If you have quite a few ideas for products, but none are knock-out-most-compelling to dive into, how do you decide which one to pursue? How much time do you invest in each one's research and feasibility?
Starting points from the answers (so far):
Thanks to Jason's answer, I found this link about Eric Ries and MVP
Use the Lean Development strategy as espoused by folks like Eric Ries.
Specifically: Decide what the Minimum Viable Product is (MVP), meaning the simplest, ugliest, crappiest, least-features thing it can be, and then (if it's not an overwhelming amount of work), build it and see what happens.
If it's a good idea, even that will be enough to iterate your way to something you'll be proud of.
Set up a single web page that describes the product you would build. Briefly! Something you could read and know if you're interested in 30 seconds.
Then put a form there where people can sign up to be notified when the beta is available. And maybe a field for notes/questions.
Then put out AdWords and put out the word on your social network to get people to the site.
If the response is not good, ditch the idea. If a lot of interest, you not only know what to start building, you even have a list of alpha testers!
Not the whole of a strategy, but it should enter into it:
Use Google's Keyword Tool to gauge general interest in the market.
Additionally, once you have an idea of what people are searching for, you can go see what leads on different searches and if there's an unexploited hole in the market. If you're good, you can start doing SEO before your product is finished so that when you're ready to ship, you have a ready audience.
Obviously, this doesn't work in all cases. But it's a good idea for a lot of endeavors.
Pick an idea that you are really excited about and provides some personal benefit. Implement something simple, even if it is crappy idea, but has some value. Once you get something working and you use it you will see features that are not working as planned new features that provide a lot of value.
In my case with the code generator I have been adding modules to validate the database structure, extract one record with all its depended records, create a MySQL entity relationship diagram, create extract scripts. When I started out it was just going to be a code generator.
Of course you might not want to listen to my advice since I have not pulled the trigger and launched the product, but I had a lot of fun developing it and I use it personally on a daily basis.
Here is a technique that I've advocated before on the forum and it's used to solve exactly this type of a problem (along with others :) Do this once for each idea that you have
Do this for each idea. Ideas may share outcomes... That's fine. If they do, comparison is easier.
Now, once you have these models choose the ones with the most compelling outcomes/benefits/value whatever metric you want. If outcomes are shared you'll have one large graph which is easy to parse in the head and see which path is the most promising.
You can augment this with the following (you could start with this step as step 0 or iterate back and forth, doesn't matter)
List out your 'conditions of satisfaction' for pursuing an idea. This set of criteria should be uniform, irrespective of your choice. Use this as a reference to choose a 'path' from the above graphs.
If you are having a tough time 'ranking' your criteria you can do this:
Now you can compare your ideas/paths/goals/benefits to this set of criteria and see which one satisfies (or satisfies) them the most.
Go after that product which you are more sure than the rest.
This way you spend quality time with the product that has a better rate of success...