The Gulfs of Execution and Evaluation

Definition

The Gulf of Execution is the gap between a person's goal and the means available to act on the world; it is bridged with signifiers, constraints, mappings, and a conceptual model. The Gulf of Evaluation is the gap between the resulting state of the world and the person's ability to interpret whether it matches their goal; it is bridged with feedback and a conceptual model. Both gulfs are crossed by way of the "seven stages of action" — goal, plan, specify, perform (execution) and perceive, interpret, compare (evaluation) — which Norman offers as a simplified but generative model of how any deliberate action unfolds.

In the Book

Chapter 2 introduces both gulfs through a broken filing-cabinet drawer: the handle's design clearly signified "pull," bridging the Gulf of Execution — but when the drawer jammed and nothing happened, the cabinet gave no further information about what to try next, leaving a wide, unbridged Gulf of Evaluation. Norman generalizes this into the seven-stage action cycle using a mundane example — the light growing dim while reading, prompting a new goal ("get more light") that must be planned, specified, and performed, then its result perceived, interpreted, and compared back against the goal. He stresses that people habitually explain gulf-crossing failures as personal inadequacy ("I'm being stupid") for simple household devices like faucets and stove tops, or give up entirely on complex-seeming ones like sewing machines and digital watches — in both cases wrongly, since "the difficulties reside in their design, not in the people attempting to use them."

Why It Matters

Splitting "can't use it" into two distinct gulfs turns a vague usability complaint into an actionable diagnosis: is the person stuck because they can't figure out what to do (execution), or because they did something and can't tell what happened (evaluation)? Each gulf has a different toolkit — the first needs better signifiers, constraints, and mappings; the second needs better feedback and a clearer model of system state. This distinction transfers to any interaction with a complex system — software, bureaucratic process, physical control panel — where diagnosing which gulf is actually open prevents wasted effort fixing the wrong half of the interaction.