Welcome to the Q&A Forum

As well as parameters mentioned you may possibly have heaps of duplicating categories, tags etc.  What I would also do is start searching Google with something like site:www.example.com/directory/ or possibly site:www.example.com/category/directory/directory/ so you are tightly narrowing down the results, switch to 100 results per page and manually look for clues.

How many domains do those 24 million links come from?  It's possible you could find that they distill down into a manageable number of domains to manually review.  I think that tools like Link Detox can help in an initial review of links and perhaps you can use that as a first pass to find as many spammy ones as possible.  Then, I would take all of the links that were marked as suspicious or healthy, and visit one link from each domain manually. In my experience, the only sites that I am seeing recovering from link based issues (either manual penalty or Penguin) are ones who have been EXTREMELY thorough on their audits.  It sure is a lot more work to evaluate links by hand but I think it is necessary.

Hmm, good point. I would like to think that they would have something in place to recognize that. e.g. when a crawler from CA crawls and the keywords change to CA, G would recognize that the listings are generated based on geoip (which should be a considered a good thing for relevance). In addition, you said it was a small portion of the page's content so it probably wouldn't affect rankings much. In G's own words, focus on the user instead of the search engine. In this scenario, your users and you have a lot more to gain from having much more targeted listings show up on your site. Cheers, Oleg

Firstly, you might want to create landing pages with better URL's for certain manufactures like: www.example.com/classifieds/honda-bikes/ www.example.com/classifieds/kawasaki-bikes/ And no-index 0 result pages programatically.