Recognizing an option from a list is much easier than producing it from memory. Interfaces that rely on recall (unlabeled icons, blank fields, codes you must remember) push work onto the user. On this site the good examples use logos, labeled chips, thumbnails and copyable details; the bad ones hide meaning behind icons or random strings.
Roger Shepard (1967) showed how strong recognition memory is: after looking through several hundred words, sentences or pictures, people picked the ones they had seen with roughly 90% accuracy for words and sentences and about 98% for pictures. Nielsen made the idea heuristic 6: minimize memory load by keeping elements, actions and options visible. Raluca Budiu at NN/g notes that long up-front tutorials are a weak substitute, because users forget them; cues on the current screen work better.
New users have nothing to recall yet. In onboarding and setup, every icon without a label, every empty field without an example and every value you expect them to remember later is a place they can stall.
How to use it
- Label icons, at least during a user's first sessions, and use recognizable logos or thumbnails for choices like data sources or templates.
- Prefill or suggest values from what you already know (company name, previous answers) instead of asking users to type them from memory.
- Keep reference details such as hostnames, keys and settings visible and copyable at the point where users will need them later.
Watch out
- Showing every option to avoid recall can tip into clutter and choice overload; show the likely ones and make the rest searchable.
- Generated defaults that users can't recognize (a random URL slug) are worse than an empty field, because they look intentional.
