(Founder of Neeva here) -- Chris -- I love the idea. Neeva does this in a contextually relevant manner to the intent of the query. For example, on health queries, we label all health sites as "trusted", "ad-supported" etc. and allow you to filter down to the appropriate subset of results. For programming queries, we label sites as "official sites", "forums", "blogs", "code repos", "programming websites" (the SEO-ed ones). We work with human raters to do this. Would love to hear if you find it useful and what other labels would be of use.