Running To Stand Still: How Organizations Motivate Users To Participate In Maintenance Upgrades of Packaged Software
Abstract
Maintenance upgrades are IS projects undertaken to replace an existing version of packaged software with a later version from the same vendor for the purpose of supporting IT maintenance tasks. Although maintenance upgrade projects can be large organizational projects with high risks of failure, little research has examined them. This paper investigates the question of motivating user participation in maintenance upgrading projects. Motivating such projects is especially challenging because users typically anticipate little benefit from the project. Via an in-depth case study using the lens of communicative framing, we discover that: (1) a negatively valenced communicative frame characterizing an external party as a threat is best for motivating users, (2) framing the maintenance upgrade simultaneously shapes user motivation and IT Support Group behavior, and (3) carefully managed anxiety can facilitate the success of a maintenance upgrade project.
Key words: packaged software upgrade, strategies, IT implementation, framing
1. Introduction
A recent trend in organizational computing is the increased adoption of packaged software in-lieu of custom software to perform business operations.The scope of computing applications available in packaged versions range from enterprise applications designed to support core business processes to operating systems designed to support productivity applications for individual users. While the merits of packaged vs. custom software can be debated (Cassidy, 2006), implementing packaged software applications has become common for organizations of all sizes (Satzinger, et al., 2004). By using packaged software, organizations overcome the need to develop computer applications with an in-house staff. However, this advantage does not remove the need for organizations to maintain packaged software. The maintenance processes associated with packaged software differs substantially from those of custom software. Instead of rewriting customized code, maintenance for packaged software is usually performed via software upgrades or service pack applications in which a company installs a new version of the same software provided by the same vendor.
For organizations, upgrading packaged software is a non-trivial activity that can take a year or more and expend resources equal to 20-30 percent of the cost of the original software. Packaged software upgrades may not work as planned and can create serious organizational problems by interrupting service for various lengths of time (Beatty and Williams, 2006). For example, Blackberry’s e-mail network experienced a four-hour outage as a result of software upgrades (Miller, 2008). Also, when the UK’s Department of Work and Pensions performed a routine software upgrade, it caused 80 percent of its 100,000 PCs to be inoperable for a full day (Rohde, 2004).The incident was one of the biggest computer crashes in government history. In another case, Lockheed Martin had to down its SAP servers for 71 hourswhen upgrading from SAP 4.7 to 6.0, after which they experienced slowdowns in database queries and user lockout problems (Bjorlin, 2008). Software upgrade projects are therefore expensive, high risk propositions.
Sometimes packaged software upgrades are justified on the basis of increased business benefits (Ng, 2001). However, in many cases, upgrades are conducted merely for maintenance reasons. In a maintenance upgrade, the software is upgraded for the express purpose of facilitating maintenance rather than business improvement. In the most common case, upgrades occur because software vendors establish “sunset” dates beyond which they do not support earlier versions of packaged software.When these sunset dates arrive, adopting firms are forced to either upgrade to a subsequent version, or maintain the packaged software themselves (Light, 2005).Because organizations often do not have access to the packaged software source code, or lack the capability to understand it, upgrading to a new version becomes the only realistic choice available (Khoo and Robey, 2007).
Maintenance upgrades are unusual in two ways. First, a maintenance upgrade is typically not about improving users’ productivity or business process change. Organizations that choose to upgrade only when software reaches its sunset date implicitly show that revisions to core business process are not sought or even welcome. Maintenance upgrades are therefore unusual in that a company undertakes the risks of upgrading software even though the project provides few intended benefits for business users. This does not imply that maintenance upgrades are unimportant for organizational success; a failure to conduct a maintenance upgrade could cause necessary software in a company to malfunction. Second, user input is often irrelevant to the decision to perform a maintenance upgrade. A maintenance upgrade is typically proposed by the IT department to facilitate system maintenance rather than bringing visible benefits to users. Thus users may be excluded from the decision process regarding maintenance upgrades, which are seen as necessary and inevitable.
Given these distinct characteristics, it is unclear how business users can be motivated to accept and cooperate with a maintenance upgrade project. We assume that user acceptance and cooperation is necessary for a successful upgrade, yet most users would be reluctant to devote effort to an upgrade project that promises them little benefit. The objective of this study is, therefore, to identify effective strategies for eliciting willing user participation in a maintenance upgrade. To answer this question, we perform an in-depth analysis of a single maintenance upgrade case using the lens of communicative framing (Benford and Snow, 2000; Snow and Benford, 1988).
The study makes three contributions. First, we demonstrate the value of a negatively valenced communicative frame characterizing an external party as a threat for encouraging willing user participation in maintenance upgrades. Second, we show that framing the maintenance upgrade simultaneously shapes user motivation and IT Support Group behavior. Finally, we demonstrate that carefully managed anxiety can facilitate the success of a maintenance upgrade project.
in maintenance upgrades,a message that motivates willing user participationsimultaneously shapes user and IT support group behavior. To motivate users, the IT support group must behave in a manner consistent with the message. Second, one effective strategy to motivate user participation is to articulate (1) the negative consequences of failure, and (2) that the root cause of the project is an external threat. Third, our study highlights how carefully promoted anxiety can facilitate maintenance upgrades.
The paper is organized as follows.First, we review related literature in two areas: research on software upgrades and communicative framing theory.We propose communicating framing theory as a response to neglect of user motivation in the software upgrade literature and summarize the insights from this theory in a process model for maintenance upgrades. We then present our research methodology. Findings from the case studies are then presented, following which we analyze the case.We conclude with suggestions for future research.
2. RELATED LITERATURE
2.1Packaged software implementation and upgrades
Packaged software is usually chosen over custom software to reduce development costs, shorten implementation times, acquire state-of-the-art “best practices,” reduce maintenance, and obtain extended functionality (Boehm and Abts, 1999; Fenema, et al., 2007; Light, 2005; Light and Sawyer, 2007; Talbert, 1998; Voas, 1998). However, software problemsmay arise after packaged software is installed. One ongoing challenge in package software management is coping with the evolution of software packages (Boehm and Abts, 1999), as vendors change software capabilities by introducing new versions and terminating support for older ones (Kohl, 2005). Client organizations may influence vendors, but lack direct control over software support. Upgrades initiated by vendors can introduce unwanted additional functionality or eliminate existing desired functions (Talbert, 1998). In addition, upgrades are costly. An average ERP upgrade can easily cost the implementing firm USD 1.5 million (Bonasera, 2002).
Academic research that addresses packaged software issues has predominantly focused on initial implementation (Fenema, et al., 2007; Gefen, 2002; Hitt, et al., 2002; Lucas, et al., 1988; Robey, et al., 2002; Sia and Soh, 2007; Thong, et al., 1996). Of the little research on packaged software maintenance, the vast majority address descriptive research questions, instead of building theory to explain “how” or “why” organizations cope with packaged software maintenance (Hirt and Swanson, 2001; Light, 2001; Nah, et al., 2001). The sparse research that attempts to build theory focuses on the decision to upgrade (Khoo and Robey, 2007; Ng, 2001; Sahin and Zahedi, 2001) rather than the process of upgrading. To our knowledge, only Nah and Delgado (2006) explicitly explore the upgrade process by comparing the critical success factors for package implementation with those for package upgrades.
Nah and Delgado report that critical success factors related to change management, such as user training and coordination across all affected parties, were important to ensure upgrade success. Thus, user involvement in training and coordination are important to the success of a maintenance upgrade (Ngai, et al., 2008; Parr and Shanks, 2000). New versions of packaged software are typically designed to suit a generic business, so users must either adapt to these generic prescriptions or participate in the customization and configuration of the package to fit with existing business processes (Robey et al., 2002). Users are also intimately familiar with the idiosyncratic adaptations and workarounds used to overcome limitations in packaged software (Boudreau and Robey, 2005). For example, Excel spreadsheets are often employed in place of database systems to allow users more control over data. These unintended work practices are often unknown to a maintenance team and the packaged software vendor.
Furthermore, a lack of user participation in package maintenance can create difficulties. For example, users may refuse to relinquish their IT equipment to allow the maintenance upgrade. Similarly, users can complain about the upgrade to senior management, or otherwise disrupt work on the maintenance upgrade. Given the critical role of the user, it is therefore important to understand how one can motivate users to participate in a maintenance upgrade.
2.2Communicative Framing
Communicative framing theory attempts to explain how messages can be structured to motivate others to action (Benford and Snow, 2000; Snow and Benford, 1988). As such, this theory offers the potential to explain the mobilization of users to participate in packaged software maintenance upgrades. Communicative framing is related to, but not identical to decision making framing (Druckman, 2001; Sheufele, 1999). Decision making framing studies how cognitive processes influence decision making. In decision making framing, communication about a frame is assumed to reflect the genuine cognitive beliefs of the communicator (Fiol, 1994; Fiss and Zajac, 2006). In contrast, communicative framing examines how deliberately structured messages influence others’ actions (Druckman, 2001; Sheufele, 1999).
Decision making framing has been used in both IS and management research to understanddecision making processes (e.g., (Butler and Gray, 2006; Davidson, 2002; Fiol, 1994; George, et al., 2006; Hodgkinson, et al., 1999; Howard-Grenville, et al., 2003; Neale and Bazerman, 1985; Orlikowski and Gash, 1994; Qualls and Puto, 1989). For example, Orlikowski and Gash (1994), and Davidson (2002) examine the issue of technological frames, which reflect distinct stakeholders’ assumptions, expectations, and knowledge about technology. They argue that the relative congruity or incongruity of technological frames across different stakeholders influence decision processes and outcomes related to the development and use of IT. However, these studies do not address the ways in which one group of stakeholders shape decision making frames to influence other stakeholders, which is the situation in IT contexts where IT professionals wish to gain user cooperation. For such situations, communicative framing provides a stronger theoretical foundation than decision making framing.
Communicative frames have rarely been applied in IS, although they are used in the marketing literature to show how binary characteristics of advertisements influence targeted recipients. For examples, such work has studied the impact on consumer purchasing decisions of emphasizing a promotion window as restrictive or expansive (Cheema and Patrick, 2008), or of stating a discount as a percentage or in a dollar value (DelVecchio, et al., 71). Most work in marketing communication frames examines whether framing an advertisement as avoiding a negative situation, or achieving a positive outcome encourages consumer purchasing (Chang, 2007; Chang, 2008; Dardis and Shen, 2008; Loroz, 2007; Veer and Pervan, 2008).
Our purpose, however, is not to understand how messages can mobilize purchases, but rather how messages galvanize organizational stakeholders to action. We therefore turn to communicative framing research in the communications (Sheufele, 1999) and social movement literatures (Benford and Snow, 2000; Snow and Benford, 1988; Snow, et al., 1986). Research in these fields addresses such issues as stakeholder response to particular interpretations of the American invasion of Iraq (Schwalbe, et al., 2008), and explaining how social groups can motivate governments, businesses, or individuals to ban smoking (Creed, et al., 2002; Kim, 2006). The literature in these fields revolves around two dominant questions: (1) how positive and negative framing influence effort, and (2) how social context influences the construction of the communicative frame.
2.2.1 Negative/Positive Framing
Most research on negative and positive framing stems from prospect theory (Kahneman, et al., 1982). In a set of classic experiments that have since been replicated (Kuhberger, 1998; Levin, et al., 1998), Kahneman and Tversky (1982) demonstrated that describing the potential outcomes of a disease outbreak in a negative way encouraged risk-seeking behavior. Further research, has demonstrated that negative framing is associated with greater goal-directed effort (Levin, et al., 1998). Thus, a project situation framed in terms of its likelihood of loss is more likely to elicit effort from project members than one framed in terms of potential gain. Similarly, a focus on the potential outcome of performance elicits greater effort than a focus on the process necessary to achieve the outcome (Zhang, et al., 2007). Some research disputes these findings, showing that in particular situations, framing a message to accentuate the positive is more effective than framing it to accentuate avoiding a negative situation (Chang, 2007; Chang, 2008; Loroz, 2007; Veer and Pervan, 2008). This research concludes that context determines whether a frame should be constructed positively or negatively.
One insight from this literature is that one need not frame a maintenance upgrade positively, i.e., that it will result in a “better life.” Instead, a maintenance upgrade can be framed as necessary to avoid a “worse life.” Such a frame is likely to fit maintenance upgrades more, since neglect of maintenance can reasonably be associated with some form of system failure, and hence a “worse life” for all concerned. In contrast, when maintenance is performedwell, users often fail to see how their lives are improved.
While the communicative framing literature argues that a negatively framed message is useful for motivating maintenance upgrades, the change management literature would suggest otherwise. IS projects, including maintenance upgrades, create disruptions to the social order. This in turn creates negative emotions and anxiety (Bordia, et al., 2006; Tvedt, et al., 2009), that in turn causes performance to fall(Fulgate, et al., 2008). It is thus, critically important to manage and reduce these negative emotions and anxiety to encourage project success(da Cunha and Orlikowski, 2008; Kotter, 1996). Therefore, the change management literature would suggest that positively framed messages should be employed during a maintenance upgrade.
2.2.2 Communication Context
Other characteristics of a communicative frame can influence its effectiveness. However, most research has generally found that the identification of these other characteristics depend strongly on the specific communicative frame being examined (Chyi and McCombs, 2004; Hoffman and Slater, 2007; Mastin, et al., 2007; Simon and Jerit, 2007; Yioutas and Segvic, 2003). For example, for temporary discounts, the question arises as to whether the window of the discount is positioned as being restrictive or expansive (Cheema and Patrick, 2008). A further question is whether the discount is framed as being in absolute monetary terms, or as a percentage (DelVecchio, et al., 71). Such dimensions of discount frames are inapplicable, for example, to a frame arguing for the restriction of civil liberties to fight a war. In the latter frame, questions of whether the frame features individuals or groups (Boyle, et al., 2006), whether the frame is written in an objective or advocacy tone (Aday, 2006), and the extent to which “heroic” images are used (Schwalbe, et al., 2008) are more relevant.
The wide body of research in using communicative frames has identified two general factors that influence a frame’s resonance with recipients: frame credibility, and frame salience(Benford and Snow, 2000; Snow and Benford, 1988). Frame credibility relates to the recipient’s ability to accept that the frame’s message contains the truth (Benford and Snow, 2000). Frame salience relates to the importance of the message to recipients of the frame.
Frame Credibility. Two dimensions of frame credibility are especially important in our research, empirical credibility and frame consistency. Empirical credibility relates to the believability of evidence presented in support of the frame. For example, frames concerning second hand smoke are supported by evidence that such smoke causes cancer (Creed, et al., 2002; Ferguson and Gallagher, 2007; Lee and Aaker, 2004). In contrast, early framings of the nuclear disarmament movement suffered from empirical credibility issues, because there was no evidence that a nuclear war would result in a nuclear winter (Snow and Benford, 1988).
Frame consistency relates to the continuity of the message as it is created and disseminated across three core framing tasks: diagnostic, prognostic, and motivational (Benford and Snow, 2000). Diagnostic frames describe a particular problem and attribute blame or causality. For example, in cigarette smoking, a diagnostic frame would identify second-hand smoke as bad and smokers as at fault. Prognostic frames not only articulate a solution to the problem but also identify strategies that will be used. A prognostic frame would articulate that smoking should be banned in public places. Motivational frames develop rationales for action that encourage groups of stakeholders to act. Motivational frames to ban smoking would, for example, target governments, restaurant owners, and others who manage public places (Kim, 2006). Effective frames must be consistent across their diagnostic, prognostic and motivational forms.
An important element of frame consistency is the valence of the frame. A diagnostic frame that is positively valenced that leads to a negatively valenced prognostic frame would be inconsistent and therefore ineffective. However, two positively valenced frames can remain inconsistent if other elements of the frames do not align. For example, a prognostic frame that suggests an upgrade to a more bloated version of software would be inconsistent with a diagnostic frame suggesting that the organization would be more efficient if processing time was improved, even though both have the same positive valence.