Practical aspects of data collection in software industry

Pekka Forselius(STTF, Finland)

OBJECTIVE - the aim is to investigate the critical success factors of project data collection. The member organisations of the Finnish Software Measurement Association, FiSMA, have collected project data since 1991. The data collection form and tool support have been rather stable and simple enough from the first years of collaboration, and the anonymity, security and other quality issues have been solved. The best of the member organisations have been rather successful, while the others have either failed to institutionalise the data collection process, or even get it started in the first place.

METHOD - the author has been involved in the data collection since its inception. Mr. Forselius has also participated in many of the major movements and changes of the software industry in Finland, and especially among the FiSMA member organisations. As the primary database manager, he has analysed the data collection activity of each member company against the five factors of the Willmanıs model, which contains the critical success factors of software process improvement.

RESULTS - only very few FiSMA member organisations seem to have serious intention or motivation to reach the quantitatively managed maturity level. The reasons may include a lack of competition, but also the low capability of acquisition processes among the customer organisations. "The biggest problem of a CMMI level 5 developer organisation is the level 1 customer," Stated Bill Curtis (NASSCOM, Bangalore, 2005).

Experiences from the Finnish software industry indicate the data collection and benchmarking are more important at the higher maturity levels, as the organisations at lower levels are more interested in acquiring basic IT skills and software engineering tools, or establishing and defining basic project management processes.

CONCLUSIONS - Project data collection is never a business goal or a separately identified activity for process improvement at any level. However, project data collection is a pre-requisite to benchmarking, which has proven to be a solid approach to help organisations to reach their business goals, or to validate their process improvement. If the data collection is not related to ambitious enough business goals, it seems to fail within two years. To get the data collection successfully started and institutionalised, the organisation should be in the right place on the software process improvement map, and select the best approaches and concepts to find its own way to excellence.