Abstract

Early experiences with Web-groupware point to new collaboration opportunities within and between organizations. We report the results of a study of more than 100 organizations that have used Web-groupware to better understand how they are using it and what advantages and disadvantages they have experienced. We then use this data to develop a framework for analyzing and assessing the fit of groupware systems to organizational needs. We close with a discussion of the evolution and future of groupware.

Introduction

The Internet[1] and the World Wide Web hold many possibilities for new forms of communication. The Internet and the Web enable individuals to collaborate via email, participate in newsgroup and listserv discussions, and to publish information electronically. The Web also enables innovative applications of groupware (e.g., Dennis, Quek, and Pootheri, 1996; Fellers, Clifton, and Handley, 1995). Full feature groupware with tools for voting and information structuring, however, has historically been available only on private internal corporate networks which were often LAN-based and required expensive proprietary software.

The Web is changing this. There are now dozens of Web-based systems available that bring the power of groupware to any desktop with a Web browser (Woolley, 1999). The question, of course, is what advantages and disadvantages do Web-groupware systems have compared to proprietary groupware and to other commonly available Internet-based systems (e.g., listservs, newsgroups)?

In this paper, we report on a study of more than 100 organizations that are early adopters of Web-groupware. We first report on their uses and perceived advantages and disadvantages for Web-groupware. We then use these results to build a framework for assessing groupware capabilities and discuss the trends we see in the evolution of groupware.

Groupware and the Internet

Research on computer support of cooperative work dates back to J. C. R. Licklider, Douglas Englebart, and other researchers of the 1960s (Greif, 1988; Licklider, and Taylor, 1968). The application of these concepts began in earnest during the 1970s with the development of several systems for threaded discussions (Hiltz and Turoff, 1978). At first, these tools were used primarily in research and education, but with the spread of LANs in the 1980s, commercial application accelerated, and the term "groupware" came to describe software intended to enhance group productivity. Commercial software was developed (e.g., GroupSystems (Nunamaker, Dennis, Valacich, Vogel, and George, 1991) and Lotus Notes), and email, listserv, and USENET News thrived on the ARPANet, CSNET, USENET, BITNET, and ultimately the Internet.

Table 1 shows examples of groupware applications that are prominent on the Internet and LANs today and how they are commonly used. Note that a given technology may used in multiple ways (e.g., one can use email for same time, different place communication) and that a given system may integrate several of these applications.

Table 1. Examples of common groupware applications

Same Time, Same Place

Brainstorming

Voting

Outlining

Writing

Same Time, Different Place

Text-based chatting

Video conferencing

Audio conferencing

Whiteboard sharing

Application sharing

Different Time

Sending email

Posting to listservs

Posting to threaded-discussions

Organizing workflow

We believe that the rapid growth of IP-based networking and protocols like HTTP will have a profound effect on groupware. The availability of these and other standards will enable low-cost, interoperable implementation on intranets, the Internet, and extranets. Indeed, many proprietary groupware packages are currently being ported to IP and the Web, and new programs are being developed to take advantage of the rapid spread of the Web. There are more than 75 Web-groupware programs available today; (see Woolley, 1999 for an updated list).

At present, we know very little about the features that users perceive to be important, and the advantages and disadvantages that Web-groupware systems provide relative to other collaboration tools. Since Web-groupware is a new research area, we felt that a grounded theory (Strauss, and Corbin, 1990) approach, in which the researchers attempt to induce a set of concepts and relationships among them by studying a phenomenon, was more appropriate than the traditional formulation and testing of hypotheses. We began with two over-arching research questions:

RQ1: For what applications are Web-groupware tools being used?

RQ2: What are the perceived advantages and disadvantages of Web-groupware tools compared to their non-Web-based counterparts?

We conducted interviews and a survey as a first step toward understanding the role of Web-groupware. These are described in the next section followed by the results of our analyses of the uses and advantages and disadvantages encountered. However, our analyses of the data also took us in an additional direction, which is not uncommon in qualitative field research (e.g., see Sutton, and Hargadon, 1996). It became apparent that many organizations had considerably different viewpoints as to what Web-groupware was and what features they expected it to provide -- features that were not always available in the software. Therefore, based on the survey data, we developed a third research question:

RQ3: What are important characteristics for Web-groupware tools?

Method

We decided the best way to answer these questions was to use a combined interview and survey approach. The interviews would serve as a pilot tests, providing more in-depth understanding of how Web-groupware was use. The survey would provide more breadth and would better enable us to understand the opinions of many users.

We began by conducting in-depth, semi-structured interviews over the telephone or via email with one to three members of four organizations we had assisted in deploying Web-groupware. The interviews began with a set of structured questions, but flowed differently for each interviewee. The responses to these questions were pursued and took each interview in a different direction. The questions focused on three general areas: how the organization used Web groupware, its perceived advantages and disadvantages, and what the functions of the system were used.

Survey Sample

We used the same open-ended, qualitative questions on the survey. Once again, our selection of organizations was purposeful, rather than random (see Strauss, and Corbin, (1990). We maintained a Web site with information about groupware (Dennis, 1998). We sent the questionnaire to everyone who had emailed us seeking information about Web groupware. Surveys were emailed to 335 potential respondents, with a second "mailing" sent out one month later to any who had not responded. We received responses from 108 organizations (a response rate of 32%); 102 from the original "mailing" and six from the second mailing.

Table 2 shows basic information about the types of organization and geographic locations of the survey respondents. We received at least one response from every continent except Antarctica, but most responses were from the United States (about 60%). Likewise, we received responses from large and small corporations, government agencies, and not-for-profit organizations, but most responses came from universities (about 50%).

Table 2 Type and location of organizations responding to survey

	U.S.	Canada & Mexico	South America	Europe	Africa	Australia & New Zealand	Asia	Total
University	28	6	2	6		7	3	52
Small Corporation	19	2	1	5	1			28
Fortune 500 Corporation	8			1				9
Government or Military	7			1		5		13
Not-for-Profit Organization	6							6
Total	68	8	3	13	1	12	3	108

Primary Analysis

In analyzing the results, we followed the general approach of grounded theory (Strauss, and Corbin, 1990). The analysis process had four steps. First, we began by drawing on previous theory and research, popular press and promotional articles, and our own experiences as developers and users of groupware to develop theoretical sensitivity to the issues under study and to develop a set of candidate seed concepts for the analysis. In this step, we developed four lists of concepts: 1. applications of Web-groupware (RQ1); 2. advantages of Web-groupware (RQ2); 3. disadvantages of Web-groupware (RQ2); and 4. other issues (which evolved into RQ4). We began with two to three candidate seed concepts in each of the first three lists (e.g., "discussion," "voting," "open standards," "costs").

Second, we used open coding to identify candidate concepts from the data and to consider whether the candidate seed concepts developed prior to the open coding had merit. We parsed every response and matched each substantive point or response fragment to a concept in one of the four lists, adding concepts to the lists as we went along. This coding process resulted in a set of initial concepts, organized into the four lists.

Third, we then analyzed the data a second time, attempting to refine the concepts and combine related concepts. We ensured that every substantive point in the responses matched one or more concepts in the four lists and that every concept in the four lists was matched to at least one substantive point in the data. Occasionally concepts were added or deleted these lists. This process yielded a refined set of concepts matched to specific parts of the survey responses. It also produced the number of organizations that mentioned each concept.

Fourth, we discussed the refined set of concepts with six respondents and with six leading academics, consultants, and developers to further refine the concepts. This resulted in the final set of concepts designed to answer the first two research questions.

Secondary Analysis

In re-examining the concepts on all four lists, it became apparent to us that many of the concepts were characteristics of groupware, some present characteristics (e.g., anonymity), some desired characteristics (e.g., multi-media), and some the opposite of the desired characteristics (e.g., high operating costs). We then decided to develop a new conceptual framework for characterizing groupware and examine the evolution of groupware using it. We followed the same four step process in examining the groupware characteristics. We first developed initial seed concepts, analyzed the responses, refined the concepts into a framework of characteristics, and discussed the framework with respondents and experts.

Results

Applications

Figure 1 summarizes the ways in which Web-groupware was used. Support for project teams was the most common application. Most of these teams (whether geographically distributed or co-located in the same office complex) used groupware asynchronously, but six percent used it at the same time and place in networked decision-support rooms. Most teams used the groupware to plan for face-to-face, non-groupware meetings or to initiate discussions to be completed in them.

A second major category of applications was educational use, where it was primarily used to supplement or replace face-to-face teaching or training (either co-located or distributed), or used to demonstrate groupware. The third major application was to replace e-mail listservs in support of special interest group discussions.

Advantages and Disadvantages

Figure 2 shows respondents’ reported advantages and disadvantages of Web-groupware. The most commonly mentioned advantages of Web-groupware were technological: open network standards to enable any-place-any-time interaction and open client standards (i.e., a commonly available, platform-independent Web browser) so that the user did not need to install any new software (i.e., a proprietary client). The use of a browser also meant individuals were familiar with the software interface and required minimal training. Most software was free or relatively low cost, so setup costs were low. A second set of advantages focused mostly on the specific functions available to support interaction, such as the type of transformations possible (e.g., the ability to structure, sort, and analyze discussions), and the specific functions of the software (e.g., anonymity, embedding HTML and graphics).

The disadvantages can be grouped into three groups. The first centered on the lack of features of current tools (e.g., narrow range of capabilities, lack of integration, cumbersome user interfaces). These were especially evident to those who had used the richer interfaces of proprietary groupware clients that support drag-and-drop user actions. Once groups began to use Web groupware, they realized that they could not support the intended application as well as first believed. The second group of disadvantages dealt with network technologies (slow and unreliable networks, security, access problems).

The final group of disadvantages are issues that often accompany IT-enabled organizational change. Concerns regarding operating costs (systems administration, training, and facilitation) and group learning (changing team skills and norms to adapt to the new possibilities offered by the software) can inhibit the use of Web-groupware. While the use of an existing browser reduces training on using the software client, it does not reduce the need to train teams to adapt their work processes to new groupware-enabled processes. We believe this is a fundamental issue that will be faced by every organization adopting Web-groupware (Orlikowski, 1992).

In summary, the survey respondents perceived value from the open technological standards of using IP networks and widely available browsers for groupware applications. The severe limitations of using HTTP and relatively lean interfaces led to some disappointments with the capabilities of Web-groupware to meet the collaboration needs for teams and education. Finally, improving the technical availability of groupware did not improve its effective use in teams. Training for computer-supported collaboration was still needed to help users and teams adapt their work habits to leverage Web-groupware. In the next section, we identify three stages of technical development for Web-groupware that offer the potential to address some of the capabilities issues and propose a research agenda. We address the need for additional knowledge sharing among groupware developers – beyond technology – to improve the value of Web-groupware.

Characteristics of Groupware

The survey data suggest that the main reason organizations adopt Web-groupware is because it is technically simple to install (open network standards, open client standards, minimal individual learning). However, the major problem that organizations encounter is that the software they install does not do what their users want (a lack of features). This suggests that a more detailed analysis approach is needed to better understand the fit between the many characteristics of groupware and an organization’s needs.

Figure 3 presents the groupware framework that emerged from our survey and interview data. We believe it can be used to analyze both organizational needs and the capabilities of specific groupware systems. The framework is presented as a radar chart with the items organized into four quadrants. Items near the bottom of the chart are related to the characteristics of the technology itself, while items nearer the top are related to technology use in organizations. Items on the left side describe the capabilities of the software while the items on the right describe the enabling conditions needed to deploy and use it effectively. The next sections explain each quadrant and the items it contains.

Figure 3 Groupware Framework

IV. Organizational

Requirements

Collaborative Capabilities. Quadrant I represents Collaborative Capabilities. It addresses the question of what will groupware do by describing the fundamental activities for which groupware is used. Communication can be understood in terms of two activities that appeared regularly in our data: conveyance and convergence. Conveyance is the exchange of information among participants with interpretation, understanding, and use of the information left to the individual. For example, organizations in our survey used groupware for the communication of marketing intelligence, project schedules, or ideas for quality improvements. Convergence is the development of shared meaning among participants. Organizations in our survey used groupware to converge on the selection of projects to be funded or to develop outlines for project reports. One additional fundamental activity that emerged from the data was organizational memory, an organized “place” for automatically storing and retrieving participant interactions over a long period of time (Press, 1992). Organizations in this study used groupware to maintain documentation of bugs and bug fixes among distributed and co-located software development teams and for questions and responses at help desks.

While many traditional Internet tools (e.g., listserv, threaded discussion) and current Web-groupware tools proved useful in supporting conveyance, several respondents cited problems in supporting convergence and organizational memory needs. Threaded discussions, for example, offer few practical ways to organize, structure, and focus information to reach consensus, or to filter and search information in an organizational memory.

Technical Capabilities. Quadrant II describes the Technical Capabilities of groupware by asking how is collaboration done? Numerous respondents mentioned many specific capabilities provided by groupware systems. We grouped these into five fundamental dimensions. The first is a set of presentation features for manipulating the format of contributions to help impose meaning. These include sorting or filtering comments (e.g., by date, contributor, discussion thread), searching (e.g., by key words, full-text), and expanding or contracting discussion branches.

The second is a set of transforming functions that add meaning to data by changing it from one form into another form. For example, analysis tools such as voting (e.g., ranking, rating) transform individual expressions of opinion into a collective result. Structuring or modeling tools (e.g., conceptual mapping) transform lists of ideas into graphical images by depicting relationships among them.

The third capability, interaction modes, refers to the ability of participants to interact in many different ways. For example, some groups want anonymity for some discussions but to have contributions identified for others. Similarly, some discussions benefit from an ability to impose restrictiveness and guidance (Wheeler, and Valacich, 1996) to limit how participants can interact, while others need a more free-form approach. For example, the ability to introduce a top-level discussion topic, delete comments, or invite new users into a discussion may be limited to only the leader. Interaction modes also includes the granularity with which this control can be shared with participants.

The final two capabilities, programmability and multimedia data types, refer to specific technology characteristics. Some groupware applications support a scripting language to automate routine tasks, to develop more sophisticated applications, or to provide agent capabilities. For example, users may configure agents to monitor various discussions and alert them via email when a certain topic is mentioned. The support for multimedia data types beyond simple text (e.g., graphics, audio, live video) can also be important.

Technology Conditions. Quadrant III describes the Technology Conditions to enable the groupware by asking what technology is required? Web-groupware, like the Web itself, uses a client-server architecture. The dominant clients in use today, Netscape’s Navigator and Microsoft’s Internet Explorer, support existing HTML standards. Each, however, has added proprietary extensions to HTML creating de facto standards that, in the short run, are not supported by other browsers. These extensions enable new features at the expense of cross-vendor compatibility. Similarly, it is possible to develop a proprietary client to enable a particular set of features (e.g., Lotus Notes Client).

Use of open client standards with widely available browsers increases the reach of Web-groupware. Use of proprietary clients or extensions can limit cross-vendor or cross-platform (e.g. Windows, UNIX, Mac) support thus reducing reach. For several organizations in our study, the requirement to install proprietary client software on each user’s desktop eliminated certain groupware systems from consideration. Any benefits from these systems were quickly outweighed by the added expense and system administration problems caused by proprietary clients. Customization enabled by Java and Active-X applets will likely blur capability distinctions between open and proprietary clients.

Similarly, groupware that requires anything other than open networking standards (e.g., TCP/IP) will also have a much more limited reach. Proprietary networking requirements (e.g., Novell Netware’s IPX/SPX) may work fine for uses within an organization and offer increased network speed, but they increase the complexity of interorganizational collaboration. For the organizations in our study, open networking standards was important.

The level of real or perceived security of groupware technology can also influence the ways in which it is used. Use of key-based encryption algorithms and certificates for authenticating participants can be essential technologies for some applications needing reasonably secure communication via groupware.

Finally, issues of bandwidth requirements and asynchronous connectivity were important to some organizations. The network requirements of some groupware systems assume a high capacity intranet and would suffer performance problems in low bandwidth environments. Similarly, the requirement of a continuous connection to a network can reduce groupware’s reach since this may not be possible for mobile participants. Replication of groupware databases (the process of keeping multiple copies synchronized over time and distance) onto local PCs, notebook computers, or servers can allow participants to work off-line and then reconnect to upload their contributions.

Organizational Conditions. Quadrant IV addresses the important Organizational Conditions with the question of what conditions are necessary to use groupware? Low startup costs can be important for getting unproven groupware in the door. Startup costs include the direct costs of hardware and software purchases as well as the indirect costs of installation, training, and setup. Web-groupware that uses existing browsers for clients can offer very low startup costs. This was a major factor in enabling several organizations in our study to consider groupware.

Once groupware is technically operational, team members must learn how to effectively use the software to collaborate via electronic media where communication differs from synchronous face-to-face meetings. Organizations and individuals within teams differ in their learning capacity, or the speed with which they can assimilate procedural and technology changes. There are two learning challenges to be overcome. The first, new software learning requirement, is the ability of individuals to quickly learn the operation and use of new software. Many organizations commented that the use of a familiar Web browser greatly simplified training issues.

The second learning challenge, process change requirement, is the necessity to adapt an organization’s work processes, values, and rewards to leverage the capabilities of new technologies (Orlikowski, 1992). This focuses on understanding how to derive value from the use of groupware, rather than understanding which buttons to push. Teams must learn how to design effective groupware “meetings” and how to use groupware to contribute in meaningful ways to their objectives. Some teams can quickly assimilate these changes while others require more time and investments in formal training. Several organizations in our study commented that the greatest challenge in adopting Web-groupware was in changing behavior, not in teaching what browser buttons to click.

Organizations must also consider the ongoing operating costs of using groupware. The use of nonproprietary clients and servers can reduce operating costs because these add no additional groupware-specific costs beyond the groupware-server software itself. However, we suspect that the majority of costs will arise not from the hardware and software but from the management costs of training users and creating, structuring, facilitating, and archiving group projects.

Discussion

This study found that Web-groupware is most commonly used to support project teams that work in different places at different times. However, about half of those project teams are co-located in the same office complex; Web-groupware simply makes it easier for them to share information without having to meet at the same place and time. Other common applications include the replacement of traditional e-mail listservs and in education.

The major advantages of Web-groupware compared to traditional propriety groupware include open network standards, open client standards, minimal individual learning, and minimal setup costs. The major advantages compared to other Internet-based tools such as e-mail and newsgroups included transformations (e.g., the ability to structure, sort, and analyze discussions), and specific functions such as anonymity and embedded graphics. The major disadvantages compared to propriety groupware included a lack of features and network speed and security. The major disadvantages compared to other Internet-based tools included operating costs, and the difficulty in changing team skills and norms to adapt to the new software.

The data also suggested that respondents held different views of Web-groupware and its important characteristics. From these responses, we developed a four-part groupware framework that included collaborative capabilities, technical capabilities, technology conditions (i.e., architecture) and organizational conditions.

Implications

In our opinion, these results have several important implications for practice and future groupware research and development. First, while many Web-based systems are being used to support work in different places and different times, many projects are co-located, with team members working in close geographic proximity. So while we may view Web-based tools as ways to reduce barriers among geographically dispersed teams, they can also play an important role in teams that can work together at the same place and time, by expanding the work that can be done before and after face-to-face meetings. The implication for researchers is to move beyond the study of just face-to-face teams or just distributed teams to include research of teams that work both face-to-face and distributed. The implication for developers is to consider how the results from the Web-based discussions can be seamlessly moved into face-to-face meetings and vice-versa: how to extend -- not replace -- face-to-face meetings via the Web.

Second, the pattern of advantages and disadvantages presents an important picture for current practice and future groupware development. The strongest driving forces propelling the investigation and adoption of Web-groupware are the open network standards and open client standards. However, the biggest disadvantages are the lack of needed features and the poor speed and reliability of the network that requires a pause after each button is pressed. In other words, Web-groupware draws organizations from proprietary solutions because of its open standards, but once they begin to use it, they are underwhelmed. The implication is that groupware researchers and developers need to understand what elements are missing from Web-groupware and work to develop new systems that address these problems.

This is where the groupware framework can play an important role. One of the issues in qualitative research is verifying or testing the principal findings (Strauss, and Corbin, 1990). The principal goal of our framework is to provide a visually compact way to assess an organization’s needs relative to the capabilities of specific groupware systems. Each of the 17 items in the framework can be assessed using a five point Likert scale and can be plotted to build a multi-dimensional profile on the radar chart. This can help organizations understand their requirements and understand the capabilities and conditions associated with different groupware tools. By comparing the shapes of the organizational requirement profile and a specific groupware capability profile, one can see the areas of good and poor fit.

However, this framework may also be useful at a more macro level of analysis if we consider the evolution of groupware over the past two decades. The first generation of commercial groupware was built for private networks using third generation programming languages such as C, Pascal, and BASIC. We will refer to this as proprietary client groupware (see Table 3). GroupSystems, VisionQuest, and Lotus Notes were examples of these systems. They offered rich user interfaces with many functions (e.g., casting votes, editing postings, or closing windows) being custom-defined for each tool. Their focus was on Quadrants I and II in our framework, providing the best set of collaborative capabilities and technical capabilities (although different tools emphasized different aspects). However, each user had to install proprietary software on all clients or go to a special room where this software was available. These high costs contributed to a limited diffusion of these tools in organizations – especially for meeting room types of groupware that required dedicated facilities for face-to-face group work.

Table 3 Three Generations of Groupware

Groupware Generation

Framework

Focus^*

Server

Communication Protocols

Client

1^st Generation:

Proprietary Client

Proprietary software

Proprietary (e.g., DOS, IPX)

Proprietary software, pre-installed