A case study on the impact of customer communication on defects in agile software development

Author(s): Korkala, M.; Abrahamsson, P.; Kyllonen, P.
Venue: 2006 Agile Conference
Date: 23-28 July 2006

Type of Experiement: Case Study

Quality
3

Reference: http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=1667565
SUMMARY:

This study focuses on the effects of different levels and types of customer communication on the quality of a software project. Four different projects were studied, each using a different set of communication mediums to interact with the customer. All four development teams logged the amount of time spent fixing defects; the fourth team further broke down the time fixing defects by the cause of the defect. Defects were categorized as either customer-independent (developer and platform defects) or customer-dependent (cosmetic and requirement defects).

The first team had constant access to an on-site customer. The second team had an on-site customer available when necessary, and preformed some communication via email. The third team had face-to-face time with the customer at the beginning and end of the iterations, and communicated through email mid-iteration. The fourth team had an on-site customer for the first two weeks of the project, then communicated via email and phone.

It was found that over the course of the project, the fourth team had 60% more customer-dependent defects than customer-independent defects; it was further found that half of all defects on the project were requirements-related. Overall, it was found that the teams with more face-to-face customer communication had lower defect rates - 6.3% of development time spent fixing defects for the first team, versus 41.8% for the fourth team. There was little difference in software quality between the first two teams - one with an on-site customer, one with an on-site customer available when necessary, suggesting that having an available customer in the same facility may provide many of the same benefits as an on-site customer.

0