We get a lot of requests to conduct usability testing on a website as part of the research in preparation for a major redesign. While testing at any time can provide valuable insight into the user experience and help identify problem areas, I have to question the expense at this stage unless one focuses on specific functionality that is likely to carry over with little modification. Often there is such significant change on the new site that early testing becomes meaningless.
I would rather see budget dollars reserved for testing a prototype based on interactive wireframes for the new design. Even though we apply best practice and expect to see major improvements in the user experience, I have yet to conduct such testing and not discover opportunities for even better approaches. Sometimes a navigation label that seems obvious to the team is unclear to users or perhaps a button is placed in a area of the page where it is easily overlooked.
I recall testing on a medical site (not the one below) designed for both consumers and professionals. We decided to shorten the label directing healthcare professionals to an area with content focused on their unique needs. We dropped the word "For" in front of "Healthcare Professionals," thinking it was unnecessary. However, consumers then thought they could click on that link to find a provider. That simple word was critical to the user experience and we would have missed it without testing.
The above image shows an example of interactive testing at the wireframe stage using a number of pre-defined user scenarios. However, user testing can also be done even earlier in the development process using card sorting exercises to test navigation labels and proposed taxonomy or paper prototypes to test page layouts. Think carefully about what sequence of testing will be most helpful to your redesign efforts.