Product Usability

Most of us in product development agree that usabilityis important, yet we still encounter many products in our daily lives thatcause us to curse the unseen designer, crying, "The person who designedthis @!#*&! thing should be forced to use it!"

I will present some of the concepts of usability and tryto persuade you how important it is. As examples, I will use audio productswith a predominantly hardware user interface, including the Orban DSE-7000.

Background

Usability is only one dimension of new product development,and unfortunately it is often given short shrift. Professionals skilledin product usability are generally found in larger companies, so that usabilitydesign often falls to engineering and marketing. Unaware of usability engineering'sunderlying principles and lacking the perspective that a day-to-day workingpractice in usability would give, designers frequently create productsthat are difficult to learn and awkward to use. This leads to dissatisfiedcustomers with negative attitudes toward the manufacturer, operationalerrors, and a surface level use of the product that leaves hidden featuresforever unused. In some industries, poor usability leads to dangerous productsthat injure people. The usability literature contains sobering examplesof aircraft crashes, nuclear power plant accidents, and tragic medicalerrors that were traced to poor usability design.

On a more positive note, companies are striving to createcompetitive new products, products that offer advantages in performanceand features. What could be more competitive than products that delightthe user by their ease of learning and use? Give your customers equipmentwhich reveals its operation via its user interface, and they will be anenthusiastic virtual sales force.

In 1987, while I was the Executive Vice President of AKG'sDigital Product Division, I conceived a digital audio workstation forradio stations and began directing a small development team to design it.As I write this, in 1997, the DSE-7000 is not only in the marketplace still,but is the leading product in its class, widely loved by its users, asthis testimonial from John Chickering attests: "The other distinctadvantage of the DSE is the well thought-out hardware interface. I hadbeen working with analog production for years, and I discovered that Iwas doing more and better work on the DSE in a couple of hours. The largework surface and sturdy controls were a refreshing change from Tinkertoysystems where you squint at a screen while pushing virtual faders witha mouse. No thanks."

Here are some of the things that we did to make the DSE-7000so easy to use:

Selected topics from usability engineering

Given the space constraints of this application note,I will present only a few ideas from usability engineeringóenough to giveyou a feeling for the field and hopefully to raise your level of interest.

Affordance Whena control device's appearance reveals how it is to be used, it is saidto be an affordance. A button with a concave surface invites pressing,while a cylindrical knob suggests grasping and rotation.

Mappings The manner in which control layout andaction match those of the controlled elements is called a mapping. Theclassic example is the control layout for the burners of a stove. In myexperience, stove controls rarely map well, often requiring a second attemptto get the correct burner lit.

Control sense Users expect an increase in the controlledparameter (e.g., volume) when they move a slider upward, push a slideraway, or rotate a control knob clockwise.

Consistency When multiple occurrences of similarinterface features are handled uniformly, the interface exhibits consistency.For example, a music system with separate button pairs for CD track selection,tape search, and FM tuning should have the right hand buttons all increaseor advance the media. Consistency is generally highly desirable, althoughthere are exceptions. See Grudin for an interesting presentation of consistencyin its three domains: internal to a product, among similar products, andbetween a product and an analogous, perhaps earlier, product.

Mental model A product should encourage the developmentof a mental model of its operation. In the DSE-7000, a computer based audiostorage and editing system, the dedicated user work surface enforced themodel of a multi-track tape recorder and a mixing console. The graphicdisplay further emphasized the model with both a local, detailed trackdisplay that scrolls slowly from right to left, and a global display ofthe total duration of the "tape."

Errors Good design will result in a product thatis so clear that there will be few errors. Those errors that do occur shouldbe handled by the product with grace, forgiveness, and a way of backingout. Certain Hewlett Packard calculators have several features that reducethe pain of errors. If you enter a digit of a long number incorrectly,you can backspace and correct it. If you press a function key and suddenlyget the sinking feeling that you wanted COS, not SIN, you simply hold thekey for a second or so until the display shows the function name, thenrelease the key, aborting the operation. In good design, you should anticipateall the errors that you can and provide means to recover from them. Surelythe days of "Abort, Fail, Retry?" and "General ProtectionFault" should be numbered....

Minimalism Minimalism emphasizes thebenefit of simpler products, fewer features, and more concise documentation.In the competitive world of audio and video products, not to mention software,manufacturers vie to provide the most features. Software manufacturersare driven by a desire to populate their row of the large product/featurematrices seen in magazine reviews with a solid string of check marks.But sometimes less is truly more. Does the CD player in a boom box reallyneed to allow the creation of a sequence of track playback? Try givingthe user less to wade through, less overhead, less to read and she willachieve more.

Iterative design When designers test their productwith real users and modify it several times until hesitations and errorsare at a comfortably low level, they are practicing iterative design.

Visibility Controls ideally should remain visibleand available at all times, reminding the user of their existence and facilitatingthe formation of a mental model of the product. When a product's interfacecan be in more than one mode, control visibility comes and goes and errorsoccur easily. If there must be multiple modes, then a clear display mechanismto indicate the current mode should be provided.

Transparency The nirvana of a user interface isthat it vanishes and the operator simply uses the product to accomplishhis tasks. A well designed automobile may present the driver with nearly100 controls, but he rarely has to think about "the user interface."

The total environment Good product design is impossibleunless engineers put themselves in the shoes of their users by visitingthem where they work. One must learn how the product is learned and used,where its input comes from, where its output goes, and what other equipmentis in the work environment. When we were developing the DSE-7000, we observedradio production engineers preparing broadcast spots. I saw several strikingand unexpected things. One was that these people were under so much timepressure that they worked quickly and instinctively, despite using whatwe regarded as older and cruder tools. Another was the energetic ways thatthey moved while working: rolling around from one machine to another onchairs with castors, jabbing buttons, grasping tape reels, and manipulatingmultiple faders. These visits led us to abandon a mouse and graphic screenas the primary interfacewe took what had been an optional dedicated worksurface and made it a mandatory part of the system. I just couldn't imagineconfining these production engineers behind a mouse, keyboard, and graphicsdisplay (yet this is just what most competitive systems had done at thattime).

Task oriented design It's good practice to explicitlystate the tasks that users expect to perform with the equipment you aredesigning, then order them according to importance and frequency of use.The user interface should support the execution of these tasks, reflectingtheir relative importance.

Stages of user interface evolution It has beenobserved (Grudin and Gentner) that as new products evolve, they are atfirst modeled according to their underlying technology. Gradually, theybegin to become more oriented to user tasks. A particularly intriguingarea of user interface design is deciding how much the interface shouldbe based on user models of earlier tools, as opposed to new and bettermodels. This came up several times in the design of the DSE-7000 and inmost cases we opted for mimicking earlier tools. For example, we consideredusing what I called the "audio snippet" method for locating aparticular instant in a recording by ear (the way your CD player cues andreviews within a track). But because we knew that all our users were accustomedto "rocking" the tape reels and locating the cue point by "scrubbing"(listening to growling audio), we replicated this instead. It was probablythe right choice, although it is arguable that faster and more accurateresults might have been obtained with the use of audio snippets. And, asthe years go by, fewer and fewer engineers will remember open reel recorders,while they all will be familiar with snippet cue/review from their CD players.One has to chart a thoughtful course between the old and the new. It iscertainly the case that the possibilities opened up by new technology eventuallymust force user interfaces to break free of old models

Testing methods Formal usability testing involvescarefully structured experiments involving real users asked to carry outcertain tasks under controlled lab conditions. Often the users are videotaped through a one way mirror for later use in focusing on particularusage problems. A good program of formal testing will result in a superiorproduct.

Heuristic usability effort Even if you can't afforda full blown usability project, a scaled down treatment is still far betterthan none at all. Nielsen and Molich advance a convincing argument forat least having a few usability professionals audit a new product, testingit against a short list of usability heuristics. Such testing can uncovera significant percentage of the more glaring usability problems.

It's not the user, stupid Many people assume thatits their fault when they are flummoxed by a piece of equipment or software.They feel stupid, clumsy, and out of date. While the inexorable forwardmotion of technology with its ever-expanding possibilities does indeedchallenge the user to learn new skills and master new devices, a majorportion of user interface difficulties stem from bad design..

Two perspectives on usability from literature

I think often of these lines from Yeats' Adam's Cursedescribing the creation of poetry:

I said, "A line will take us hours maybe;
Yet if it does not seem a moment's thought,
Our stitching and unstitching has been naught.

The reward for hours of usability optimization shouldbe a product whose user interface vanishes. The product is effortless touse...usability nirvana.

I'll close with an excerpt from "Wind, Sand, andStars," by the French pilot/author Antoine de Saint Exupéry.Writing in 1939 of his experiences as a mail pilot in France, Africa, andSouth America, he describes the fit between himself and his plane's "userinterface" in language that is pure poetry:

Meanwhile, startling as it is that all visible evidenceof invention should have been refined out of this instrument and that thereshould be delivered to us an object as natural as a pebble polished bythe waves, it is equally wonderful that he who uses this instrument shouldbe able to forget that it is a machine.

There was a time when a flyer sat at the centre of a complicatedworks. Flight set us factory problems. The indicators oscillating on theinstrument panel warned us of a thousand dangers. But in the machine oftoday we forget that motors are whirring: the motor, finally, has cometo fulfill its function, which is to whirr as a heart beats--and we giveno thought to the beating of our own heart. Thus, precisely because itis perfect the machine dissembles its own existence instead of forcingitself upon our notice.

Notice

Orban and DSE-7000 are probably trademarks or trade namesof Harmon International or its subsidiaries.

Andersen, Ken, "To define easy-to-use...," PersonalEngineering & Instrumentation News, December 1993.

Chickering, John, "Orban: Big Payoff in Small Markets,"Radio World, March 5, 1997

de Saint Exupéry, Antoine, Wind, Sand, and Stars,Reynal and Hitchcock, 1939.

Gentner, Donald R., and Jonathan Grudin, "Why GoodEngineers (Sometimes) Create Bad Interfaces," Proceedings CHI '90,April 1990.

Grudin, Jonathan, "The Case Against User InterfaceConsistency," Communications of the ACM, October 1988, Volume32, Number 10, page 1164ff.

Grossman, Sandra J., et.al., "Team approach improvesuser interfaces for instruments, EDN, June 4, 1992.

Knox, Stephen T., et.al., "Directed Dialogue Protocols:Verbal Data for User Interface Design," CHI'89 Proceedings,May 1989, ACM.

Nielsen, Jakob, and Molich, Rolf, "Heuristic Evaluationof User Interfaces," Proceedings CHI '90, April 1990.

Nielsen, Jakob, Usability Engineering, AP Professional,1993.

Norman, Donald, The Design of Everyday Things, Doubleday,1990.

Wiklund, Michael (ed.), Usability in Practice, AP Professional,1994.