A Magic Way to Enterprise Software Success ;-)

Joe McKendrick has an interesting piece about the cultural issues of implementing SOA in an enterprise- this got me thinking about successful IT projects more generally: he's based much of his commentary on a challenging article "Finding the Real Barrier to SOA Adoption" by Ronald Schmelzer.
My observation is that everyone seems to accept that enterprise implementations fail between 30% and 80% of the time. However, we've made around 20 investments in very early stage software companies and ALL OF THEM delivered the product both on time and on budget.
I don't think this is too hard to understand: these companies have some common characteristics:

• They're small teams of very very good programmers (you can hire them more easily/cheaply in the North of England!)
• They live very close to the customers
• Their motivation is very highly geared towards their customers' satisfaction
• They have brilliantly supportive investors ;-)

It seems then especially ironic that large enterprises are very hesitant to use very small software vendors as they view them as having higher risk- my answer to that would be risk of "what". A vendor that's still around for a project that failed is perhaps no more useful than having a successful implementation where the vendor has gone out of business!
Finally, what I think large companies often neglect is that the "failure risk" of that small innovative vendor is strongly linked to the small vendors ability to sign up the large enterprise (on good payment terms!)