The analyzed object is a fixed method; exploration, hesitation, and learning are outside the model

What it is

The GOMS model analyzes the execution path once a user already knows the goal, method, and selection rules. It does not model exploration, hesitation, or learning — "where should I look," "is this icon clickable," "how do I learn this the first time." This card is about what gets fixed on the input side before any modeling starts: "the user already knows how to do this" is a precondition that must already hold, not some output limitation discovered after the model runs. That is a different-stage question from "the model predicts only error-free skilled performance," which is about how to read the output.

Why it happens

GOMS requires the object of analysis to be an already-fixed method because it is designed to sidestep a combinatorial explosion. If "the user does not yet know which method to use" were also brought into the analysis, the model would have to simultaneously characterize the user's current mental model, the cues the interface offers, and how the user would guess based on those cues — this is no longer a problem task decomposition can handle; it requires an entirely different cognitive architecture to answer "how does a person decide under incomplete information." GOMS's solution is to move that whole tangle of complexity outside the scope of analysis entirely: assume the method has already been chosen, and analyze only the execution sequence that follows. This is also why exploration, hesitation, and learning are singled out by name — exploration corresponds to search behavior before a method is selected, hesitation to the uncertainty during that selection, and learning to the method itself changing with experience. All three share the trait of happening before the precondition "the method is fixed" holds, and so none of them was ever part of the question this model set out to answer.

Where it stops holding

This is not a retreat from reality; it is a statement of scope. In high-frequency internal enterprise tasks and control-console workflows, users who have used the system for a long time genuinely do converge on a fixed method, so "the method is already fixed" essentially holds and the model can be applied directly. But for first-time use, choosing among several unfamiliar options, or creative tasks that deliberately encourage exploration, this precondition fails to hold from the start, and any conclusion drawn from applying the model has no footing — it is not that the model computed something wrong, it is that the object of analysis was chosen incorrectly to begin with. The first step in deciding whether GOMS applies should be confirming whether target users have genuinely converged on a stable method, not building the model first and discovering the assumption failed afterward.

Applying it

• Before modeling, confirm from logs or interviews whether target users have actually converged on a stable method; if different users take very different approaches to the same goal, the method has not stabilized yet and GOMS is premature.
• Explicitly record the task goal, the set of available methods, and the proficiency level of the users the model targets, and put these preconditions in the report, rather than letting readers assume the conclusion applies to everyone.
• For interface sketches during the exploration stage or novice onboarding design, use observational usability testing instead of GOMS, since exploration and learning are precisely the part this model excludes.
• How to check: sample-observe target users' real paths through the task; if most people do follow the method the model assumes, the precondition holds. If you instead observe extensive probing, method-switching, or repeated confirmation, the task is not yet at a stage where GOMS analysis applies.

Related

• Same group: B4.11.2 When multiple methods exist for one goal, correct selection rules determine whether predictions hold · B4.11.3 Decomposition granularity is optional, but alternative rankings should remain consistent across granularities · B4.11.4 The model does not predict error rate, satisfaction, or learnability · B4.11.5 Modeling cost is high; returns are positive only for high-frequency repetitive tasks
• Nearby: B4.04 GOMS · Q2 Usability Evaluation
• Search terms: GOMS assumptions · skilled performance · exploration · scope of analysis