Obviously, it is not feasible to develop every single feature request people have. There are various reasons to say NO most of the time, and this article give good advice on how to argue with requesters. I highly recommend it.
Some applications seem to be finished. They do everything that is conceivably possibly. Consider Excel, for example. Is there anything that’s not yet possible but should be? Or imagine a calendar application. Everything is already developed, after all, what could there be added?
Still, there are some enhancements that can be made, mainly by linking the original purposes to related purposes. This post on Little Big Details shows how Outlook displays the weather for each day directly in the schedule. Originally two application, the calendar and the weather application where separate, but the integration saves a lot time and effort in seeing whether the weather fits to a planned activity.
I conclude that for such finished applications, integration with other applications can still enhance the product.
An interesting way of deriving features from actual business goals is Impact Mapping. The centre of the process is the business goal, which several groups of people contribute to. Each of the groups can contribute in certain ways, and therefore needs specific features. This four-step approach cannot replace any planning, wireframing, etc., but if the goal is fixed, it helps you see which features are needed by which people.
There has been some discussion on whether the Who or the How should be the second step, and the author decided for the Who. Of course, anyone using this technique is free to adapt the concept.