Optimizing how people use geospatial systems requires an understanding not only of geospatial technologies used but also of how human beings respond to maps and graphics, how they achieve their goals using devices, and how they come to choose the particular actions they choose. More generally it requires knowing why and how users do what they do when they do it. The payoffs of understanding the user generally falls into three categories.
- Understanding the user can save lives. I do not have a geospatial example but in other areas there are numerous examples illustrating where ignoring the user can lead to loss of life. Understanding the user (i.e., the pilot and air traffic controllers) has made airplanes the safest transportation per passenger. Medicine has numerous examples as well. For example, it was found that for nurses to type in the digits of a drug dose is inherently more dangerous than when users dial them in using a wheel for each digit. When typing, a repeated digit can increase the dosage by a factor of ten. This mistake is just not possible with the dial-based interface.
- Understanding the user can lead to new and better products. Understanding the user can make a system that is more usable, more learnable, and more efficient. For example, the rise of Internet mapping and mashups has to be acknowledged as one of the driving forces in the increase in the use of digital geospatial technologies. Traditional GIS historically targeted formal applications that require an informed user. By contrast, the web's new neogeography tends to apply to the areas of approachable, colloquial applications. The two realms can have overlap as the same problems are presented to different sets of users: experts and non-experts.
- Understanding the user can save money. Designing to support users can save development and use cost. One of the best known examples comes from Nynex, the New York telephone company in the early 1990's. The toll and assistance (T AO) operators are the people who help you when you dial "0". They help customers with collect calls, billing, and other more complex calls. In the early 1990's Nynex was considering upgrading their T AO workstation. They had a room with about 100 of these operators located on the East Coast and it was believed that new graphical user workstations could improve productivity. The cost of upgrading all the workstations was going to be about $500,000. To be on the safe side, they had a team of applied psychologists look at just how much faster the new workstations would be. Their analysis suggested that the new workstations would not be faster, but would, in fact, be 4% slower to operate. This may seem like a small difference, but a 4% reduction in productivity was going to cost Nynex $2.4 million a year - in addition to the cost of the workstations. Of course, Nynex did not find it easy to accept this analysis, and so they ran a user study to discover how much faster the new workstations would really be. After allowing time for the operators to learn the workstations, the operators' performance plateaued about where it was predicted - 4% slower. NYNEX now claims that this study saved them $2.4 million per year. It's been ten years since the analysis and the user study - a study that paid for itself in the first week (Gray, John, & Atwood, 1992; Gray, John, & Atwood, 1993).
However, understanding the user does not guarantee success. Getting the usability right may increase the time to get the product to the market, it may make the price inappropriate, or reliability may be poor. Likewise, systems with poor usability can still be successful since they may offer a functionality that is unique. DOS computers, shoebox-size satellite phones, and command line GIS interfaces were all difficult to use, but successful because of their unique functionality.
When do you need to study the user? As early as possible! It is a well known and true statement in the area of design that changes to a design cost much more later in the process. The take-away message from this lesson should be that support for the users and their tasks should be incorporated from the start. Having said this, it is particularly important to get the users involved early when:
- There are lots of them or they are uniquely important
- You (the designer) are not like them
- Lives are at risk