Maintaining coherency of the project's core ideas is my key. I was introduced to this through a philosophy of truth course. If the ideas avoid contradiction and are of genuinely new ground, pursuing the project is a matter of executing until either the project fulfills the concept or some unexpected barrier was discovered. In doing this you have to set testable benchmarks for what makes it "done" and "successful". List them all out. Sometimes you learn your principles through your benchmarks, other times the reverse. The project is often better motivated by a study goal ("I learn this") rather than a utility goal ("I earn this").

If contradictions slip in, then scope becomes unlimited. Most businesses will eventually end up in a contradictory position - and this sets a definite limit on their lifespan. But you can avoid this fate in a side project by removing some of the assumed requirements of business and allowing it to be an obscure toy or a money-loser, instead setting other benchmarks for success.