Deborah J. Mayhew points out you get what you pay for: discount usability will not do the entire job. She goes on to say that guru's also aren't worth the money. While I do think she generalizes a bit, I agree with her on several points, especially the fact that websites are far more complicated and need more extensive testing beyond typical discount methods.

However, there are aspects of discount I still think are right on, including ditching labs and using smaller samples when time is limited. Some usability is still better than none. I think the key problem is that discount usability should have been a fix for projects with tiny budgets and timelines-- instead they've become the defacto standard, and that isn't the way a grown-up company should go.