The perfect business would be fully integrated and have no staff (actually the perfect business makes money without staff or customers!) but we must balance the cost of that integration against the return it is going to give. In the beginning of Internet time (1999) a successful businessman set up WebVan, an online grocery delivery business in the US. It was fully automated with 300,000 sq ft distribution centres and special vans. Ultimately the business failed because the margins were too thin and as a start up it was costing $210 to acquire each customer. Tesco started a pilot web grocery business at the same time. After running it for 6 months they shut it down and reflected. I was lucky enough to be at the Manchester Business School at the time and listened to a talk on the lessons learned from one of the directors. They re-launched with in-store picking, special vans and sold to existing customers; they were profitable by 2001.
You can read more about this in John Mullins excellent book ‘The New Business Road Test’ (ISBN 0-273-70805-8). What this shows us is that we need to look at the savings to be made from integration against the volume and margin of sales expected. For new Internet ventures what is the cost of acquiring a customer? The Field of Dreams approach; ‘I will build and they will come’ only works in the movies!