Editor's Note: Remember that if you're interested in SaaS or other Web 2.0 issues, then you cannot miss ebizQ's upcoming virtual conference on Enterprise 2.0. Sign up today.

What is so hard about making software? Surely you just collect the market requirements, generate a high level functional specification, build the product, and then test it? All Marketing needs to do is put together some screenshots and sales should be as easy as sitting by the fax machine and waiting for the order?

Unfortunately for software vendors it doesn’t quite work that way. It takes a long time, many years in fact, to get software right, and the problem stems from the first step – understanding the market. Part of the issue is that the market does not understand itself. Go into any large company, and ask them to detail a business process such as order entry, and you will get a variety of differing responses.

Management will describe the process as they think it should be. Unfortunately, this utopian vision tends to ignore the inevitable trade-offs that need to be made in the real world and is at times rife with logical impossibilities. The user population will give a different story. A small number of staff members can be counted on to provide a fair description, but many will innocently leave out critical nuances that are essential to effective outcomes, while others will deliberately conceal activities that are considered out of bounds. Unfortunately, workarounds, however necessary, are frowned upon by management, so are often left out. The net of this is that the definition of the business process that emerges is often inaccurate and/or incomplete and any effort toward automation is bound to fail.

Even when processes are accurately described by the users, if the underlying process was poorly designed, automating it will only make it worse. Many companies have a wide spread in the quality of their internal process. They may have excellence in one department and mediocrity in another. For example, a given company may have great customer support procedures but lousy billing processes – or vice versa. Automation of a poor process like that solves nothing and will likely make the situation worse by creating faster and more efficient ways to fail.