And, remember that "average" is a very poor statistic to use. I've never seen a site where a few incredibly long visits doesn't pull the "average" up high. You can have half your visitors spending 40 seconds or less on the site, and the "average" could still be 3 minutes.
If your stats package has a "median" visit length available, that's much closer to what most people would call "typical." In the above example, the median would be 40 seconds.
If your stats package has a report listing lengths (under 1 minute, 1-2 minutes, 2-3 minutes, etc) with the number of visits for each, you can get a much better idea of how long most visits are, and you can use that table to calculate the median.
Also, remember that the "average" may or may not include all the very short visits that are the single page visits. A single page visit has no length at all because there is no second page view to show how long the first page was viewed. If the stats package doesn't include these ultra-short visits in its calculation of "average," then your "average" is, again, far too high and nowhere near a good representation of "typical."
The e-comm sites I have experience with averaged 7 minutes, with a median of about 2.75 minutes. But, there are lots of different kinds of e-comm sites with difference ideal visit lengths.
This is about the point when somebody else should jump in and say "conversion is the only thing worth measuring."