Good afternoon/evening from beautiful Seattle, where day one of our sold-out SMX Advanced conference is wrapping up with the traditional Google keynote conversation.
Tonight, our founding editor, Danny Sullivan, will be speaking for about an hour with Gary Illyes, Webmaster Trends Analyst at Google. We’re expecting them to cover a lot of the recent hot-button SEO topics — like RankBrain and keyword-based top-level domains — as well as some long-running industry concerns like the Penguin algorithm and more.
The conversation is due to start at about 5:00 p.m. PT, so feel free to come back then and refresh this article page for our live blog coverage.
Okay, we’re about to begin. As always, I’ll use “DS” for what Danny Sullivan says and “GI” for what Gary Illyes says. And I’ll try to insert some good tweets from our audience, too. Follow along as we get going!
DS: I thought we’d start off with something easy, so could you just explain everything about RankBrain to us. (Laughter) Seriously, is it really a ranking factor?
GI: If this is easy, I’m in trouble. I will stick to what we said — basically, it’s a ranking factor. It’s part of machine learning. (Asks who knows what machine learning is.) It’s something that tries to identify patterns and bucket data. It looks at data about past searches and based on what worked well for those searches, it will try to predict what will work best for a certain query. This works best for long-tail queries and queries we’ve never seen.
An example might be “can i beat Mario Bros without using a walkthrough.” Without RankBrain, we give interesting results that don’t meet my needs. But with RankBrain, we can give results that satisfy my question.
RankBrain will understand better what results work for queries. It’ll understand that certain stop words should not be dropped. Sometimes the word “with” is dropped from a query, but RankBrain will understand that we need to keep it.
GI: It’s less about understanding. It’s more about understanding how to score the results.
DS: Is there a RankBrain score?
GI: You don’t have a score. I think the root of your question is whether you can optimize for RankBrain — (Laughter)
DS: All of my questions will be about optimizing to some degree. (Laughter)
DS: Does RankBrain leverage your existing signals, or is it a new signal?
GI: It’s a new signal. But the reason I asked about optimizing for RankBrain is because you don’t. It’s about making sure the user gets the result that is deserved for the query. If you write in natural language, you’re all set. If you keyword-stuff your content, that will almost certainly not be good for you.
DS: Last year, you said RankBrain handles 15 percent of the queries. What’s it doing now (in terms of how many queries)?
GI: I have no idea.
DS: Last year, you said RankBrain was the third factor, and then you eventually said content and links are the top two. But which one is first?
GI: Seriously? (Laughter)
GI: The order depends on the query. I can’t give you a concrete answer because it depends on too many things.
DS: What’s the deal with Google assistant?
GI: Frankly, I have no idea. I know we’re still trying to wrap our heads around what it should look like. We like to experiment with these cool ideas. (Missed some discussion on machine learning.) Who has seen the movie, “Her?” I think it’s freakishly creepy, but also a good example of what you could do with an assistant.
DS: Asks about how Google looks at keywords in domain names, specifically whether Google looks at keyword-based TLDs….
GI: TLDs do not play a role in how we calculate relevancy for a specific piece of content or a specific URL. Country TLDs can play a role in queries in specific countries. But TLDs like “attorney” and “news” don’t play a role.
DS: Do you look at the domain name at all?
GI: There could be certain cases where we look at it, but in most cases, no. I would not try to buy domain names that are keyword-filled. Going for keyword-rich TLDs … that’s just weird. Don’t do that.
DS: In 2013, Google said you’d extend the amount of data in Search Console beyond 90 days. Can we just get a year’s worth of data?
GI: We’re looking at 91 days of data now. (Laughter) We are still looking at how we can do it. We heard the feedback from SEOs at the Google Dance earlier this year.
DS: In May, you started bringing Search Console data into Google Analytics. Will that be archived longer than 90 days?
GI: I would assume it will be the same.
DS: We would like that to be much longer. (Laughter)
DS: Let’s move on to Penguin real quick. Last update was in December 2014. Will we get another update before Elon Musk lands on Mars?
GI: I won’t say a date because I was wrong too many times, and it’s not good for business, I’ve heard.
DS: Some time this year?
GI: I won’t say any time frame anymore.
DS: What’s up with Panda? You said it was part of the core ranking algorithm… (Danny goes on to ask a very long question about how it’s run.)
GI: It’s not real-time. It runs and takes months.
DS: (Asks about secure search)
GI: I think about 30 percent of pages are using HTTPS. We’re looking to boost secure as a ranking factor, but we don’t want to do it too soon.
DS: Mobilegeddon 2 was in May. How’d that go? When’s the next one?
GI: I so hate this “mobilegeddon” thing.
DS: It’s a good name.
GI: No, it’s not.
DS: Just name it yourself. Just don’t call it something stupid like the “mobile search update.”
DS: Can you give me an update on social signals. Is it still the case that you don’t look at Facebook likes or Twitter retweets?
GI: It is. We have a problem with social signals because we don’t want to rely on something that someone could pull the plug on.
(Missed question about Google+.)
GI: By the way, we are not using authorship anymore.
DS: At all?
GI: At all.
DS: Can we get a breakdown on voice search vs. typing a search? Can we get that data?
GI: You could. Possibly. I’m trying to decide if we’re already working on it or trying to figure out how to do it.
(And now we move on to the audience questions.)
GI: We use clicks for very specific things, like personalization. If you first search for “apple,” we may not know if you mean the fruit or company. If you regularly click on pages about the company, we’ll learn that’s what you’re interested in.
GI: These click experiments that people are running sometimes interact with our own click experiments, which is not nice. It’s really annoying.
DS: (Audience question) Should we move to HTTPS2?
GI: It’s still newish. I would look at what it would take to implement this, but only do it if you can gracefully downgrade back to HTTPS. We’re working on making sure Googlebot handles it correctly, but there are browsers that cannot handle it. You could be shooting yourself in the foot if you cannot fall back to HTTPS.
Should you switch to HTTPS? That depends on how much you care about your users and the integrity of your website.
DS: Why isn’t my site fully indexed when I use HTML and XML sitemaps?
GI: When people ask me about this, the most common reasons their pages aren’t getting indexed are the noindex tag, robots.txt, rel=canonical or cloaking.
DS: Do you have different algorithms for different industries?
GI: For Universal search, we have “query deserves images” and things like that. But no, we do not have different algorithms for different industries.
DS: Last year you stopped displaying emoji in the search results, but they still work in Google News. Can we have emojis back in title tags?
DS: What’s your final advice you want everyone to know?
GI: Two things. Pay attention to AMP. It’s going to be really big. Figure out with your developers how to implement it. Second, look at developments around assistants and chat bots. They are going to be huge, and you want to be among the first people who get on those features.