by: Bob Sakayama
updated 31 December 2014
Some technical knowledge is required to properly evaluate links. The vetting process has evolved with time - the information posted below is the result of many years of work while restoring hundreds of sites. We recommend that vets be conducted by experts because experience really counts when the standards are many and technical in nature. But if you have responsibility for a website's performance, you already know that knowledge of those factors that contribute to risk is invaluable.
You don't want to manually vet a humongus list. Link profiles can be daunting in size, so no one will be manually vetting all the links of a large website. But the information below is invaluable to evaluate a small link profile, a set of questionable links, or the work of a link builder. Although many rely on automation, we know from having tested the available tools that humans are better at this - if they know what to look for. One of the reasons for this is that "intent" is what you're trying to discover - why was that link placed there?
Rely on automation to reduce the numbers of links that have to be vetted.
You don't want to waste time discovering nofollow links, or links from images, or links that have been deleted. All of these are in WMT data, so use tools to quickly discover links you can ignore. Because Google's Webmaster Tools data is sometimes not inclusive enough, we also retrieve ahrefs.com and majesticseo.com link data. We then run a bot to visit each url and return all do follow links, anchors and other data. Since link penalties are often keyword specific, the links that need to be reviewed can be quickly limited to those using specific anchors.
There is no such thing as a link that any link builder posts that is really natural, although it may pass the test for relevance. Natural means the original writer of the content, or the webmaster posted the link because it was useful to the reader. This is a very simple standard.
That said, links that survive an unnatural link review or Penguin appear to be 'natural' - that is, they are related to content or theme in some way, supported by content, and not obviously placed merely to pass PR. I would use the following information to determine what to remove:
Assuming that the link passes index and suppression tests:
Directory links are acceptable if the anchor is not a valuable keyword.
Links that redirect to a client's site are ok, unless the source domain is penalized
Permit legitimate directory links (not using valuable anchors)
Permit iframe links - check iframe source
Ignore nofollow links
Ignore image links