POST-IMPLEMENTATION USABILITY OF ERP TRAINING MANUALS: THE USER'S PERSPECTIVE

PROBLEMATIC ERP IMPLEMENTATIONS are often associated with inadequate user training (Brown and Vessey, 2003; Roberts et al., 2003; Scott and Vessey, 2002). In a recent study of 30 manufacturing firms, user training was their top ERP problem (Duplaga and Astani, 2003). Training is consistently an under-budgeted item and is often the first item cut in the budget (Slater, 1998), despite Gartner's finding that each hour of effective training is worth five hours to the organization because well-trained users (1) reach the required skill level in less than a quarter of the time, (2) require less assistance from peers and help desks, and (3) spend less time correcting errors (Aldrich, 2000). Organizations spend as much as 50 percent (Davenport 2000) to as little as 5 percent of their ERP project budget on training (Wheatley 2000). Gartner research asserts that companies allocating less than 13 percent of project costs to training are three times more likely to have their ERP projects fall short of business and project goals compared with companies that spend 17 percent or more on training (Aldrich, 2000; Burleson, 2001).

Vendor-provided ERP training, at an estimated $2.8 billion annual revenue, is generic because it is not practical to tailor the course to each customer attending a class (Wheatley, 2000). However, a focus on software-specific training, rather than the business processes that underlie that software, is ineffective (Neiderman and Webster, 1998). At an ERP user seminar, 75 percent of respondents reported that next time they would allow more time for training, and would tailor it more around their own business processes (Wheatley, 2000). Delivering training too early or too late in the project, combined with the lack of tailoring, compromises training effectiveness. For users, learning the software is only one part of training; relearning their jobs is even more challenging (Schaaf, 1999) Users in new roles need support to perform their transformed work tasks (Roberts et al., 2003). Without adequate user support, the ERP project can fail, as it did the first time at Motorola's Semiconductor Products Sector. There was a smooth implementation on the second attempt when a change readiness team began with an audience analysis, designed new roles to support new business processes, and then developed role-based training.

A non-trivial form of user support is the ERP training manual, which ideally also documents organization-specific processes to support users' tasks.The objective of this research is to assess the perceived usability of manuals in the post-implementation period. This study could lead to improvements in the design of ERP training materials and the effectiveness of ERP training.

TRAINING MANUALS

Training manuals have evolved steadily. Early software-oriented manuals provided a reference to all the commands in the system, but were not useful in helping people complete their tasks (Rettig, 1991). The next stage in the evolution was task-oriented manuals, which gave more explicit assistance because all the steps to complete a task were specified. While perceived as useful by some, others believe the "recipe" approach may lead to deskilling and a decrease in discretion (Brockmann, 1990).

Specifying detailed directions results in task-oriented manuals becoming bulky, which often discourages adults who resist new learning (Rettig, 1991) and may not see the value of detailed instructions. A minimalist design gives the learner more to think about but less to overcome (Rettig, 1991). Research shows that business application training should go beyond step-by-step instructions for a task by explaining the rationale for the process embedded in the software (Huang, 2002). The minimalist manual was created in response to research on the poor fit between task-oriented manuals and how adults learn (Carroll et al., 1987/1988). While the minimalist manual design may help motivated users start using the software, reduce user errors, and coordinate user attention, some users may choose ineffective and unattainable goals and find learning by self-discovery to be ineffective and unsatisfying (Brockmann, 1990). Although research has shown that most people do not read documentation except as a last resort when they are stumped using the software and cannot get help from another person (Keirnan et al., 2002; Rettig, 2001), training new users with minimalist manuals reports impressive gains, such as reducing learning time by 40 percent when the number of pages was cut by 75 percent (Rettig 1991). Fewer pages reduces information overload and search effort for users.

Thus, finding ways to improve documentation is an important issue for both practice and research. Documentation is increasingly available online to take advantage of lower printing costs, easier distribution, hypertext, and display of current content. Nevertheless, Gartner recommends the paper-based channel for inexpensive handouts of quick references (Aldrich, 1999). Integrating documentation into the product, and integrating the design and documentation roles, could also lead to improved documentation usability (Keirnan et al., 2002; Rettig, 1991). Assessing the perceived usability of ERP training manuals is a logical starting point for research focusing on improving ERP documentation.

PERCEIVED USABILITY

Early pioneers in the usability field pointed to the fundamental principles of good design, such as having a good conceptual model before designing a product. Usability in simple terms is making sure that something works well, and that a person of average ability can use it for its intended purpose without frustration (Krug, 2000).

Usability was later applied not only to the design of everyday products, but also to computer interfaces and training manuals. The International Standards Organization (ISO) defines usability as "the extent to which a product can be used by specified users to achieve specified goals with effectiveness, efficiency, and satisfaction in a specified context of use" (ISO, 1999).

Effectiveness measures the "goodness" of the output by the user or whether users can perform their tasks, whereas efficiency has more to do with the quantity of work output in relation to the time, effort, and resources involved (Dillon and Morris, 2001).The satisfaction of a user is a measure of the user's attitudes and perceptions about the usability of an item. Products with good usability allow users to complete their tasks well in a timely manner and with minimum frustration.

Prior research has used a practitioner instrument to assess usability (Agarwal and Venkatesh, 2002). Similarly, in this study, the items assessing usability of the training manuals were adapted from the Questionnaire for Usability Evaluation (QUE), designed by US West Information Technologies (1996) to assess the usability of interfaces.The dimensions of usability assessed using QUE are navigation, presentation, learnability, and task support.

Navigation

Navigation refers to the methods by which users move around a system or documentation. Usability research recognizes the importance of navigability (Gillan and Bias, 2001; Nielsen, 2000; Palmer, 2002). Documentation should be easy to search (Nielsen, 1994), so efficiently finding information in the training manuals is not a problem. The manual should facilitate finding -what is needed quickly without too many steps.

Presentation

Effective presentation uses good graphic design principles for fonts, colors, layout, graphics clarity, and white space, while making sure not to distract users from their primary task (Gillan and Bias, 2001; Keirnan et al., 2002; Nielsen, 2000; Palmer, 2002; Shneiderman, 1998). Presentation organized with tabs, a table of contents, and an index help with navigation. Structure and organization facilitate ease of use (Agarwal and Venkatesh, 2002).

Learnability

Learning is an important issue for usability (Gillan and Bias, 2001). Unnecessary technical jargon and confusing acronyms deter learning. Terminology should use the language of its users (Keirnan et al., 2002). In QUE, learnability is defined as how easy a system is to learn so that users can begin to get their work done quickly. In this study, learnability' is applied to how helpful the manuals were to users in learning the ERP system. Documentation should not be too large (Nielsen, 1994), as a large manual with too many details is likely to cause information overload and hinder learning. Reactions to information overload include frustration and stress. As a result, users might avoid using the manuals. The use of hyperlinks in online documentation is a potential solution to information overload.

Task Support

Task support refers to the capability to help users perform their tasks quickly, effectively, and economically (Keirnan et al., 2002). Documentation should focus on the user's task, and list concrete steps to be carried out (Nielsen, 1994). The content should be relevant, have the appropriate depth and breadth, and have current and timely information (Agarwal and Venkatesh, 2002). Tailoring content would be effective for increasing relevance.

Navigation, presentation, learnability, and task support dimensions contribute to users' effectiveness, efficiency, and satisfaction. While the terminology sometimes varies, QUE dimensions are analogous to those proposed by usability experts and researchers (Agarwal and Venkatesh, 2002; Gillan and Bias, 2001; Krug, 2000; Nielsen, 1993, 1994, 2000; Palmer, 2002; Shneiderman, 1998).

THE ASP PROJECT

The University of Colorado (CU) has campuses at Boulder, Denver, Colorado Springs, and the Health Sciences Center (near the Denver Campus). The goal of the Administrative Streamlining (ASP) Project at the University of Colorado was to implement an ERP system to solve three major problems: (1) minimize the amount of administrative effort and expenditures; (2) replace technologically inferior, inflexible administrative information systems, many of which were over 20 years old; and (3) increase the availability of management information to personnel responsible for managing the human resources (HR) and finances of the university. The University of Colorado selected PeopleSoft's ERP system, which was widely adopted by U.S. universities. PeopleSoft modules for Finance version 7.0 public sector and HR version 7.51 were implemented at the University of Colorado on July 1, 1999, and July 1, 2000, respectively.

The ASP Business Plan budget was $32.7 million for the ERP system implementation, of which approximately $2 million was allocated to training costs (University of Colorado, 1998).1 According to the ASP Final Report, the project stayed within the overall budget although the project scope was reduced by about 10 percent, and almost $4 million was spent on operating (non-salary expenses) and training combined, which was $192,917 more than the original estimate (University of Colorado, 2001).

Another aspect of the ASP project was business process redesign. Prior to the ASP project, each campus handled its own procurement and human resources was administered independently. During the project, the Procurement Service Center (PSC) and Payroll & Benefit Services Center (PBS) were created to centralize procurement and human resources. The office of Change Management was created in 1998 to organize training and change management, and disbanded in 2002 after Finance training transitioned to the PSC and Human Resources training transitioned to the PBS.

Training was mandatory before new users were allowed onto the ERP system. The office of Change Management developed a variety of training opportunities for University of Colorado staff. Online tutorials allowed staff members to learn introductory PeopleSoft material at their own pace. Classroom-based training was available on each of the four campuses. Two of the campuses had dedicated training labs with updated computers and a live practice database available to users. Change Management staff, who worked on the training and help desks, developed the manuals in loose-leaf binders, for the training sessions and for reference. In the 44-page Training Overview manual in April 1999, financial administrative processes are named and a hyperlink directs users to narratives and flows of redesigned processes. The manual for Financials was about three inches thick, consisting of 473 pages, of which 163 pages were for Financial Core Training and 310 pages were for Procurement and General Ledger. The HR manual was of similar size. Most of the pages in the manuals listed steps for tasks accompanied by screen shots. A glossary, descriptions of commands or terminology, error messages, and troubleshooting guides were not included. Pages were numbered only within each chapter to allow for updates, which were often based on user feedback. User alerts informed users by e-mail of changes to the training materials and other training and change management issues.

After training, open labs provided drop-in assistance to users. Help desks were located in the Procurement Service Center and Payroll and Benefit Services Center. A Citrix front end to the ERP system provided shadowing capability on the help desk (Orten, 2001). Finally, Change Management developed about 50 stepby-step guides for Human Resources, 20 to 25 for centralized procurement, and a somewhat smaller number for the Finance module. (General Ledger was decentralized so a step-by-step guide was not available.) The guides, customized for the ASP project, were delivered online as user alerts originally for Financials, and then included in HR online and printed documentation because of their popularity.

The Finance module of the PeopleSoft ERP system implemented at CU includes Accounts Payable and Procurement. Beginning in January 1999, a PeopleSoft tutorial for the Finance module was made available on the ASP Web site. The training of over 1900 users of the finance system began in April 1999. An overview lecture, a two-day classroom curriculum, a desktop practice area, and training materials were developed for end users. Supplemental course sessions were developed and delivered for core department or heavy users of the system in Finance, Accounts Payable, and Procurement. The staff reacted favorably to the handson sessions (Dodge, 1999a) and had access to open labs provided for them to visit any time with issues or questions. Nevertheless, when the system went live, many users were traumatized. Wait times for response by the help desk was sometimes as long as 59 minutes (Dodge, 1999b).

Learning from the Finance module training a year earlier, Change Management incorporated improvements for Human Resources training. Change Management conducted a computer basic skills needs assessment and arranged workshops for staff that required basic computer skills prior to PeopleSoft Human Resources training (Orten, 2001). Workshops were also organized on managing imposed change, and a PeopleSoft tutorial was made available again on the ASP Web site. Human Resources classroom training began in February 2000. More than 1200 users were trained in a two-day class. Because approximately 80 percent of HR users also were using PeopleSoft Finance, most users were familiar with basic PeopleSoft security, log-on, panel structure, and PeopleSoft navigation. The classes used non-directed exercises to "make people think" (Orten, 2001).A desktop practice area was provided, and documentation was designed for business procedures and policies as well as online step-by-step guides for PeopleSoft transactions.

HR "go live" was delayed twice to reduce the risk of overburdening users who were still "suffering" from unresolved issues with PeopleSoft Financials (Dodge, 1999c, 2000). Because of the delays, a mandatory half-day refresher session was scheduled for employees in May and June 2000, with a review of key concepts, an overview of reporting, payroll timelines, and updated step-by-step guides. Extensive support and one-on-one assistance were provided for payroll- and HR-related activities in open lab sessions around each payroll deadline from July 2000 through December 2000. Change Management attempted to tailor the training as much as possible, given the resource constraints. In April 2001, training for new users was transitioned from the ASP training team to the PBS Center.

PERCEIVED USABILITY OF ERP TRAINING MANUALS FOR THE ASP PROJECT

In November 2002, a Web-based survey was designed to investigate the post-implementation usability of the ERP manuals. The Director of the Office of Change Management Services at CU collaborated on the survey, which had two parts. Part 1 of the survey asked users for information on their hardware and operating system in preparation for an ERP system upgrade to PeopleSoft 8.3. Part 2 of the survey included several usability questions, followed by two open-ended questions on what users "liked" and "disliked" about the manuals.

This article focuses on the responses to the two open-ended questions and reports on this part of the larger study. Of the 480 respondents to the survey, 289 of these responded to one or both open-ended questions; however, if negative comments were recorded under the "liked" section or positive comments were recorded under the "disliked" section, these responses were not included in the analysis. The specific survey questions and number of responses are provided in Table 1.

Users' Attitudes toward ERP Manuals

As mentioned above, we asked users on the survey (1) what they liked and (2) what they disliked about the training manuals. For most users, their responses were collected more than two years after they had attended the ERP training class just prior to implementation. Content analysis of user comments revealed many usability issues, which are discussed below.

A total of 135 usable positive comments were provided about the ERP training manuals. These positive comments were coded for the four dimensions of usability introduced above: navigation, presentation, learnability, and task support. Analysis of the positive comments showed that users commented on task support much more than on navigation, presentation, or learnability. As shown in Table 2, only 5 percent of the positive comments mentioned navigation, 13 percent presentation, and 10 percent learnability, while task support was mentioned in 88 percent of the comments.

Additionally, 157 usable negative comments on the ERP manuals were received. The negative comments were also coded for the four dimensions of usability, as shown in Table 2. Analysis of the negative comments showed that users again mentioned task support much more than navigation, presentation, or learnability.

In the next section, we explain in more detail users' likes and dislikes for each of the four usability characteristics: navigation, presentation, learnability, and task support.

Navigation in ERP Manuals

As shown in Table 2, although 5 percent of responses to the "like" question expressed satisfaction with navigation in the manuals, 25 percent of responses to the "dislike" question complained they could not find information easily in the manuals. They found it "challenging to locate which step-by-step guide is needed" when, for example, one is new to the CU system and does not understand what terms such as "Payroll Expense Transfer" mean. Other comments included:

* "It's not always easy to find what one needs to do because it may involve multiple entries."

* "Everything ran together and it was difficult to find a particular subject."

End users often did not want to read extensively to find what was relevant. Following are related comments:

* "I find myself tending to skim over a page in an attempt to find exactly what I'm looking for."

* "I sometimes have to look through an entire section to locate the exact piece of the process I need help with."

* [It is] "confusing to find correct information because of the many "alerts'."

* "I get tired of having to leaf through everything (which is quite thick) to find the one thing I'm looking for."

Other respondents reported problems finding answers to their questions:

* "The few times I've tried to use the manuals to answer questions, I could not find the answer."

* "Sometimes it is difficult to find the answer to a certain question and requires scanning multiple pages that have to do with the broader topic of a question."

Navigation would also be easier if the presentation of the manuals was improved, as discussed below.

Presentation in ERP Manuals

As shown in Table 2,13 percent of the responses to the "like" question and 20 percent of the responses to the "dislike" question mentioned the presentation of the manuals. Users suggested that the structure and organization of the manuals could have been improved with tabs and cross-reference indexing, as well as a better table of contents and index. Some of these comments follow:

* "I think there should be tabs for each process to help user easily find instructions."

* "Maybe a cross-reference index in case you don't know the exact name of the procedure you are looking for."

Users complained about having to search within the sections to get the information needed and suggested improved formatting within a section:

* "The table of contents was of limited use since I wanted to find specific processes quickly, not go to the section and hunt for what I needed."

* "The contents could be expanded a little more to describe what is actually in each section."

* "The format within a section can be improved as I find that I have to search within the sections to get the informatin (sic) that I need."

Often, the users who complained about difficulty finding what they need in the manuals suggested an improved index. For example:

* "There needs to be a better index so if we don't use the correct word to find the process, we can still look it up without any trouble."

* "Also it would be helpful to include an actual divider index in the binder rather than a bunch of yellow sheets of paper."

* "PBS manual would be better organized if it were more a reference manual rather than a training manual."

The following users suggested a FAQ (frequently asked questions) section for advanced problem solving:

* "Maybe it could use an index or FAQ section."

* "I wish we could have a FAQ section for a little advanced level, including the problems which wouldn't make it saved."

A FAQ section might also improve ERP learnability using the manuals, as discussed below.

Learnability of ERP Using Manuals

Ten percent of user responses to the "like" question reported that the manuals helped them learn the system (see Table 2). However, 47 percent complained about aspects of learnability such as unclear explanation (25 percent) and information overload (26 percent). A few users reported that they learned the system through trial and error rather than from the manuals. Some respondents complained about the learnability of the system, regardless of the manual:

* "The PeopleSoft system is not easy to learn, even for people who have no trouble learning other programs. There are too many ways that you can leave out information that is needed, creating problems later."

* "The [PeopleSoft] system is hard to learn and the layout of the manual is indicative of that. It's easy to forget steps."

Some users preferred learning with a person, rather than a manual:

* "I tend to learn more quickly when I've had hands-on training while looking over someone's shoulder or working with someone else rather than using a manual."

* "It's easier to ask than look in a book. Sometimes the book is just too cumbersome and what one needs is an abbreviated outline to recall how an entry is made."

One user pointed to the value in multiple methods of learning: "The online training, Step by Step Guides, Manuals, and the PBS & PSC Call Centers are all necessary to learn the PeopleSoft system." Another user reported the need for more thorough explanations in the finance manual: "The finance manuals do not teach how to read the reports, how to reconcile accounts, or what can and can't be done with certain funds and WHY"

Criticisms of technical jargon and acronyms were mentioned by three of the respondents. The following is an example:

* "It is frequently apocryphal, sometimes wildly inaccurate, and does not actually offer insight into processes on campus but rather clouds them by informing us of unexplained (and inexplicable) jargon."

Many users (25 percent of responses) complained about information overload and were overwhelmed by the size of the manuals. Comments included:

* "Too big and bulky."

* "I feel as if it is too much information."

* "One section that I found a little confusing and overwhelming was the report reading section."

* "Also would like to see the manuals less 'dense,' more concise directions."

* "Too cumbersome. Too much is assumed as far as understanding, probably due to the fact the people who put them together have forgotten how much they had to learn when they first came to the University."

* "I would prefer simplification for tasks that I am required to do rather than having to sort through and learn tons of things I never will do."

One user suggested, "budget and journal entries info could be in one binder and the purchasing in another." Another user recommended separating beginner from advanced content.

Some of the comments suggested that more tailoring of the documentation would help users learn the system and improve task support.

Task Support in ERP Manuals

Task support was mentioned in 88 percent of the "like" comments (see Table 2). Positive and negative perceptions of different aspects of task support are shown in Table 3. The aspects of task support that the responding users liked were (1) having a manual for reference, (2) the step-by-step guides, and (3) the screen-shot illustrations, as described in more detail below.

Users liked having the manuals as a reference; 45 percent of the "like" comments mentioned appreciation for having a hard copy to access, especially for procedures that were used infrequently. When users did not do a procedure on a regular basis, they tended to forget the details and the manual helped refresh their memories. A few users commented that the manuals were designed more for training rather than as an ongoing reference tool and therefore the content was in the order used in training. One of the users considered the step-by-step guides as reference tools.

The step-by-step guides were mentioned by 39 percent of the users who submitted "like" comments. They found the guides easier to follow than descriptions of the process and were glad to have help without bothering anyone. Many users commented that the online step-by-step guides were more current than the printed manual that they received at the training session.

The screen prints were mentioned by 13 percent of the users who responded to the question on what they liked about the manuals. The screen prints showed users where they were supposed to be in the software and clarified the instruction step. Screen prints are categorized as task support rather than presentation because their content is much more important to users than their appearance.

As shown in Table 2, task support was mentioned in 71 percent of the "dislike" comments. Users disliked aspects of the manuals relating to (1) difficulty understanding processes (explanations), (2) incomplete information, and (3) problems keeping the documentation updated, as summarized in Table 3 and described in more detail below.

Mention was made of problems in understanding processes in 46 percent of the responses. They complained that the manuals did not "explain why something is done a certain way"; did not give the implications of selecting various answers; did not help with troubleshooting; did not explain the reason for errors in the system; did not explain who does what; did not explain CU-specific processes; and assumed that users are already' familiar with the CU processes. Some examples of user comments follow:

* "The process on each campus outside of PeopleSoft is important and wasn't given enough attention in the desk manual."

* "Most of the training seemed to be 'click your mouse here' and not WHY this needs to be done and this is what you do if it doesn't work."

* [Suggested improvements] "explain what 'dispatching a PO' means and better explain how to receive in a PO."

Incomplete information and documentation of processes was mentioned by 33 percent of the responses. Many complained that the offline processes were left out and that the manuals did not indicate "what other processes need to occur in conjunction -with the PeopleSoft processes." Others complained that the manuals leave out special circumstances; leave out quirks in the system; leave out little details that make "all the difference"; have no flowcharts or descriptions of paper flow; and do not provide data dictionary information so users are "constantly guessing at field content and interpretation." For example, one user commented:

* "If you find it and follow the instructions, there is invariably a step that is left out or assumed in the instructions, etc."

Problems keeping the manuals current were mentioned by 22 percent of the respondents. They complained that they are not always notified when there are updates; they are unsure if the manuals are current; the manuals were obsolete "the day we went live"; too many "alerts" were published after training; information in the manuals is different from that online ("mine is so out of date that nothing is relevant to today's work"); and because the manuals were not updated, they had "incorrect information related to panels and data."

Perceived Usefulness of the ERP Training Manuals

The four perceived usability characteristics discussed above are likely to affect the perceived usefulness, use, and user acceptance of the EElP training manuals. However, 65 percent of respondents who provided "like" comments and 57 percent of those -who provided "dislike" comments answered better than neutral (on the seven-point Likert scale) to the following survey questions:

* "The manual provides useful information."

* "I would recommend that other people in my group use the manual as a resource."

As suggested in some of the responses above, however, 42 percent of respondents to the "dislike" question indicated the usefulness of the manuals could be improved. The manuals would be more useful to users if the offline processes were documented in an integrated way with the online procedures in the software. Furthermore, troubleshooting content needed to be provided. The following examples suggested other problems that affected the usefulness of the manuals:

* "I also would find it more useful if it indicated what other processes need to occur in conjunction with PS processes. I often find I've done something only halfway because I followed only the PS manual."

* "Sometimes, steps seem to be missing; and when you have problems, the manual is rarely useful."

About 10 percent of the "dislike" respondents reported that they did not use the manuals, either because they are obsolete or not helpful for troubleshooting.

SUMMARY AND LESSONS LEARNED

The relatively small ERP training budget for this implementation hampered efforts to prepare more than 2000 ERP users at the University of Colorado for their new roles. With more resources, CU users may have had more role-based training like at Motorola, which had learned from a prior ERP implementation failure (Roberts et al., 2003), and training materials would have been more customized to user roles. Under this scenario, users would not have been intimidated by the bulk of manuals containing irrelevant material, but would have been able to find the information they needed quickly. Two years after training, usability problems with the manuals still plagued staff that used the ERP irregularly and relied on this documentation. On the other hand, many users appreciated having a printed manual to help them do their jobs. The complexity of the software, the radical change in tasks, and the large numbers of users affected meant that having no documentation was not an option.

The positive comments showed that users considered task support much more important than presentation, navigation, and learnability. Among the aspects of task support, the most significant were availability of a reference, step-by-step guides to carrying out the task, and, to a lesser extent, illustrations of screens from the ERP software. However, problems with presentation, navigation, and learnability were also obstacles to task support.

The negative comments also emphasized task support - specifically, not enough explanation, and incomplete and out-of-date information. Learnability and navigation received the next most complaints and presentation the least. The overwhelming size of the manual and information overload was a common complaint that implied obstacles to learnability.

Our findings have implications for improving the design of ERP documentation, which would decrease the pain and cost of ERP implementations. Although most ERP training documentation has moved online at CU and elsewhere, there are lessons to be learned from this study:

* Documentation should focus on organization-specific business processes. Providing a hyperlink from online documentation to process flow diagrams would help users understand who does what and the rationale behind the ERP transaction. This would also address the complaints about incompleteness of information in the manuals.

* Role-based training (see Roberts et al., 2003) would provide knowledge integration and better mapping to users' needs. More tailoring of documentation would increase perceived usability, especially learnability and task support.

* FAQs (frequently asked questions) available online would explain advanced tasks and provide troubleshooting advice.

* Out-of-date information should be avoided by printing only the more static content in manuals and posting the dynamic content online where it is more easily updated. A learning management system (see Roberts et al., 2003) would help organize training-related content.

* The most recent version of the manuals could be available for download. The download could be by chapter or the entire manual.

* The online version of the manuals should include advanced search capabilities so users can easily find the help they need.

* Popular online step-by-step guides could be made more usable with a search engine.

* An obvious and relatively easy change would be to restrict the size of the manuals. Much like desktop software documentation, there could be a "getting started" manual, a reference manual, and an advanced manual. This solution would give users more control over information overload.

* Using hyperlinks in online documentation is another potential solution to information overload.

* Providing a data dictionary online would help users who want an explanation of field content and interpretation.

* Use of color-coding, tabs, more section breaks, and a good layout for the table of contents and index would improve the presentation of the printed documentation.

* Soliciting feedback on the manuals - in the design phase, at the end of classroom training sessions, and at regular intervals - would provide a means of monitoring user attitudes.

CONCLUSION

In conclusion, this research study identified users' perceived usability of ERP training documentation for the University of Colorado's PeopleSoft implementation. Because user-training issues are common in ERP implementations, it is likely that the results of this study are generalizable to other organizations and other industries. The results are also applicable to training in other categories of complex software that have large numbers of enterprise-wide users whose jobs have changed radically.

The survey responses confirm a need for process-centric documentation, as mentioned in the literature. This is the most important finding for both research and practice. While it may seem obvious that users are most concerned with task support, the fact that many users were not satisfied with the manuals' facilitation of their work activities more than two years after training suggests major shortcomings. Analyzing the users' likes and dislikes yielded some clear guidelines on ways to enhance the usability of printed training manuals and online documentation.

Documentation is a relatively small part of a typical ERP project budget, yet the impact on learning and troubleshooting the ERP system is potentially high. If the usability of the documentation is poor, users waste more time trying to correct errors, and also are more likely to call the help desk or ask colleagues for help. These forms of support may not be immediately available, especially at go-live. However, this study clearly shows that organizations should not neglect the usability of their documentation after go-live, because it impacts the future returns on their investment in the ERP system.