Ask HN: evaluating project ideas, how do you do it?

I've read some articles in the past days about idea verification before diving into it. I think that nails the problem I have when building stuff - it looks like it might be useful for someone to pay for it, but I don't really know. I've given up on "build it and they will come" model because there's simply no time to execute false ideas while maintaining a full time job.

My last app is youtube -> mp3 converter, I've spent 3 weeks of midnight grinding building a python powered backend server on top of ffmpeg and another ~2 weeks building the frontend website. Launched in mid Jan it now has very few subscribers, adsense rev. was laughable so I took it off completely. Should I have known how hard it was to gain search engine rankings, how unwilling users are to pay for such service - doubt I'd do that again. But I did not verify it, I simply hoped there's still room in this space (well, maybe I'm just doing it wrong, who knows). How could have I verified it?

My next idea is to build a self hosted client, workflow, invoice and preview manager targeted at small digital service firms, such as countless psd2html companies so that they could install it, have a link on their website called "client area" where their customers could login, see progress, invoices, give feedback on shown work. Now this does seem useful to me as I've been into these type of services, but would anyone pay at least 50 bucks per year? To verify it, do I just create a wufoo form and approach target audience directly, promising some incentives for filled survey?

Appreciate any insights.