To better harness the power of their own data, many firms have invested in data warehousing technology. A data warehouse enables the collection and storage of vast amounts of data that is then extracted and analyzed by end users. Sophisticated software allows data mining, and gives analysts and managers unprecedented visibility into business operations. For instance, business intelligence software can use the data warehouse to track results as well as better understand the factors that led to them. Moreover, sales can be enhanced via customer relationship management, which uses the data warehouse to extract and explain complex buyer behaviors. It is no surprise that with its potential to generate extraordinary gains in productivity and sales, the market for data warehousing software and hardware was nearly $200 billion by 2004. The market for data warehousing technology will continue to expand because businesses highly regard its decision-support functionality and because it has been adapted to better meet the needs of a growing number of small- and medium-sized businesses.
Potential benefits notwithstanding, data warehouses are still large, expensive, and risky undertakings. For example, the median installation cost of a data warehouse is $1.5 million and can exceed $50 million, which does not include annual operating costs. In addition, it has been noted that 40% of all data warehousing implementations fail [11]. Even if the implementation succeeds, end-user success with the data warehouse can still be problematic.
Indeed, a troubling pattern in the growing literature about data warehousing is that end users experience substantial difficulty with their firms’ data warehouses [2, 3]. This seems paradoxical, as data warehouses are engineered to be user-driven, allowing end users to be in control of their data [8]. It is this paradox—that data warehouses are used even though end users find them difficult—we explore in this article.
Acceptance of Technology by End Users
End-user acceptance of information systems is one of the most relevant and researched issues in the IS literature. Corporate managers and IT departments have long been frustrated with the implementation of expensive information systems that end users underutilize. This pattern has led to the development of research models that seek to better understand this phenomenon.
One of the most successful of these models is the Technology Acceptance Model (TAM), which posits that acceptance of a new technology (measured as self-reported system usage) depends on the extent to which end users perceive the new technology will be useful (enhance job performance) and easy to use (require little effort). Research has shown that perceived usefulness and ease of use do an excellent job of explaining the extent to which end users will use the system [9].
While initial TAM research focused on the acceptance of smaller, simpler systems, recent research has begun to focus on more complex systems. Moreover, additional constructs have been added to the original TAM to better understand the interconnected relationships that explain its ability to predict system use [10]. One of these constructs important to the current study is flexibility, or the ability to align system inputs and outputs with changing end-user needs. Flexibility is embedded within the system’s features.
For example, data warehouse users would likely describe flexibility in terms of the system’s ability to analyze data in various dimensions over different periods of time, to roll up or drill down to different levels of detail, and to perform “what-if” analyses using a variety of output formats [7]. Thus, a flexible data warehouse would include a query system that allows users to modify existing queries to perform new functions by easily adding/dropping fields and/or changing the query’s time frame. A flexible data warehouse would also allow users to easily view query results in a variety of ways, such as in tabular or graphical form or by creating an Excel spreadsheet for further analyses. Thus, to end users, the ability of the data warehouse to adapt to changing user needs is an essential characteristic of its flexibility.
It is not surprising then that it has been suggested that system flexibility indirectly affects system usage because of its effect on end-user perceptions of ease of use [4, 6]. Nonetheless, other researchers have posited that flexibility has its effect on system usage through its effect on usefulness rather than ease of use [7], and still others suggest that flexibility directly affects system use [1]. We compared the basic TAM with alternative models that included the different relationships posited for flexibility to determine which model best explained system usage.
Research Methodology
A survey including the original TAM items adapted to fit a data warehousing environment was sent to managerial-level data warehouse users in a number of major Midwest U.S. corporations. The survey also solicited other information, including the industry and size of the user’s company, the user’s position and department, the amount and type of system-related training the user had, what system-support was most useful to the user, and the amount of experience the user had with the data warehouse. The survey items (grouped by construct) are shown in Table 1.
Items regarding perceived usefulness, ease of use, and flexibility were measured using a five-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). Data warehouse usage was measured using a similar five-point Likert scale, with 1 indicating little usage (“a few times per month”) and 5 frequent usage (“several times per day”). Sixty-three managers from 40 organizations were contacted, with 33 managers from 29 organizations agreeing to participate. Surveys were emailed to the 33 managers who disseminated the surveys to another 59 in-house respondents. Of the 92 returned, seven surveys were incomplete, which resulted in 85 usable surveys.
Results
Survey results were analyzed using regression analysis, factor analysis, and structural equation modeling. First, data warehouse usage was regressed on usefulness, ease of use, and flexibility to determine the variance explained by the three constructs. The resulting adjusted R2 value (0.223) is well in line with results reported in previous studies using perceived usefulness and ease of use to predict actual system usage [9].
The revised TAM items were then factor analyzed to determine how well they grouped together into three separate constructs. The results shown in Table 2 indicate that the revised items loaded cleanly on their respective constructs.
Finally, the revised TAM items were analyzed using structural equation modeling methodology. This allowed us to determine how well the data fit the basic TAM model as well as three alternative models. Model 1 is the basic TAM that includes measures of ease of use, usefulness, and system usage. Models 24 examine the indirect effects of flexibility on data warehouse usage through ease of use and usefulness respectively and on data warehouse usage directly. Model 2 uses Model 1 as a starting point, adds flexibility as a third construct, and assesses its effect on ease of use based on research that suggests the effect of flexibility on usage is through ease of use [4, 6]. Model 3 is the same as Model 2, but assesses the effect of flexibility on usage through its relationship with perceived usefulness as suggested by other TAM researchers [7]. Finally, Model 4 is the same as Model 2 except Model 4 hypothesizes that flexibility directly affects data warehouse usage [1].
All four models indicate that perceived usefulness has a much larger effect on usage than perceived ease of use. Moreover, all four models indicate that perceived ease of use is a significantly positive antecedent of perceived usefulness. These findings are consistent with the literature on the TAM [4, 9, 12].
Models 24 represent the three different research hypotheses about the role flexibility plays in system usage. The hypothesized results for flexibility were obtained for Model 2 only—flexibility had a positive, significant relationship with ease of use. Conversely, flexibility was not significantly related to usefulness (Model 3) or usage (Model 4).
All four models were then analyzed using goodness-of-fit statistics to better determine how well each model fit the data. Because there is no one accepted statistic for determining model fit, several fit statistics were used. The conventionally accepted threshold for each statistic and the fit statistics for each model are presented in Table 3.
A comparison of the fit statistics indicates that Models 1 and 2 are preferable to Models 3 and 4. Both Models 1 and 2 appear equivalent in terms of fit, as both meet or exceed most of the recommended thresholds for the fit statistics. However, the Comparative Fit Index is less sensitive to errors caused by smaller samples, and based on it, Model 2 has a slight advantage over Model 1. The best-fitting model (Model 2) is shown in the figure here.
User perceptions of how easy it is to use a data warehouse are important because they directly and indirectly affect system usage.
Discussion
This study examined the impact of system flexibility on data warehouse usage by analyzing three alternatives to the basic TAM model suggested by the related literature on system flexibility. All four models were also compared to each other to determine which model fit the data best.
The research evidence that flexibility affects system use either directly or through its effect on usefulness was not supported. The best-fitting model supported the research evidence that flexibility affects usage indirectly through its impact on user perceptions of how easy the system is to use.
Increasing users’ understanding of the data warehouse (especially its flexibility) is likely to facilitate usage sooner. We believe this can be accomplished best by providing users with adequate training and support, both initially and on a continuing basis. We also believe it is important to provide all end users with adequate training and support whenever important features or components of the data warehouse are changed. This is especially important when those features/components might affect perceptions of the data warehouse’s flexibility and/or ease of use.
One might argue based on the results reported here that leveraging system flexibility will only increase system usage marginally. Although this may be accurate, even a small increase in usage can result in substantial monetary gains for the firm. For example, Wal-Mart estimates that its few power users earned the store over $12,000 per data warehouse query, and that those power users performed approximately 1,000 queries per day [5]. Adding just one more user to the “power user” category sooner by providing timely training and support would likely both pay for itself very quickly and produce substantial future earnings for the firm.
User perceptions of how easy it is to use a data warehouse are important because they directly and indirectly affect system usage. The flexibility embedded within the features of the data warehouse contributes directly to these ease-of-use perceptions. Interventions that help users better understand how to leverage a data warehouse’s flexibility can both increase system usage and produce a significant return on the firm’s data warehouse investment.
Recommendations
Recommendations for increasing data warehouse usage by leveraging its flexibility include:
- If users perceive that the data warehouse enhances their job performance, it will be used; it will be used even more if it is also perceived to be easy to use. The extent to which the data warehouse is perceived to enhance job performance is the most important determinant of its usage. However, to increase system usage, users must both perceive that the system enhances their job performance and is easy to use. System flexibility contributes to user perceptions of the latter. A data warehouse that users believe is flexible will be used more often than one that is not.
- Although the effect of flexibility on usage is marginal, it can nonetheless produce a significant financial return. Flexibility is not a major determinant of usage, and users will not use a data warehouse just because it is flexible. Nonetheless, when you consider the potential payoff of the data warehouse, even a small increase in system usage can have a significant financial impact on the organization. As evidenced by Wal-Mart, the returns produced by the query statistics of just a few sophisticated analysts (“power users”) were extraordinary; Wal-Mart’s return on investment far exceeded the cost of the initial implementation of its data warehouse. Imagine what the return would have been had Wal-Mart been able to accelerate the development of users to the “power user” level.
- Timely training and support can help users leverage flexibility sooner. System flexibility is embedded within the features of the data warehouse. Sophisticated users are more likely to leverage system flexibility because they are savvy enough to know where the flexibility in the data warehouse is; naïve users are more likely overwhelmed by the sheer enormity of the data warehouse. Efforts that accelerate users along the data warehouse learning curve (targeted training and support) should positively influence user ease-of-use perceptions sooner.
Join the Discussion (0)
Become a Member or Sign In to Post a Comment