All of these things come into sharp focus where there is interaction with the customer: multiple aspects of dealing with customers, or would-be-customers are designed in isolation, obvious bridges between aspects are not even considered, never mind built and yet, YET, this is exactly the kind of use algorithmic analysis is perfect for. 

Businesses do, of course, build their systems with an objective in mind, start here, finish there. There is a simple, single, minded focus on the end of that particular tunnel. But it's not a tunnel: it's an open road. 

But.... 

When those systems operate in isolation, they often operate at different speeds: part of it, devoid of obstacles, races ahead; part of it, designed to identify and address obstacles and, even, to block progress in appropriate places. 

This happens across organisations: it happens when S&M want to bring a new product to market; it happens when a deal is being done and, crucially, it happens during the customer acquisition process, often called "onboarding".

The result is that one part of your company is pressing the customer to get on with certain activities, it might even be under pressure that a discounted period will soon expire while another part is finding reasons to reject the customer. Sooner or later the customer is rejected or gives up. 

Worse, if you look at what some simple, single-minded commentators say, they advise that abandoned applications are prima facie a cause for suspicion. Who says that? The UK's financial sector fraud body, CIFAS. 

The fact is that it's not the customer's fault, in most cases: it's because there's hub and spoke arrangement when there should be a spider's web. Connections should be made so that progress is coordinated and, here's a shock, tech can make that happen. 

Or you can do that by looking at real world examples of what's gone wrong and make sure your company doesn't make the same mistake.