Internal to the development team (cheap but inaccurate), while the product is being developed.

The basic rule of evaluation is: the earlier you assess the system, the less accurate the results but also the less costly to fix issues. Therefore:

• Evaluate Early and Often: Identify major errors early to prevent costly fixes later.
• Prioritize Early Evaluations: Reserve later evaluations for minor issues or formal assessments.

What to evaluate: sketches, mock-up on a pc, wizard of oz (high quality moved by an human), prototype, the working system.

Activities in the inspection phase:

1. Cognitive walkthrough
2. Action analysis
3. Heuristic analysis

1 and 2 are task-oriented evaluations, 3 is heuristic evaluation

Cognitive walkthrough

A fictional and step by step execution of a task, and the empirical evaluation of the likeliness of the fiction.

It is a method used to anticipate a user's actions and thoughts when interacting with an interface for the first time. It involves creating a credible story of how a user would complete a task using the interface. This story must align with general user knowledge and interface feedback.

1. Ingredients Needed:
   • Detailed interface description or prototype.
   • Description of a representative task.
   • Complete list of actions required to complete the task (often called the "Happy Path").
   • Clear description of the user, their skills, and expectations.
2. Building a Story and Evaluation:
   • Task by task, construct a story of how the user would complete the task.
   • Evaluate the story's credibility based on user assumptions and interface feedback.
3. Identifying Issues:
   • The walkthrough can uncover problems like assumptions about user reasoning, commands not obvious to users, label or prompt issues, and feedback problems.
4. Avoiding Common Errors:
   • Distinguish between listing actions and actually storytelling.
   • Differentiate the cognitive walkthrough from real user tests, as both serve different purposes.
5. Differences from Scenarios:
   • Scenarios don’t include a prototype or the happy path.
   • Unlike scenarios, cognitive walkthroughs focus on evaluating interface usability rather than constructing it.
   • Scenarios are inherently believable, while cognitive walkthroughs aim to become believable through iterations.
6. Self-Evaluation Questions:
   • Assess whether the actions by the character are realistic.
   • Check for control availability and clear linkage between action and control.
   • Evaluate the appropriateness of feedback provided by the interface.

The goal is to uncover potential issues with the interface by realistically imagining user interactions and refining the design based on these insights.