Browsing the web using a one of a kind user agent is not a good way to avoid tracking.
A more convenient approach might be to remote desktop at random to a pool of machines that aren't his and that other people also use, the way he uses Google search.