Jennifer Slegg challenged Gary Illyes with some excellent questions at Pubcon Vegas this year. Of all of the Google employee AMA’s I have attended, this was the most valuable. I want to thank Gary for giving us such helpful answers.
I was personally in the session and live tweeted the whole thing. Because there was so much helpful information I decided to write it all down in a blog post. Wherever possible I did my best to quote Gary verbatim (as indicated by quotation marks). For those who are interested, we also added our thoughts, interpretations and links to helpful sources.
Nofollowed link changes (rel=sponsored and ugc)
Google’s blog post announcing nofollow changes
Is there any benefit to the site owner to implement rel-sponsored and ugc?
Gary: No. It is to help Google understand the web.
Gary said, "It is up to the ranking teams to decide how they want to use nofollows". In the past, any link signals surrounding a nofollowed link were not allowed to be used in ranking algorithms. Now, "it is up to the ranking teams to decide how they want to use nofollows". This change takes effect as of March 1, 2020.
Won’t the changes to nofollow encourage link spam?
Note from MHC: Now that Google treats nofollow as a hint, some people have theorized that spammers could go full barrel with spamming nofollowed links. The idea is that Google could start counting some of those, so why not make as many as you can? We believe that Google will indeed start using the signals from some nofollowed links. However, in our opinion, they will likely only do this when there are links that appear to be true recommendations for a site that come from authoritative sites.
Gary answered that Google does not think the nofollow changes will lead to link spam working.
Links and disavowing
Will there be any changes to Penguin soon?
Gary: Penguin is just working as it should. We've had a few launches since we last spoke about Penguin extensively. But they were "boring". Gary said that there have been some patches made to the Penguin algorithm, but for the most part the Penguin algorithm “runs on autopilot”. He also said that some algorithms, such as core updates are constantly being worked on.
Should sites invest in link audits?
Gary: "If you know you had a shady SEO company or you were buying links, it's probably best to do." He did also say, however, that just randomly disavowing spam links you find will not be likely to improve your rankings.
Note from MHC: John Mueller confirmed that disavowing unnatural links can potentially help a site algorithmically. We recently wrote an article with some disavow case studies that we are really excited about. If you have been making links rather than focusing on earning links, there is a good chance that those links made for SEO reasons are keeping your site from ranking at their full potential.
How often do site owners hurt themselves with the disavow tool?
Gary: “It's often enough that if it were me I'd remove the disavow tool. If you don't know what you are doing you can shoot yourself in the foot.”
Do we need to worry about links not showing up in GSC for link auditing?
Gary:”What we like to believe is that those links we show there are enough to get out of a manual action." He also cautioned though that what you get in GSC is a sample of your links.
Note from MHC: We recommend gathering links from several sources when auditing. We have seen manual actions where Google has given us example links that are not in GSC. Also, Google can still see links that are not in GSC. They have never said that those links are not used in their algorithms. With that said, we have decided that for some sites with massive link audits we may start focusing primarily on GSC links in our auditing process as doing a manual link by link audit of tens of thousands of links can take a very long time and be quite costly. If we end up seeing positive results after just disavowing GSC links, we’ll report on that in the future.
If someone changes a link so it's now pointing to a new page, would Google consider it still as valuable?
Gary: “I really don't want to answer this. If I give a non-PR answer I'll hear from the people who buy links.” He said that if you change links, it can change the relevancy. Google would reconsider how they use that link in ranking algorithms "and the weight will very likely change."
E-A-T and Core Updates
Here is our document on E-A-T.
Google’s advice on core updates.
Is there an E-A-T score?
Gary: “There's no internal EAT score or YMYL score. The Quality Raters’ Guidelines are guidelines for raters. EAT and YMYL are concepts that allow humans to dumb down algorithms. There is no one algo that looks for YMYL. He said that Google has “a collection of millions of tiny algorithms that work in unison to spit out a ranking score. Many of those baby algorithms look for signals in pages or content. When you put them together in certain ways, they can be conceptualized as YMYL. It's not like we have a YMYL score though.”
He also said that “Multiple algorithms conceptualize E-A-T”
Does Google recognize authors on sites?
Gary: “If you are asking whether Google does it, I think yes but for Google News." In Google scholar they have the concept of author. In web search, "we have entities for very popular authors like if you were an executive [inaudible] for the Washington Post, then you probably have an entity. It's not about the author, it's about the entity.”
Note from MHC: We have given advice in the past that every important article on your site should have an author bio. We do not believe that Gary is saying that author bios are not necessary. Extolling your authors’ E-A-T can often help users to trust that the content they are reading is accurate and good.
Do all changes to the algorithm get run past quality raters?
Gary: We tend to do tests for most of the changes." Core algo changes "kind of have to be tested". But a launch of nofollow changes would not. Gary went on to say that when he has changes on his projects that he wants to run by the Quality Raters, he will send them a fixed set of queries and it will generally take them a few days to get results back to him.
Does Google look for a site’s about page or contact page? Would not having those be hurtful to a site?
Gary: The Knowledge Graph is most likely looking for those, especially with larger entities.
Can people increase their E-A-T for their site and increase rankings?
Gary: It depends because there are way too many things that can come into play. He added, “Increasing your EAT is probably a good thing if you are in the YMYL space.”
Note from MHC: The next question you are likely asking right now is how you can “increase your E-A-T.” When we try to do this for our clients, what we do is comb through Google’s Quality Raters’ Guidelines to see if we can find things that are identified as a sign of high or low quality that can be changed on our client’s site to make it more trustworthy or to better demonstrate expertise and authority. Gary has said in the past that E-A-T relies heavily on links and mentions as well. As such, gaining good links and recommendations on authoritative websites can increase your E-A-T as well.
Does crawling increase just prior to an algorithm update?
Gary: “It depends. We have millions of baby algorithms and they might trigger something that increases crawl rate on certain sites, but it’s not a blanket thing."
Is Google still updating Panda?
Gary could not recall any update in the last couple of quarters. He said that possibly earlier in the year Panda updated, but he can't remember.
Information specific to medical sites
Were medical sites targeted specifically by recent updates?
Gary: “I don’t know. I’m leaning towards saying they were not targeted. But, it was known that they will be affected."
Does Google use consensus for things like conspiracy theories and things like "baby oil cures cancer"? How does G recognize that?
Gary: “This is an area where I specifically don't want to give too much information. Some of our algorithms use entities. Entities translate to terms. Certain algos can use medical terms. Anything from Project Owl can rely on that. For me, medical terms are close to my heart...Those who provide medical information and they are highly authoritative they know how to do, [mumbled]."
Unfortunately, it was challenging for me to hear the exact words Gary used at the end of this sentence. From what I understood, he was saying that Google only wants to recommend medical content that is highly authoritative.
Other interesting quotes
Can having emojis in your title tag hurt a site?
Gary: “I don't know. The algo that chooses the title will be more open to rewriting titles if it considers it less readable.”
In other words, if your title tag has emojis in it, Google may decide to just write you a new one.
How should we handle content on our site that is evergreen, but potentially outdated
Gary: “If you're writing about dinosaurs, there's not new news in that area, so you don't need to update it." He mentioned a [something] and date algorithm. Unfortunately I could not hear what the first word was.
How long does it take for manual actions to expire
Gary: They can expire as quickly as one month. Some can last two years.
This was a great part of the Q&A. Jennifer asked Gary to give very quick answers on whether the following were ranking factors.
Is domain age a ranking factor?
Gary: The age itself doesn't matter, but the signals collected over the years might matter. When Perficient Digital bought Stone Temple, the redirect would likely be helpful as the topic is similar. But, otherwise, he said, “we likely will drop most of the signals.”
Is having too many ads a negative ranking factor?
Gary: "I will go with yes...The Page Layout algorithm is still live and still working."
Can keywords in domain names help a site rank better?
Gary: I think it can help a little but not for the reason that spammers think." They can help in the sense that people will understand the site better. You're more likely to find them ranking well for low query terms.”
Is CTR a ranking factor?
Gary: “For personalization.”
Is ranking out to authoritative sites like Wikipedia a ranking factor?
Gary: “Reference your sources. If you are writing about how carrot juice can cure cancer, you sure as hell better have a resource for that with extensive information."
Is accessibility a ranking factor?
Gary said that no, it isn’t. However, he said that some things that are done for accessibility such as having descriptive alt attributes can help contribute towards rankings.
Can unlinked brand mentions benefit a site?
Gary: "Indirect. A better understanding of your brand."
Note from MHC: Gary is likely referring to entities here. If your brand is mentioned in an authoritative place, Google can likely recognize that even without a link.
Is word count a ranking factor?
Gary gave an example of "how to boil an egg". That doesn't need a lot of words. There may be more you can say about something, (like which chickens give the best eggs) but that doesn't make it more relevant.
Is content accuracy a ranking factor?
Gary: “YMYL so yes. We go to great lengths to surface reputable and trustworthy sources, so yes."
Again, I want to thank Jennifer Slegg for this great AMA and especially thank Gary Illyes for giving such helpful answers. I’ll be discussing my thoughts on many of these points in our next podcast episode. I’d love to hear your comments below as well!
If you like this, you'll love our newsletter! You can sign up below.
Google update newsletter
Want an update when Google makes a big algorithm change or other announcement? Sign up here!
It’s interesting that Gary contradicts himself a little bit about EAT. First saying they don’t have any internal EAT scores, then a couple of questions later saying if you increase your EAT its a good thing.
Has Googles algorithm learned and developed so much over the past decade that it now essentially runs the show? Have they lost a firm grip on it to where they would have to do something drastic to get it back under their control?
May sound like tin foil hat stuff but it just seems like nobody from Google can agree on some answers and they are just as confused as we are.
Gary said there was no single E-A-T score, but that the concept of E-A-T is something that is used by many of Google’s algorithms. I don’t think he contradicted himself at all.
It’s not like there is a single score for E-A-T.
Thanks Marie for really detailed coverage of Gary’s QA session. Some thoughts:
1. Nofollow as a hint is really going to make things spammy for webmasters but for Google it might not be a worry point. However, a great move by Google because sometimes it is hard earned link but still nofollow because of the website policy over any outgoing links
2. Link audit, I believe is best to be run with an aid of popular tools like ahref or majestic, specially for big sites as GSC sample links is a small set
3. EAT is a thing now – Would love if you can publish a check point list which webmasters can follow and see if there is something webmasters can work themselves
Thanks once again.
Thanks for these thoughts. Here are my ideas on your points.
1. I don’t think the nofollow changes will change spam. If Google starts counting the signals in some nofollowed links, they’re not going to be from blog comments and other easy to make links. Rather, if an authoritative publication that Google trusts in terms of linking out links to you with a nofollowed link, that could start to count.
2. We definitely do use ahrefs, majestic and Moz along with GSC links for link auditing. However, for some sites with massive link profiles if we are going to manually check each link it’s just not possible to use all sources. We’re going to experiment with using just GSC links for some sites and see if we can get equally good results.
3. We’ve published loads of things on E-A-T. If you’re looking for a checklist, there are many things in our book on E-A-T. (It’s soon going to be updated as well.) To be fair though, a full assessment of a site’s E-A-T generally takes my team and I a good week or more. It’s not a simple task!
Thanks Marie for this summary of the conversation.
You’ve given a lot of important information here from Garry.
The most significant information for the YMYL industry is this information
“Google only wants to recommend medical content that is highly authoritative ..”
Did Garry mention anything about how we can present Google more that we’re trustworthy?
This really comes back to E-A-T in my opinion, most particularly the A. Authority is closely tied to links. But, keep in mind that Google is good at determining which mentions on the web truly are there because an authoritative site is recommending you and which are self made. Other things that could help include having a Wikipedia page, having news mentions, having a good reputation, etc.
You can read more on this here: