How about a little insight into web analysts' behaviour instead of visitor behaviour today? 

The web analytics vendor (Omniture) we use provides us with four different tools that offer varying degrees of complexity.  From a "big picture" tool, ClickMap, all the way to a deep dive tool, Data Warehouse.

In discussion with a couple of other Sun web analysts recently I made the comment that there are "too many tools.  I guess you can't build a house with just a hammer."  The tool you need really depends on how much precision, data confidence and expense you want.  We do this decision making continually and think nothing of it.

It is very important to consider your confidence level with each tool's output.  Most of my work is done somewhere in between with SiteCatalyst and to some extent, Discover.  I just don't trust ClickMap very much.  The reason I hardly use Data Warehouse is because of time constraints.  This is another factor web analysts consider when choosing a tool.  I usually don't want to wait days for information.  So I usually prefer a lower level of confidence over a long wait time.

Another factor is cost.  The smaller the level of confidence, the shorter the wait and the smaller the cost.  Cost is really a factor of wait time and is not determined by the level of confidence.  The level of confidence is a by-product of the wait time.  Quick is cheap and dirty.  Lengthy waits beget precision and are more expensive.  You can see that precision and level of confidence are intertwined.  Cost begets precision which begets confidence level.

What I find interesting is that we usually skip the first step mentally - Cost.  We just automatically assume that we can't wait and can't spend money.  While the latter is usually true, wait time isn't always as expensive as we think it is.  In this fast-paced world, faster is assumed to be better even if precision is lost, so long as our level of confidence doesn't run dry.  I guess this is just how business works.  We strive for a balance between precision and confidence level to get answers as quickly as possible at the smallest cost.

I'm curious about how much you think about this on a day to day basis.  Do you prefer low cost, high precision or high confidence levels in your analysis?  In which situations would you prefer one over the others?

Happy analyzing your analysis behaviour!