Something that I find a real struggle is staying focused on developing features that actually matter to the end user. Working on NewsTrapper it’s extremely easy to continually work on the more technical Solr/NLP side of things, after all that’s the real heart of the system and where all the challenging work is. The trouble is, improving that part of the system isn’t what’s holding us back from being able to launch. The point of an MVP is to have the minimum features possible for it to be useable. It doesn’t matter if our indexing is terrible as long as someone can come along and use the system in some form of intended fashion.
I find it really useful to review what it is I’m working on and what actually needs to be done almost constantly. Many times I’ll actually drop what I’m doing and move onto something less appealing, but more pressing. Break your system down and answer the question of “What does a user need to achieve their goal?” and do the bare minimum, no matter how bad, to deliver that. If you’re doing a small private beta then it doesn’t have to be perfect, it just needs to prove the point and confirm whether your idea is useful or not.