The Layer Cake or... The Eight Levels of Analytics?

Ron Shevlin once again has some great comments on some confused article in the 1 to 1 blog.
The article presents an implies hierarchy for analytics that lumps analytics (and a wide range of those) with reporting, leading the reader to believe that there is some kind or hierarchy or continuum in these.
Aside from Ron's remarks, here's a stab at straightening these ideas. Reporting is more conveniently divided into the following main groups:

• Pre-defined reports - this corresponds to "canned reporting" that is issued on a periodic basis and in which the reports and their content have been pre-defined. The user has little to no freedom in changing them (an exception is if the reports are issued as a spreadsheet - a bad practice that should be discouraged unless the client explicitly asks for it). These reports seldom have interpretative comments
• Campaign or activity reports - issued at the end of each campaign (again, interim reporting sometime during the campaign is also to be discouraged) and that summarize the results, present the conclusions from whatever tests took place. These reports are issued in a pre-defined format and only the content changes
• Summary reports - quarterly, semi-annual or yearly reports that summarize the results for the time period that they cover. These are an extension of the pre-defined reports with the main difference that they tend to be heavier on interpretation and tend to refer to trends and also track results to targets
• Ad-hoc reports - as the name implies, these reports are single instances or very infrequent instances at best. Their content and design is custom for each.
• Self-serve reports - generally OLAP or some similar variant, these reports are generated by the end user, at will and according to a pre-defined set of measures, attributes, dimensions and levels. There is little to no commentary from the analyst in these reports.

And now I have more content for a post tomorrow: types of analytics!