Pros and Cons of the Approach

The GOMS approach to user modeling has strengths and weaknesses. GOMS can be used to model how skilled people will use a system. Designers can develop laboratory usability studies as an alternative to GOMS. However, designers may have to provide extensive training for usability study participants. GOMS gives designers the ability to make quantitative predictions about skilled behavior without having to train people. The keystroke-level model provides excellent quantitative fits to the performance times of skilled users during error-less performance. The keystroke-level model parameters have been proven to be stable, thus performance in similar new situations can be accurately predicted. Parameter-free estimates makes the GOMS approach useful in design because it allows comparisons of different design alternatives. GOMS can be used both quantitatively and qualitatively. If, for instance, you have to choose between two systems, GOMS can be used to predicted the performance time and learning time trade-offs of competing systems. GOMS may be used to discover usability problems. Analysts are forced to focus on the structure of the task as the user sees it when developing the goals and methods definitions in GOMS.

Card et al. (1980) provided the most detailed list of the weaknesses of GOMS. The weaknesses are as follows:

1. The model applied to skilled users, not to beginners or intermediates.
2. The model doesn't account for either learning of the system or its recall after a period of disuse.
3. Even skilled users occasionally make errors; however, the model doesn't account for errors.
4. Within skilled behavior, the model is explicit about elementary perceptual and motor components. The cognitive processes in skilled behavior are treated in a less distinguished fashion.
5. Mental workload is not addressed in the model.
6. The model doesn't address functionality. That is the model doesn't address which tasks should be performed by the system. The model addresses only the usability of a task on a system.
7. Users experience fatigue while using a system. The model does not address the amount and kind of of fatigue.
8. Individual differences among users is not accounted for in the model.
9. Guidance in predicting whether users will judge the system to be either useful or satisfying, or whether the system will be globally acceptable is not included in the model.
10. How computer-supported work fits or misfits office or organizational life is not addressed in the model.