It's extremely difficult to self-manage if you haven't been in a safe situation to experiment with it.

Personally, I find that the only sensible and generally applicable solution is to set short-term goals based on a formally written project plan to yourself.

You don't ever have to give that project plan to anyone else, but you should write it as well as you possibly can. It should be clear on what kinds of things are feasible to do, what is required, what you can do first to prove the concepts, and so on.

I do this for every project I start, after struggling with the problem in graduate school a bit. It takes about a week per proposal to put something together of reasonably high quality, but by the end you actually probably have a very good sense of the strongest argument you can make to do the project as well as the weaknesses, as well as ideally a logical argument for why it might work based on citable evidence. This would also include things like "this experiment I can do" (translated to "this market idea I can test" in business) will generate some result that is clear and helpful in moving towards your medium term goal -- which is probably to either make enough money or be able to raise enough money to pay yourself and enough staff to move to the next medium term goal.

Then take that medium term goal and break it down not into time-based goals but sensible sub-goals. If you want to build a web service, a first goal might be "I think that people have a use for... I can test that by building... and collect data using... if that works, then it's worth doing the second goal..."

Break it down as small as is sensible to do, but limit this kind of strategizing so that you don't go really far down a road before putting it aside to sit and let you get a fresh perspective. Bounce it off others to see if they can poke holes in your strategy that you didn't think of. And so on.

Anyway, it's not so much self-discipline as it is realizing that big goals are always changing and medium term goals are uncertain. You have to always be able to change those goals based on the actual data you collect. But the short term goals should be achievable. And it will take time to isolate what good short term goals should be.

A good short term goal might be, to start, to make a good start on identifying short term goals...

That's a good idea - even though I'm in the middle of the project, and the project is big enough where a plan of the entire project wouldn't be feasible, I can at least do it for the chunk I am working on right now. I think just being explicit with my tasks would be a huge step. Right now everything is vague and kind of just in this "idea soup" in the back of my mind. Every time I pull out the next idea to work on, I realize I had all the wrong expectations the whole time.