(a) generate and *polish* the idea;
(1) make sure it is technically sound
(2) not trivial, i.e. not easy to copy
(3) maps well on business opportunities
(4) does not have IP (patents, licenses, etc) problems
(b) implement it about 80-90% through, prepare solid beta
(c) release beta and see how it's does
(d) if all goes well, evaluate the progress and estimate how long until the first revenue comes in
Steps (a.2) and (a.3) filter out A LOT of ideas. Step (b) also tends to screw things up as some ideas work out to be hard or even impossible to implement alone or on a modest budget. Steps (c) and (d) are pretty much the last mile before the full-time involvement. If the idea was thought through, the chances of failure on last two steps are not very high.So I'd say 2+5+6 (+8) is the combination I had. Step (a) took about 2 years, (b) - under a year, (c) - 6 months. Your mileage may obviously vary :)