One recurring problem with discovery is that users may only be using a subset of features they are comfortable with, and not know about other features, especially if the features have been added recently.
- Appealing to psychology, people's desire to complete the set, can be one way to alert users to the existence of features they have not used.
- New features added to a web app can be advertised through use of a blog.
For discovery there often needs to be an assumption that users will not read the manual, because that is often so. That can make it important to have more than one way of doing things - an obvious way that is discoverable, and a more efficient way that it does not matter much if casual users don't find. Keyboard shortcuts are often a problem in desktop apps. Making keyboard short cuts configurable is one way to advertise their existence and the existence of the power-user commands to which they can be bound.
Discovery is often thought of in terms of discoverability of particular features, but actually it depends on the whole UI. For example features in a UI with a good undo facility are more discoverable than in one without, because exploring is safe.
Closely Related Tags: affordance
Creating a 'Discovery' Question
Constraints on what is allowed to change in an application are particularly important in 'discovery' questions. Saying what's behind a constraint can improve the relevance of answers.