Article Preview
TopIntroduction
Since 1994, the Standish Group has periodically produced the CHAOS report, which presents the state of performance of information technology (IT) projects. From 2011 to 2015, between 27% and 29% of the projects were successfully completed on time, on budget and with satisfactory results, but on average 71% were either challenged or considered a failure (Hastie & Wojewoda, 2015). The annual CHAOS reports have been the target of criticism for their methodology. However, since their first edition, these reports revealed that a substantial proportion of IT projects have presented significant problems.
Since the first edition of the CHAOS report, requirements have continued to be a key concern in IT projects (e.g. Bormane, Grzibovska, Berzisa, & Grabis, 2016; Meth, Mueller, & Maedche, 2015; Rosenkranz, Vranesic, & Holten, 2014). While clear requirements are a major success factor, incomplete and changing requirements are identified as a major cause of the problems leading to project failure (e.g. Alflen, Prado, & Grotta, 2020; Babar, Bunker, & Gill, 2018; Batra & Bhatnagar, 2019; Davey & Parker, 2015; Jukic & Velasco, 2010; Pacheco & Garcia, 2012a; Rosenkranz et al., 2014; Sandhu & Weistroffer, 2018; Taghavi & Woo, 2017). Despite this, little effort has been made within organizations to tackle this issue (PMI, 2014).
Most studies on IT requirements from a business perspective were conducted in the context of transactional projects, which are undertaken to automate business processes such as payroll or accounting systems, ERP, CRM or e-commerce web sites. However, new types of systems associated with business intelligence (also called descriptive analytics) have emerged (Foley & Guillemette, 2010; Jukic & Velasco, 2010). Since these systems are designed to support decision-making and managers in their business performance monitoring activities, their requirements are very different from those of transactional systems. Business intelligence requirements are based on decision-making processes rather than on the conventional day-to-day operational processes of the organization (Jukic & Velasco, 2010). They are loosely structured, poorly shared and unsynchronized with the evolving organizational context (Jukic & Velasco, 2010).
Although context is important, when it comes to determining technique effectiveness, a technique that is effective for one project is not necessarily effective for another. (Lane, O’Raghallaigh, & Sammon, 2016). Most of the published studies on requirements do not consider these differences related to the type of systems, i.e. whether it is transactional or business intelligence. However, professional literature is paying more attention to these differences and tends to indicate that business intelligence requirements are often more difficult to satisfy than the requirements of transactional systems (Briggs, 2015; Sherman, 2015; Wells, 2008;Whitehorn, 2012). This suggests that different approaches should be adopted by analysts depending on their project’s development context (Prakash & Gosain, 2008; Taghavi & Woo, 2017; Wells, 2008a). Nevertheless, the literature does not provide any guidance on which techniques to use specifically in the context of business intelligence.