r0ml has an article on service quality over at Doc Searls IT Garage. The two processes that are important are measuring service quality and then using those results to improve the service quality. As r0ml says:

The objective was then (and still is) to measure the availability of the system as perceived by the users.

From Service Quality | Doc Searls' IT Garage
Referenced Tue Jun 01 2004 10:19:52 GMT-0600

This is harder than it sounds. Most system monitoring tools are built to tell you if the machine is up, but that's a far cry from being able to tell if the service is working. I remember well our own evolution toward delivering 24/7 service quality at iMALL. My goal was to never have a user (especially the user with the title CEO) tell me my service was down before I knew. This requires money and discipline. I've documented some of what we did from an organizational and process standpoint in a white paper on tiered support.

A funny story regarding our trip down this path at iMALL is that when we got bought by Excite@Home we were far enough along that we knew how much we didn't know. We took advantage of the opportunity of being acquired to spend a lot of time with the Excite.com operations folks and basically soak up everything we could learn from them. They were good and had decades of experience.

Later, after Excite@Home had gone through numerous reorganizations and staff changes (it seemed like years, but it was really only about 10-12 months later) all of those people were gone and we had become the corporate memory for many of the techniques and tools and basically retaught it to a completely new group. It was eery to see how fast a company could lose all of its skill and capacity in a crucial area. The turnover was just too great and too fast for the skills to be passed along.