At last week’s SMX East conference, Google’s webmaster trends analyst Gary Illyes took questions from the dual moderators — Barry Schwartz and Michelle Robbins — as well as from the audience in a session called “Ask Me Anything.”
In this post, I will cover that question-and-answer dialogue, though what you’ll see below are paraphrases rather than exact quotes. I have grouped the questions and used section headers to help improve the flow and readability.
Off-site signals
Barry: You’ve been saying recently that Google looks at other offsite signals, in addition to links, and some of this sounded like Google is doing some form of sentiment analysis.
Gary: I did not say that Google did sentiment analysis, but others assumed that was what I meant. What I was attempting to explain is that how people perceive your site will affect your business, but will not necessarily affect how Google ranks your site. Mentions on third-party sites, however, might help you, because Google looks to them to get a better idea what your site is about and get keyword context. And that, in turn, might help you rank for more keywords.
Imagine the Google ranking algo is more like a human. If a human sees a lot of brand mentions, they will remember that, and the context in which they saw them. As a result, they may associate that brand with something that they didn’t before. That can happen with the Google algorithm as well.
Mobile First, AMP, PWAs and such
Michelle: Where should SEOs focus their efforts in 2018?
Gary: If you are not mobile-friendly, then address that. That said, I believe the fear of the mobile-first index will be much greater than the actual impact in the end.
Michelle: When will mobile-first roll out?
Gary: Google doesn’t have a fixed timeline, but I can say that we have moved some sites over to it already. We are still monitoring those sites to make sure that we are not harming them inadvertently. Our team is working really hard to move over sites that are ready to the mobile-first index, but I don’t want to give a timeline because I’m not good at it. It will probably take years, and even then, will probably not be 100 percent converted.
The mobile-first index as a phrase is a new thing, but we have been telling developers to go mobile for seven years. If you have a responsive site, you are pretty much set. But if you have a mobile site, you need to check for content parity and structured data parity between your desktop and mobile pages. You should also check for hreflang tags, and that you’ve also moved all media and images over.
Michelle: Where does AMP fit? Is AMP separate from mobile-first? Is the only AMP benefit the increased site speed?
Gary: Yes, this is correct. AMP is an alternate version of the site. If you have a desktop site, and no mobile site, but do have an AMP site, we will still index the desktop site.
Michelle: If half a site is a progressive web app (PWA), and half is responsive, how does that impact search performance?
Gary: PWAs are JavaScript apps. If they can render, they will do pretty much the same as the responsive site. However, we are currently using Chrome Version 41 for rendering, and that’s not the latest, so there are newer APIs not supported by V41. If you’re are using those APIs, you may have a problem. Google is working to get to the latest version of Chrome for rendering, which will solve that issue.
Barry: I’ve seen desktop search showing one result and a mobile device showing a different page as an AMP result.
Gary: This happens because of our emphasis on indexing mobile-friendly sites. AMP is an alternate version of the regular mobile page. First, the mobile page gets selected to be ranked. Then the AMP page gets swapped in.
Michelle: So that means AMP is inconsequential in ranking?
Gary: Yes.
Michelle: Will there be a penalty for spamming news carousels?
Gary: We get that question a lot. I do not support most penalties. I (and many others at Google) would like to have algorithms that ignore those things [like spam] and eliminate the benefit. I’ve spoken with the Top Stories team about this, and they are looking into a solution.
Michelle: What about progressive web apps (PWAs)? Do they get the same treatment as AMP, i.e., no ranking boost?
Gary: If you have a standalone app, it will show up in the mobile-first index. But if you have both a PWA and an AMP page, the AMP page will be shown.
Michelle: What if the elements removed from your mobile-first site are ads? [Would that make the AMP version rank higher?]
Gary: Your site will become faster [by adopting AMP and eliminating these ads]. The “above the fold” algorithm looks at how many ads there are, and if it sees too many, it may not let your site rank as highly as it otherwise might. But when we’re looking at whether sites are ready for the mobile-first index, we’re more concerned about parity regarding content, annotations and structured data than ads.
Michelle: What about author markup?
Gary: Because AMP pages on a media site can show up in the news carousel, the AMP team said that you shouldn’t remove the author info when you’re creating AMP pages.
Search Console
Barry: When will SEOs be able to see voice search query information in Search Console?
Gary: I have no update on that. I’m waiting for the search team leads to take action on it.
Barry: How is the Search Console beta going?
Gary: It’s going well. There are a significant number of sites in the beta. We’re getting good feedback and making changes. We want to launch something that works really well. I’m not going to predict when it will come out of beta.
Barry: When will they get a year’s worth of data?
Gary: They have started collecting the data. Not sure if it will launch. The original plan was to launch with the new UI. [Gary doesn’t know if plans have changed, or when the new UI will launch.]
Barry: Why is there no Featured Snippet data in Search Console? You built it, tested it, and then didn’t launch it.
Gary: There is internal resistance at Google. The internal team leads want to know how it would be useful to publishers. How would publishers use it?
Barry: It would give us info on voice search.
Gary: I need something to work with to argue for it (to persuade the team leads internally at Google that it would be a good thing to release).
This question about how the featured snippet data would be used was then sent to the audience.
Eric Enge (your author) spoke from the audience: I’d like to use the data to show clients just how real the move to voice search is. There are things they need to do to get ready, such as understand how interactions with their customers will change.
Michelle: So, that data could be used to drive adoption. For now, that sounds like more of a strategic insight than immediately actionable information.
Gary: The problem is that voice search has been here for a couple of years. Voice search is currently optimized for what we have, and people shouldn’t need to change anything about their sites. Maybe there will be new technologies in the future that will help users.
Michelle: I think that it’s more complicated than that. There are things that you can do with your content that will help it surface better in search, and brands can invest resources in structuring content that can handle conversations better.
Ads on Google and the user experience
Michelle: As you (Google) push organic results below the fold [to give more prominence to ads and carousels] … is that a good user experience?
Gary: I click on a lot of search ads. (Note that Googler clicks that occur on our internal network don’t count as clicks for advertisers, so this costs you nothing.)
I believe that ads in search are more relevant than the 10 blue links. On every search page, there’s pretty aggressive bidding going on for every single position. Since bids correlate to relevance and the quality of the site, this does tend to result in relevant results
Barry: Sometimes the ads are more relevant than the organic results …?
Gary: Especially on international searches.
Michelle: How is that determined?
Gary: This is done algorithmically.
Michelle: How can you compare ads to organic if the two aren’t working together?
Gary: The concept of a bidding process and the evaluation of quality are used by both sides. The separation between the groups is more about keeping the ads people who talk to clients away from the organic people, so they don’t try to influence them. The ads engineering people, they can talk to the organic side; that’s not forbidden.
Ranking factors and featured snippets
Michelle: Does Google factor non-search traffic into rankings?
Gary: First of all, search traffic is not something we use in rankings. As for other kinds of traffic, Google might see that through Analytics, but I swear we do not use Analytics data for search rankings. We also have data from Chrome, but Chrome is insanely noisy.
I actually evaluated the potential for using that data but couldn’t determine how it could be effectively used in ranking.
Barry: What about indirect signals from search traffic, such as pogosticking? Previously, Google has said that they do not use that directly for ranking.
Gary: Yes, we use it only for QA of our ranking algorithms.
Barry: At one point, Jeff Dean said that Google does use them.
Gary: I do not know what he was talking about. The RankBrain team is using a lot of different data sources. There was a long internal email thread on this topic, but I was never able to get the bottom of it.
Michelle: Is RankBrain used to validate featured snippets?
Gary: RankBrain is a generic ranking algorithm which focuses on the 10 blue links. It tries to predict what results will work better based on historical query results. The featured snippets team uses their own result algorithm to generate a good result. I have not looked into what that means on their side. RankBrain is not involved, except that it will evaluate the related blue link.
Barry: Featured snippets themselves are fascinating. You said that they are changing constantly. Please explain.
Gary: The context for that discussion was about future developments for featured snippets. The team is working around the clock to improve their relevancy. The codebase underlying it is constantly changing.
Michelle: Does the device being used by the searcher factor in?
Gary: I don’t think so.
Schema and markup
Gary: I want to live in a world where schema is not that important, but currently, we need it. If a team at Google recommends it, you probably should make use of it, as schema helps us understand the content on the page, and it is used in certain search features (but not in rankings algorithms).
Michelle: Why do you want to be less reliant on it?
Gary: I’m with Sergey and Larry on this. Google should have algorithms that can figure out things without needing schema, and there really should not be a need for penalties.
Michelle: Schema is being used as training data?
Gary: No, it’s being used for rich snippets.
Michelle: Eventually the algo will not need the schema?
Gary: I hope so. The algorithms should not need the extra data.
Barry: Is there a team actively working on that?
Gary: Indirectly, absolutely. It probably involves some sort of machine learning, and if so, it’s the Brain team that works on it. I do not know if they have an active project for that.
Barry: How did you get entity data in the past?
Gary: From Freebase and the Knowledge Graph.
Panda and thin content
Barry: You said that pruning content was a bad idea. If you’re hit by Panda, how do people proceed?
Gary: Panda is part of our core ranking algorithm. I don’t think that anyone in a responsible position at Google thinks of Panda as a penalty. It’s very similar to other parts of the algorithm. It’s a ranking algorithm. If you do something to attempt to rank higher than you should, it basically tries to remove the advantage you got, but not punish you.
Ultimately, you want to have a great site that people love. That is what Google is looking for, and our users look for that, as well. If users leave comments or mention your site on their site and things like that, that will help your ranking.
Pruning does not help with Panda. It’s very likely that you did not get Pandalyzed because of your low-quality content. It’s more about ensuring the content that is actually ranking doesn’t rank higher than it should.
Barry: Pruning bad content is advice that SEOs have been giving for a long time to try and help people deal with Panda.
Gary: I do not think that would ever have worked. It definitely does not work with the current version of the core algorithm, and it may just bring your traffic farther down. Panda basically disregards things you do to rank artificially. You should spend resources on improving content instead, but if you don’t have the means to do that, maybe remove it instead.
Using disavow
Michelle: Should you use disavow on the bad links to your site?
Gary: I have a site that gets 100,000 visits every two weeks. I haven’t looked at the links to it for two years, even though I’ve been told that it has some porn site links. I’m fine with that. I don’t use the disavow file. Don’t overuse it. It is a big gun.
Overusing it can destroy your rankings in a matter of hours. Don’t be afraid of sites that you don’t know. There’s no way you can know them all. If they have content, and they are not spammy, why would you disavow them?
Sites like this are very unlikely to hurt you, and they may help you. I personally trust the Google filters.
Barry: Penguin just ignores the links.
Gary: Penguin does that, too (Gary’s phrase implies that there other algorithms that might filter bad links out, as well).
Some opinions expressed in this article may be those of a guest author and not necessarily Search Engine Land. Staff authors are listed here.
