are invited to visit the SMS Ltd site www.measuresw.com or contact information@measuresw.com. For more information on the MeterIT tool suite contact enquiry@telmaco.co.uk.
Top 

To contact us E-mail: enquiry@telmaco.co.uk
your comments and views are always welcome
Copyright © 1999-2008, Telmaco Ltd.