CHI '95 ProceedingsTopIndexes
TutorialsTOC

Designing for Complex Products

Jared M. Spool Carolyn Snyder

User Interface Engineering 800 Turnpike Street, Suite 101 North Andover, MA 01854 (508) 975-4343 spool.chi@xerox.com

© ACM

Abstract

Increasingly, software interfaces bear the burden for com- municating new and complex concepts to users. To do this, the designer can employ create wizards, cue cards, drag-and- drop, or other similar techniques. Interfaces contain an ever- increasing amount of functionality. But, in order for these methods to be effective, they must be properly designed. How does the designer determine which approach is called for? How can the designer determine what the user needs? How can the designer create these tools so that they are effective?

This tutorial examines the latest interface tools and the techniques for developing them for complex products. Keywords Prototyping, low-fidelity prototyping, process management, product development, contextual inquiry, modeling, practi- cal techniques, wizards, cue cards, guided learning, com- plexity management.

DESIGN CHALLENGES IN COMPLEX PRODUCTS

An interface is far more than a collection of widgets. An interface communicates its content - the product's func- tionality - via the multiple channels of the screen, help, tutorials, etc. An interface that uses these channels wisely will help users get work done, even as they are learning new concepts.

Successful interface design is like magic. [2] The designer must first decide what mental model the user should have, and then consciously create an "illusion" with the product interface that communicates this model to the user.

ADVANCED INTERACTION TOOLS

Designers of sophisticated products have an arsenal of state- of-the-art interface tools at their disposal. "Wizards," "Cue Cards," interactive tutorials, drag and drop, and intelligent defaulting can help users work more easily and productively. Unfortunately, we have observed that a poor implementation of these tools can make the situation worse than having no tool at all. Obviously, these tools all have important design considerations.

"Wizards" and "Experts"

They go by many names - "wizards" in Microsoft prod- ucts, "experts" in Borland, "advisors" in Meca - but the concept is the same: an on-line tool to walk the user through the steps of a complex process.

Many developers are considering wizard-like enhancements to their product. Unfortunately, some wizards fall short of their promise. We have seen instances of wizards that frus- trate and confuse users. Obviously, there are important de- sign issues for wizards. [1] For example, if the wizard pro- vides only a subset of the available functionality (this is true for Microsoft Excel's Chart Wizard), how does the wizard point users toward the rest of the functionality? Also, as the user becomes more familiar with the system, will he or she still want and/or need to use the wizard?

"Cue Cards" and Interactive Tutorials

These methods help new users learn product concepts. Implementations that use the users' data allow them to get work done while they are learning, but make the designer's job more challenging. Users often stray from the directions provided on the screen (either by mistake or out of curiosity) so the implementation needs to be "bulletproof" and forgiving. Other design issues include finding out exactly which information new users need, and giving users a choice as to whether they want to learn the information.

Drag and Drop

Drag and drop can make users more productive by elimi- nating the need for some keyboard and/or menu commands. One problem with drag and drop is that this technique in- herently lacks affordances - users may be completely un- aware that this method is available. If an interface relies on drag and drop for certain functions, the designer must ensure that users have a way to learn of this method.

Advanced Data Entry

There are a number of tricks that designers can utilize to facilitate data entry tasks. For complex formulas, the inter- face can provide assistance with syntax. Repetitive entries can be shortened to fewer keystrokes. With intelligent de- faulting, the system "looks" at previously entered data to determine the most likely entry for subsequent fields.

DESIGNING ADVANCED INTERACTION TOOLS

Often, the most difficult part of product design is under- standing what the users need. Contextual inquiry allows designers to gain insights into the users' job and work envi- ronment, and the information thus obtained can substantially influence the design of the product.

For example, a designer who wants to lower the products' learning barrier can search users' work environments for diagrams, handwritten notes, and other "cheat sheets" that users rely on to get their job done. The designer can then incorporate this information directly into the wizard. Before creating the tool. designers should establish its goals. For example, is the proposed wizard intended as a learning aid for new users, or a productivity aid for experienced users? With the goals clearly in mind, the designer can then focus on design issues such as clarity and affordances. Paper prototyping is a useful method for evaluating the success of a particular approach, and for obtaining feedback on how to improve it. For example, when developing a wizard to assist new users with a complex process, designers can determine whether these users actually understand the steps and terminology in the wizard. Since this is a make-it- or-break-it issue for this type of wizard, designers can nail down a workable approach before investing the effort in writing code.

MODELING

Developers often develop "myths" about their users and how they use an interface. For example, "the person who installs this product is the same one who uses it." Since these myths form the basis for many of the developers' assumptions, they can substantially affect design.

Myths often exist at a subconscious level - asking a developer to "tell me your myths about users" is unlikely to elicit any useful information. However, there are ways to unearth myths and use them as the basis for a model of the user. The designers continually refine this model as they acquire more knowledge of users.

PROVIDING ACCESS TO FUNCTIONALITY

Toolbars

As users gain expertise, they can benefit from shortcuts. Toolbars seem to keep growing. Unfortunately, having more icons on the toolbar does not necessarily make users more productive - there seems to be a point of diminishing returns. Surprisingly, many users don't customize their toolbars when this option is available, so the designer still bears the burden for determining which functions should be on the toolbar. Icon recognition is also an issue. Putting words on toolbars and/or having a readily available way to determine the purpose of an icon aids users.

Live Status Bars

Similar in concept to toolbars, live status bars serve a dual purpose: to show the user the current status of something (the selected object, the current view, etc.) and to let the user change this status.

Tabbed Dialog Boxes

Tabbed dialog boxes allow multiple functions to be collated under a single menu access. They allow the user to make a variety of disparate changes in quick succession, but add their own unique design issues: what happens if the user wishes to cancel only one of several changes made?

Right Mouse Button

The right mouse button is a handy way to put commonly- used functions literally at the users' fingertips. However, the right mouse button shares a problem with drag and drop - users may be completely unaware of this functionality. When users select with the left mouse button before accessing functions with the right, we have observed that they often move the move between the clicks. This can result in their performing the function on the wrong object.

DESIGNING ACCESS TO FUNCTIONALITY

Contextual inquiry can yield a wealth of information about the task frequency, users' terminology, usage of keyboard and mouse, steps of the task, and any patterns in the input. The users' background is also relevant - for example, people who are primarily Macintosh users are unaccustomed to multiple mouse buttons.

It is often a good idea to provide redundant access to func- tionality, but not every function needs to be available via each possible method.

In usability testing, designers should be alert for missing affordances, excessive "tool time," and should take note of where users actually look.

References

1. Snyder, C., Spool J. Building Better Wizards, User Interface Engineering, In Press
2. Tognazzini, B. Principles, Techniques, and Ethics of Stage Magic and Their Application to Human Inter- face Design, INTERCHI '93 Human Factors in Computing Systems (Amsterdam, April 24-29), ACM Press, p.p. 355-362) CHI 1993 Proceedings.