Google’s Disavow Conundrum -SEO Guide

What is the best way to identify links that should be disavowed? Well, begin with your own quality test for the linking site:

1. Does the site seem to legitimately represent someone’s voice?
2. Does the linking article provide good information regardless of how well it is written?
3. Does the linking article present what appears to be a valid opinion?
4. Do the links in the article support its points or explain vague references?
5. Does the linking site publish more than 1 article?

If you can answer “YES” to all the above my first inclination is to leave the links alone. It may not be a high quality site but that doesn’t mean it’s spam, it doesn’t mean it’s toxic, and it doesn’t mean it’s hurting your (client) site. These are the kinds of questions you should ask yourself, but come up with your own list that represents your own values.

Next, apply a toxicity test for the linking site:

1. Is it a hastily made site that obviously won’t ever be updated?
2. Do the links only use targeted competitive keyword-rich anchor text?
3. Is the site similar in design, content, tone, and style to others you find in the backlink profile?
4. Do you know that the link was paid for?
5. If you scan the backlinks for the linking site, do they look even more suspicious?

If you can answer “YES” to all of the above then my first inclination is to assume the links are spammy, easily identified, and probably toxic. These are probably common sense talking points for anyone with a few years’ or more SEO experience. Voluminous link spam looks very, very bad. People still use it but mostly to propel their “Web 2.0 links” into Bing and Google’s good graces (they hope). They use the Web 2.0 sites for “link whitewashing” or “link laundering”.

These links do (sometimes) appear to still work. I know because we have a reputation management client whose search results went to hell when his hostile entity spammed Google with a lot of crappy links. Those crappy links don’t always meet the above the criteria but they come pretty close.

There is what I would call a Character Test for the linking content.

• Does it look out of place?
• Does it feel like it was obviously created for only a specific purpose?

You’re more likely to encounter obvious link spam that fails the Character Test in a hostile reputation management campaign.

Call that Negative SEO, if you will. It’s just dirty, rotten content that would not normally be where it is if someone were not trying to torpedo someone else (or a Website). Someone once linked to my personal domain (Michael-Martinez.Com) with 30,000 links accusing me of being a pedophile and bestiophile (via their link anchors). These links were in forum comments, social media profile pages, and other easily spammed content.

Honest links may be vitriolic but the vitriol tends to be of a different nature. “He’s a lying jerk” is a more honest link anchor than “this pedophile child porn”, in my opinion.

…if you’re discussing the risks of “nofollow” links with clients or business decision-makers, consider the following order of risk severity:

1. Known spam links to known link spam recipient
2. Known spam links to UNknown link spam recipient (not yet caught)
3. Unknown spam (passes filters) links to known link spam recipient
4. Unknown spam links to UNknown link spam recipient (not yet caught)
5. [Some spam] links to non-spam link recipient [INNOCENT]
6. Non-spam links to known link spam recipient [- probably no harm -]
7. Non-spam links to UNknown link spam recipient (not yet caught)
8. Non-spam links to non-spam link recipient

Assuming these are all “nofollow” links, none of them should have any immediate, direct effect on the receiving Websites. But in the first scenario Google may be counting noses and taking names. Otherwise, I don’t think there is (yet) any risk from receiving “nofollow” links. It should be very rare, although obviously any site that is caught passing or receiving spammy links (even if they use “nofollow”) will move up in the risk scale.

Key takeaways from part one include:

1. Penguin is looking at the linking site (quality signals associated with the linking page) not the site receiving links
2. Google annotates links with “labels” such as “footer link”, “Penguin real-time”, “disavowed link”, and many more
3. Gary Illyes pronounced his name as “(e)+ISH” and Danny Sullivan pronounced it as “ee-ish”
4. Google prefers to ignore a modest number of spammy links through the algorithm
5. If they detect (maybe 70%) spammy links the Webspam team may apply a manual penalty for egregious behavior
6. Penguin 4.0 looks at other types of spam (on the source site) than just links
7. By the time the podcast was recorded only one data center had not yet picked up the new Penguin algorithm

Gary mentioned a negative case study that had been submitted to Google for evaluation. He said that Penguin was devaluing the links and the site was not being hurt by the links. In an unrelated case study, another site received a manual action because of a negative SEO campaign. The company was able to quickly disavow the links and get the penalty lifted. I think Penguin may make it easier for people to see when negative SEO is happening because of the manual actions. Of course, someone might conduct multiple consecutive negative SEO campaigns against you to build up a history of penalties with Google. Gary maintains that they still have not seen negative SEO “working as people think that they should work”.

…

Key takeaways from part two include:

1. Most people cannot identify the good directories
2. Keep your directory links to a small percentage
3. Legitimate business directories should be okay
4. Machine learning will not take over the core algorithm
5. Expect more launches around Google AMP and structured data
6. Click through rates are “drastically lower” to AMP pages
7. Larger, well known sites may see better CTRs for their AMP pages than small sites
8. Machine learning is used to identify new signals and signal aggregations
9. RankBrain “reranks based on historical signals” that “comes up with a new multiplier”
10. Panda “measures the quality of a site by looking at a majority of the pages”
11. Panda only demotes sites; it cannot promote sites (negative signal only)
12. Panda concludes the site is trying to game the system
13. Small business sites tend to adapt to new guidelines and signals faster than large business sites
14. Technical SEO tools like structured data and AMP don’t improve rankings but may get you into more widgets
15. The “HTTPS signal” just looks at the first 5 characters in the URL

For example, a trustworthy link may be:

1. Given without incentive
2. Part of critical on-site navigation
3. Using unhelpful anchor text
4. Rare

An untrustworthy link may:

1. Only be given with incentive
2. Be superfluous or unnecessary
3. Use targeted anchor text
4. Be common (sitewide or found on many sites)

So here is how I evaluate links for naturality:

1. Would I normally include a link like this on my personal Websites?
2. Would I normally place a link like this at this spot on the page?
3. Would I normally place this link next to this kind of content?
4. Is this the kind of content I would normally create for myself or my clients?
5. Does this content appear to have a non-manipulating purpose?

Option 2: Instead, let’s look at some acid tests you can apply without having to crawl the Web, build spreadsheets, etc.

1. How prominently is the linking page featured on the Website? If the Website doesn’t make the page seem important then that is a red flag.
2. Does the page get a lot of social media attention beyond self-promotion? If you can’t find gratuitous social media activity for the page then that is a red flag.
3. Does the page seem exceptional compared to other content on the site? Are the articles random in nature or is it just one out of many articles about how to bake cakes? Content that is largely diverse is a red flag.
4. How much care was given to the details of the site design? If the site owner made little to no effort to give the design some personality then that is a red flag.
5. How many links are embedded in the page content? If it’s just an article and you see 2-3 offsite links in every paragraph then that is a red flag.

Low quality sites hurt the user experience. They are most likely to:

1. Obscure information (not just “content” but “information”)
2. Use vague language in content
3. Rely on duplicate information (not just “content” but “information”)
4. Contain a lot of page clutter (widgets, images, videos, ads, etc.)
5. Diffuse PageRank through flat site architecture
6. Use PageRank sculpting (adding “nofollow” to internal links)
7. Create orphans and weaken intrasite navigational support for “leaf” pages (deep content)
8. Publish content through hard-to-render Javascript
9. Provide less annotation for graphics
10. Use more boilerplate text than similar sites

Low quality (spammy) links are most likely to:

1. Be self-placed (directly or indirectly)
2. Use preferred (“targeted”) anchor text
3. Be limited to “relevant” context (natural links are most often embedded in IRrelevant contexts)
4. Be quickly made
5. Not support and reinforce useful content or information (in the linking context)
6. Be found in similar or (near-)duplicate content across multiple sites
7. Make less sense than normal in-context links would
8. Match query terms more than they match non-query phrases
9. Be placed in “guest” or “user-generated” content
10. Use shorter anchor text than normal links (based on context norms)

…

You are most likely to find self-placed links in:

1. Guest posts (body content or bio boxes)
2. User account profile pages
3. Social media shares
4. Blog comments
5. Forum discussion posts (in the post or in the signature)
6. Widgets
7. Attribution text (such as image or infographic captions)

…

Regardless of whether you are evaluating content or links, you should be asking these kinds of questions:

1. Why is this thing here?
2. How is this thing different from similar things on the page / site?
3. Why is this thing different from similar things on the page / site?
4. Why is this emphasis necessary?
5. To a disinterested reader, does this thing add real value to the page?
6. Would this thing be here if [SEO metric] was lower or zero?
7. Why does this site exist?
8. Would the page or site exist if the link were not there?
9. Whom would a disinterested reader conclude is responsible for the link?
10. Whom would a disinterested reader conclude is responsible for the content?