3.5 Prototype Conclusions

To truly evaluate the success of our design, we need measurements such as can be obtained through usability testing. At this point, we have no such quantitative values, but there is still a way we can gauge our effort. In Phase I, we discussed what the problems with the current interface were, and included a list of several ways that we proposed to solve such problems. It will be instructive here to go back over that list to see how we did.

• Make use of multiple media to create an enjoyable, useful environment
  This goal is achieved to a large degree in our design. We use video, audio, and text smoothly throughout. These three media complement and reinforce each other. They are not used so gratuitously, however, that they distract from the task at hand. Rather, they provide a natural feeling of interaction for the customer.

• Provide ease-of-use and convenience for customers
  This is, of course, very hard to quantify, and will only become evident once we do some user testing. We believe, though, that our system will be very easy to use, and that it will require very little learning. This is really a requirement for a system that will be used by so many people.

• Be installed in multiple instances, so that several customers can order at once
  This goal does not really refer to the interface design itself, but our design definitely does not preclude the possibility of realizing this goal. In fact, since no human employee need be involved, having several ordering stations would in fact be easier to implement.

• Give the customer total control of the ordering process
  We have largely succeeded in this area. Unlike a human employee, our system will never place pressure on the user to order, especially if there are multiple ordering stations. The customer can undo any action, and can wait until completely satisfied with the order before submitting it.

• Provide many levels of information detail on menu items
  Here, a drawback of our system is brought out. The choice to use speech-recognition, and to limit it to a small vocabulary, means that the amount of information we can offer is relatively small. Although not specified in our current design, this obstacle could be overcome. When the user made an order, the system could automatically present a picture and/or textual information on the item being ordered (e.g. nutritional information). This satisfies the requirement above, but it violates the principle of total user control. Ideally, the information should only be presented when it is asked for by the user.

• Present immediate feedback on orders for customer verification
  This is a major component of our system, and one of the greatest improvements that our interface makes over the present one. We offer feedback in a variety of forms. The order is shown in the "Your order" field immediately after it is processed. The command given and the response of the system are printed in the other two fields. Finally, the system responds in an aural fashion to every user command. We claim that this feedback could drastically reduce the number of errors made, both by customers and employees.

• Allow multiple methods of payment
  Again, this is not an interface issue in the strict sense, although it is specified as part of our design. As with the multiple ordering stations requirement, though, our design lends itself to the implementation of this goal. Since the system is completely computer-driven, the payment system may be integrated into the system, again taking human employees out of the loop. Using credit cards in the present system would be quite difficult and time-consuming. With our system, however, it would become a natural, integrated part of the ordering process, and verification over the phone (the first phase, anyway) could be done while the order was being taken. Then, when the final total was accepted, the system would be ready to finish the verification with the credit card company, and no time would be wasted.

There are still improvements to be made, of course. If such a system were implemented today, it would suffer from the lack of completely reliable voice-recognition technology. Also, we did not meet our goal of providing a greater depth of information on products. We feel, though, that our advances in the areas of feedback to the customer, reduction of errors, and total user control make our interface design very desirable, and a great improvement over the current interface.

Back to Phase III index

On to Phase IV