For the example projects I gave the sales prediction one is growing the fastest, followed by the mechanical failure predictor. The sales prediction software should be 1.5 years to break even, the mechanical failure one probably a little longer. The other ones will fund my next TV / Macbooks / Holidays etc ...
My customers are usually start as friends / acquaintances / random business people I meet, then spreads through word of mouth.
Sorry if it's vague, let me be give my most recent example (The sales prediction software). I have a friend who is a manager of a small supermarket chain in New Zealand, I ask virtually every business owner I meet (anyone anywhere, including random people I meet on the train), a list of questions, one of which is "What is the biggest problem in your business at the moment" ... he said "Inventory Management". We got talking about his purchasing process, how he orders, what determines the levels he orders, all those sorts of things. I tell them to treat me like an idiot, this gets us to the root cause of the problem faster and also builds a good dynamic if I 'admit' I'm dumb ... I become less intimidating.
I then build MVP, iterate rapidly based on their feedback. MVP (v2) is always quite different to MVP (which is why I dont spend a lot of time on MVP)
Quickly my supermarket buddy is talking to his supermarket buddies saying "you have to use this software" ... so I make my customers do the selling for me. Kind of cheeky I know, but hey, it works.
I'm by no means a pro ... I have build a good 20 other projects that have all failed miserably. But leant from them all, and as I said in an earlier post, I have the sheer bloody minded determination that means failure can punch me right in the face and I will always get up again (haha)
