Visual cues predict system behavior
What this criterion checks
Evaluates whether visual elements and patterns throughout the application help users anticipate the behavior of the system. Assesses whether hover, click, and inactive states of interactive elements are helpful, whether loading and empty states are consistent, whether symbols, colors, and layouts are predictable across screens, and whether menus and options are logically structured. Also evaluates whether error and success states provide appropriate visual confirmations.
How it's assessed
The agent navigates through the application following realistic user journeys, capturing screenshots at each step. Using AI-powered vision analysis, it evaluates whether visual elements help users predict system behavior and checks whether visual design principles are applied effectively, whether visual cues serve multiple purposes creating ambiguity, whether interface elements align with users' mental models from the real world, and whether interactive controls resemble familiar real-world or digital objects. It also checks whether excessive font or color variation causes confusion, whether icons follow standard conventions and include labels, and whether buttons provide visual feedback when clicked.
How reliable it is
This criterion performs well against expert review. Treat its findings as strong leads and confirm before remediating. See Accuracy & confidence for how the confidence tags are set.