If I had to single out a reason why sites fail that really can be avoided, it would be the lack of testing.  It always seems as though sites are up against the wall to go live and do...only to fail when their users can't use the site for the sole reason it was created.  If adequate testing had been done on the site prior to launch the users' experience would be much more pleasant and better conversion rates would result.  However, it's not as simple as it seems.

What I've discovered over the years is coming up with testing scenarios, while a key to the success of the site, really comes down to how well the client really knows its own clients or potential users. While you are building a site for a client, it would be helpful to have your client coming up with scenarios to test the site.  The typical user scenario is pretty easy.  The ones that are difficult are the unusual users who may be coming to the site.  Now, my friends....I am not saying "unusual" meaning strange.  I mean unusual in the sense that they are not the typical user a client will expect to use their site. 

If you want to help find out what is the atypical user, consider these questions:

1.  Who would your typical user be, and who would your typical user be most likely to tell about your site.

2.  Consider geographical areas, meaning terms you would use that someone from another part of the country wouldn't.

3.  Consider ages.  A person who uses your site could potentially tell their elderly parent who is somewhat new at the internet...would they know exactly what to do on your site...could they get to the parts of the site you want them to.

4.  Consider gender.  Lowe's has increased the amount of sales, and sales decisions by females almost 30% more than its known competitor, Home Depot.  Why?  Because Lowe's considered the female perspective for what would make going to a hardware store (for all intensive purposes) more enjoyable for women, but yet still cater to their target market.  So they were able to keep their focus on the target market of males 25 - 40+ and still increase their sales.  So consider the fact that you may in fact have other genders visiting your site, and consider their perspective.

I don't expect you to be able to test every type of scenario, but perhaps by considering the atypical user, you may actually be able to have a much more stable application/website and really take advantage of those new site surge visitors.