Arsen Rabinovich (00:00:00):
Here we go. We are live now. Hi, everybody. Welcome to episode 44 of SEO for Bloggers webinar. Today we have a special segment, and this is the first that we’re going to be doing. So it’s Top Hat Chats, where we speak to an expert about one specific topic, and we’ll usually focus on that one topic or a specific skill.
(00:00:32):
So today with us we have Dr. Marie Haynes. She’s a leading expert in EEAT, expertise, experience, authority and trust, Google’s quality raters guidelines, and algorithm updates. Through her work, Dr. Haynes has helped understand and adapt to Google’s evolving search ecosystem. She’s helped businesses and brands all over the world understand algorithms, and help them recover from updates. Today she’s going to be talking with us about the latest updates to quality rater guidelines from from Google.
(00:01:07):
And just like with all of our other webinars, if you do have a question, we will be answering them. Please type Q in chat before your question. And we have Melissa here with us. She’s not on camera, she’s in the background pressing all the buttons. She will make sure to surface that question for us. And our regular webinar with Andrew, Casey, and Melissa still happens quarterly. We’re going to have these Top Hat Chat episodes coming out to you hopefully once a month. And with that we’re going to start with some questions for Dr. Haynes. Hi, how are you?
Dr. Marie Haynes (00:01:45):
Hey, Arsen. I’m good. I’m looking forward to this. Anytime we get to talk about Google, it’s always a good chat.
Arsen Rabinovich (00:01:51):
Right, right. So I was on Twitter, or X, and I saw your tweets and I’m like, “I got to have her on. We got to check. We got to check in with her on what’s been happening.” And we have a very big community of bloggers who are paying attention to some of these changes. And just to get us started, for some people who might not be aware of what it is, can you give us a quick rundown on what the quality rater guidelines are and why they matter so much?
Dr. Marie Haynes (00:02:22):
Yeah, so these guidelines are published by Google. They’ve been published for many years. Do you remember when they used to be hidden, right? Nobody could find them, and then a leaked copy got out one year and people tried to publish it, and then Google, the legal team would go after them. And then one year Google was like, “Well actually, why wouldn’t we publish these guidelines? Because we want people to make the type of website that we want to reward.”
(00:02:48):
So the guidelines are written for these people called quality raters. We’ll talk about who they are in a minute. And they’re people who actually rate Google’s algorithms, who rate how the search results are. And they have to be given some sort of guidelines to say, “Well, this is good, this is not so good.” So when something changes in the guidelines, we pay attention to it, because it means that that’s something that Google wants the raters to focus on.
(00:03:16):
And there were several changes in the guidelines this time. The last few times they’ve changed, it’s just been minor spelling changes or punctuation, stuff like that. And this time there’s significant, significant change, and a lot of it applies to recipe bloggers. I know you have a lot of recipe bloggers here, and so we’ll talk about those. And if you remember, I’m going to test you, Harrison. Do you remember what update, okay, the rater guidelines changed in July of 2018, and the next month there was a significant update. Let me tell you first what the change was. They made one small change, it said, “Safety of users”. That was it. The only change. They updated the whole guidelines to add something about the safety of users. What happened August 1st?
Arsen Rabinovich (00:04:05):
Was that HTTPS?
Dr. Marie Haynes (00:04:05):
No. No.
Arsen Rabinovich (00:04:07):
No? No. Okay. I have no [inaudible 00:04:09]. Don’t put me on the spot.
Dr. Marie Haynes (00:04:11):
No, [inaudible 00:04:09], it’s okay. I know, I always do that and I always embarrass the host. I don’t know why I remember algorithm update dates. But it was Medic, do you remember Medic, the…?
Arsen Rabinovich (00:04:18):
Medic, sure.
Dr. Marie Haynes (00:04:19):
And that was the kind of EAT update, before the extra E was added. And they really were focusing on websites that were writing your money or your life content, like medical content with no medical expertise. And so when something changes in the guidelines, it means look out, because Google’s going to be making algorithmic changes to try and reward content that matches that.
Arsen Rabinovich (00:04:44):
Right. And that kind of leads us into the second question. This is constantly being asked. How do these guidelines compare to the actual ranking algorithm? Do they directly influence rankings, or is this something that we’re should be mindful of, but it’s not really a part of the algorithm?
Dr. Marie Haynes (00:05:01):
It’s funny because when I first became aware of these guidelines, I think I took a lot of flack, because a lot, if you’re in SEO, we know… We’re trained to optimize for this algorithm that’s a bunch of heuristics. It’s a bunch of rules. We know that Google looks for keywords in the title tag, or, they want fast pages. We can picture these things that they can measure, but a lot of the stuff that’s in the guidelines, we couldn’t conceive of how you can make an algorithm that would determine, like, this is high quality, or how do I know that this is an expert, and does Google look at every single author?
(00:05:37):
And we’d have all of these arguments and discussions about how could these possibly be algorithmically enforced. Then there’s a line in the guidelines that says, if you look for, “Helpful and unhelpful”, there’s a line that actually says that the quality raters are providing Google with examples of helpful and unhelpful results. And later on we learned more and more about how Google uses machine learning in their systems. And one of the most important things to machine learning is to see examples. And so Google needs to show their systems, “Here’s examples of what’s helpful. Here’s examples of what we would consider unhelpful.” Like, if there’s ads that cover over everything and make it impossible for the user to see.
(00:06:26):
And then what the machine learning systems do is they look for signals that they can measure, so they can try to replicate and predict. So just like if you were building a machine learning system to predict if an email was spam, you could show it examples of like, “This is spam.” You know, like, “Do you want to buy this vacation package?” Or, “This is a real email.” And if you saw enough of the, you could start predicting. Even though you didn’t see that exact email, you could predict the signals that kind of correlate with spam.
(00:07:02):
And so it’s the same thing when Google does a search, they try to predict which result is going to be the most helpful for people. And so the rater guidelines are one way that Google’s provided with these examples of helpful and unhelpful results. And then we also learned a ton of stuff in the DOJ versus Google trial. For anybody who’s in this webinar, if you have not read Pandu Nayak’s testimony… There’s two testimonies. One is Pandu Nayak, I can find you the links if you need them later. And then the other is a professor named Douglas Ord. And they both talk about how Google’s machine learning systems are used, and Douglas Ord talks about these two systems that are put in front of the quality raters. One set of, actually, let me back up a little bit. Quality rater… You okay if I just ramble like this, are you?
Arsen Rabinovich (00:07:59):
Yeah, go on. Yeah, this is what we’re doing.
Dr. Marie Haynes (00:08:02):
Okay, good stuff. Good stuff. So the quality raters, there’s 16,000 of them apparently, in the last documentation that Google provided for us. And they used to tell us that these are for, say an engineer wants to implement a change, has this idea for making the results better. The engineer would give the raters two sets of results. One would be “Here’s the current algorithm,” and “Here’s the algorithm with my changes involved.” And then the raters would go through every single result that’s in that search result page and rate it according to these guidelines. They have sliders in there on page quality, which is synonymous with EEAT, and also on whether the users’ needs are likely to be met. And so if more raters said that the new result was better, then that gives Google evidence that we should go ahead and implement it. And so… Not me, I don’t work for Google. That they should go ahead and implement it.
(00:09:00):
And so that made sense to me, but it really didn’t make sense to me that they had 16,000 people that are employed to monitor these changes. Like, how many engineers are at Google? I know there’s a fair number, but it just seemed like a lot of people. Well, in the Douglas Ord testimony of the DOJ versus Google trial, he talks about another way that changes can be introduced into the algorithm. And he talks about two versions. One is what’s called frozen, which is the results that exist today in Google’s algorithm, and the other is retrained results.
Arsen Rabinovich (00:09:34):
Right.
Dr. Marie Haynes (00:09:35):
And that was just a mind-blowing thing for me, that the retraining, they also talk in Pandu Nayak’s testimony about how machine learning systems are at the core of Google. There’s RankBrain, there’s one called DeepRank, there’s Rank-in-Bed BERT, and those systems are constantly being retrained. A lot of that retraining has to do with click data, which we can talk about that if you want. That’s probably a whole conversation on its own. But that retrained version is trying to predict what would be best for each particular query to put in front of a searcher. And they can’t just go and run with that. They need to, because machine learning, we all know that AI can make mistakes. And so they put those two versions of the algorithms in front of the quality raters, and then the raters are determining, I mean, they don’t know the individual changes, but overall, are the results from the retrained algorithm that’s learning and trying to improve, better than what existed before? So that’s kind of how they’re used, is to kind of keep in check the machine learning changes that Google makes.
Arsen Rabinovich (00:10:45):
So it sounds like this was, quality raters guidelines were kind of the first step in the evolution of what we now know as that Helpful Content System. They’re looking at content from a perspective of some of these guidelines.
Dr. Marie Haynes (00:11:02):
Yeah. Because when they were first published, I don’t think Google was using a whole lot of machine learning in their systems, then. They were using it for spelling correction, things like that. But Google themselves have said that they’ve been AI-first since 2017, which is super interesting that they haven’t been very public with their use of AI within Google search algorithms. I’m not talking about AI overviews, but to actually determine the ranking of results.
(00:11:31):
The other part of the Pandu Nayak testimony that blows my mind that I don’t see more people talking about this, is when they talk about RankBrain, so Google’s AI brain for ranking. And they say that Google puts together the traditional search results, like, they do it with traditional core algorithms that look at keywords that look at an inverted index and then even use PageRank, although PageRank has changed significantly over the years. And that narrows results down from trillions to hundreds, and then they take the top 20 to 30 results and RankBrain reranks them. Gives them a new ranking score. So really, it’s AI that, once you’re in the top couple of pages of search results, then what matters is that you look good to Google’s AI systems. And the AI systems are trying to replicate what is in the guidelines, basically.
(00:12:24):
So I don’t know, when did that shift happen, in 2017? I think they’ve been working towards this AI-generated system for quite some time, but now more than ever, when something’s in the guidelines, it’s something that we know that Google is trying to get their machine learning systems to replicate as if they were a human quality rater.
Arsen Rabinovich (00:12:46):
And we’ve been noticing that for many years now, and on both our enterprise side and our publisher side of services. I always preach, I’m like, we need to make sure we get into top 20. We got to follow best practices to get in. Once we’re in top 20, then we can slowly start adjusting the levels of relevancy and focus towards specific topics, to get to where we need to go. Because that’s a different shuffling that happens in top 20 or 30, whatever. We don’t know. It’s what we assume. I’m assuming top 20, but it’s a different shuffling that happens 21 and lower, right, or higher. So yeah, we see this, that it doesn’t… Just like with the human layer, the click data, right? Before, Google was like, “Oh, no”, but we see it. We noticed it, right?
Dr. Marie Haynes (00:13:35):
Yeah.
Arsen Rabinovich (00:13:36):
We had a question in here about who the quality raters are, and I know somebody in the comments said that they used to be a quality rater, and you kind of answered that. So I want to move on. I want to talk about the page quality, the PQ aspect of the update. So Google recently updated its stance on filler content, and this is very important for our audiences today, especially in recipe blogging. What does that mean for bloggers, and how should they adapt to it?
Dr. Marie Haynes (00:14:01):
Okay, so quality is the whole guidelines. The majority of the guidelines talk about quality. A few years ago they changed a lot of the places where it said, “EAT”. They actually just swapped out “EAT” and put “Page quality” in there, which is a little bit hard to grasp, because to me, before that I would think, “Well, quality is about the quality of the text, the content, is it good?” Maybe the images. And it is that, but the quality also has to do with the quality of your website, the reputation of your brand. That’s something we should probably talk about. Reputation is all throughout the guidelines. And so page quality, I’m looking at the guidelines now, it starts on page nine and goes like 92 pages of what page quality is.
(00:14:56):
So in terms of filler content, this is something that was just newly added in January of this year, they added some information on filler content. And they specifically called out recipes, and the example that they gave, there’s an example that you can go into the guidelines and I think it’s for buttery beer, something like that. It’s a Harry Potter thing. And the website that is chosen, it looks like a decent website. However, they talk about the fact that there’s filler content before you get to the main content. So there’s multiple sections of the guidelines that talk about what the main content is. The main content is the part of your page that fulfills the purpose of the page. So is it all recipe bloggers here, or do we have a-
Arsen Rabinovich (00:15:43):
No, no, we have a mix.
Dr. Marie Haynes (00:15:43):
We have a mix. Okay.
Arsen Rabinovich (00:15:44):
But I think a bigger chunk is recipe publishers.
Dr. Marie Haynes (00:15:47):
Okay. So I mean, it’s easy if we think about the main content on a recipe page. If I go to a recipe page, the vast majority of the time, the purpose of that page is to help me make that recipe. And so the main content is the part that fulfills that purpose, which primarily is the recipe card. It might be if you’ve got helpful steps in your text that are related to helping me make that recipe. And then we have all the jokes about reading about your grandmother’s grandmother in Italy who used to make this and the family made it for hundreds of years. That’s okay to have that kind of stuff, but that’s supplemental content.
(00:16:27):
And so what the guidelines say about main content is that it’s easy to find. It’s also in Google’s page experience guidelines. Like, page experience, we always talk about core web vitals, and it actually, one of the points in the page experience guidelines, “Is the main content easy to find?” And so this example that they give in the guidelines talks about this page that I can’t remember exactly what it starts off with, but there’s just a lot of stuff that readers would just skim over. It’s not relevant to the recipe. And then at some point they’re an affiliate blogger, so there’s also links to where you can go to Disney World or Universal Studios or wherever it is that the Harry Potter exhibit is, which has nothing to do with the recipe. Again, it’s not wrong to have that in your posts, but it’s distracting from the main content. And there is a “Jump to recipe” button in this example, but it’s hard to find. You have to scroll down a little bit before you find it. And then the example, you know what, I should probably find what the exact words say. The word filler’s in there 16 times. Now, let me find that recipe, because-
Arsen Rabinovich (00:17:42):
While Marie’s doing that, I pasted a link to the PDF file for the Quality Reader Guidelines in the chat. For those of you who are interested, you can follow along and use that command F or control F button to find specific words within those guidelines.
Dr. Marie Haynes (00:18:01):
So page 67, they give an example of a low quality website, is what they say.
Arsen Rabinovich (00:18:06):
Right.
Dr. Marie Haynes (00:18:07):
And you can click on the link and they show you a PDF of the website. And again, it doesn’t look like spam to me. It looks like a decent website. I think the recipe looks good. And it says, “The primary purpose of the page is to share a recipe for butterbeer, while helpful main content such as the ingredients and information about butterbeer…” Oh, they misspelled it. Butterbear. “Are at the top of the page. Significant scrolling is required to find the actual butterbeer recipe. The “jump to recipe” link itself is hard to find.”
(00:18:41):
And then there’s just a couple more paragraphs I think are worth reading. “The page has a large amount of content unrelated to butterbeer between the butterbeer ingredients and the butterbeer recipe. The page has photos and reviews of many other unrelated foods at Universal Studios.” That’s the blogger trying to put in more affiliate content. “In addition, ads and supplemental content and interstitials appear throughout, causing additional scrolling to finally find the recipe towards the bottom. All of the above lead to a poor user experience.”
(00:19:14):
And then it says, “Note: A higher rating could be appropriate if the page were better organized, the recipe was prominently placed and easy to find, and there was less filler, supplemental content, and ads distracting from the helpful main content.” There’s a lot in that, right?
Arsen Rabinovich (00:19:33):
Right. [inaudible 00:19:33] to digest. So with that, you and I quickly talked before the webinar, we’ve been preaching an approach to, and especially since Google published some of the content around the Helpful Content System, and frequently we see and read that Google’s looking for a unique perspective. And we kind of see that already start taking place in search results for certain queries.
(00:20:02):
Not a lot of food-related queries, just because of how saturated the results are with the traditional recipe post layout. But the ones that are winning, and we’ve been looking at that and measuring and preaching, especially in our coaching, that the recipe card is the body of the post. It’s not a separate thing from your recipe. Those ingredients, those instructions should be enough for the user to make the recipe. And you should always ask yourself, if I land on your recipe card, there was no content above the recipe card. If I land on your recipe card, is there enough information for me to make this dish properly? If you can answer yes, that’s the recipe.
(00:20:47):
Above the recipe card… We’re not saying move the recipe card up. We understand everybody still has to monetize, and all that. But above the recipe card, that area should be utilized for you to present your unique perspective around either the ingredients, the process, the cooking, the preparation method, the tools that are needed, any substitutions. It should not be a rewording of what you have in the recipe card, because that’s redundant text. The recipe card is a part of the body of your post.
Dr. Marie Haynes (00:21:22):
Yeah. There’s another new section that they added, “5.22 Filler as a Poor User Experience”, and they speak directly to recipe sites. They’re talking about, “A high quality page has the most helpful main content placed most prominently.” And then it says, “For example, on recipe pages, the recipe itself and important supporting content directly related to the recipe should be prominently displayed near the top of the web page.” So that’s not good news if you have a recipe site, because of monetization reasons, and that’s always the struggle. But from a user’s perspective, if that’s what I’m looking for, it makes more sense.
Arsen Rabinovich (00:22:08):
I’m here for the recipe. I need the recipe. Right?
Dr. Marie Haynes (00:22:10):
Yeah. And you’re right, the Helpful Content System, the vast majority of the sites that I’ve seen that were impacted strongly by this system, you had to scroll through page after page after page to get to the part that was actually what the user went there for. So it’s important to have the helpful part near the top.
Arsen Rabinovich (00:22:32):
Right. And there’s a lot of questions popping up about “jump to recipe” button. And they think if it’s properly placed and you don’t have to look for it… Because keep in mind, Google renders your pages, so it’s not just looking at the code. If it’s properly placed and it’s easy to find, it’s a good method to jump to the content. You always also want to make sure that you’re not diluting away from the purpose of that page. More of that dilution that you do… Go ahead, Marie.
Dr. Marie Haynes (00:23:00):
I’m just looking at one of the comments that this site that was used as an example is still ranking number one for Butterbeer. This is important, because the changes were just made to the guidelines in January. So this means that Google’s currently training their systems to do better in these respects. So this will be interesting to watch for those rankings when we have another core update.
Arsen Rabinovich (00:23:24):
They have a [inaudible 00:23:24] of this.
Dr. Marie Haynes (00:23:24):
Yeah, we have another core update, which could happen. Google said they’re going to be doing them more frequently and then they didn’t do one for several months, but it’ll be interesting to see if that is still the case at the time.
Arsen Rabinovich (00:23:35):
At the same time, we have to remind people that there’s other reasons why that site could be number one. There could be other signals that are keeping it in that position. So from a content perspective, and Google has said that, like, if we can’t find a better result, we’re still going to show a result that we would probably not if there wasn’t better results, right?
Dr. Marie Haynes (00:23:51):
Yes.
Arsen Rabinovich (00:23:52):
So there’s other, they could be clicks, it could be dwell time, it could be engaged users, there could be other reasons why they’re ranking number one. Even if, a lot of times we’ll look at a result, even on our consultation calls, we’ll be like, “How is this number one? They break every single SEO rule.” They’re repeating keywords in every heading. Their headings are long sentences, almost paragraphs. How are they number one? There’s other reasons for that.
(00:24:16):
Okay, so with that, there’s been a lot of discussion about AI generated content. You and I quickly talked about this as well earlier. What’s Google’s latest take on this, and how can bloggers use AI without getting in trouble?
Dr. Marie Haynes (00:24:31):
So the guidelines actually mention AI content. There’s some new places that they added, and they basically talk about what the definition is of generative AI. Google has a bunch of documentation on their stance on using AI, and they are not against using AI. Where they’re against it, is when you use it as a quick way to make content. Throughout the guidelines… Let’s see, the number of times the word effort appears. 120 times. And so they talk about, if you see AI as a quick way to generate content without effort, then that’s spam, essentially. But there’s many ways that you can use AI to help you to make content. You can use AI to get ideas on how to improve a recipe. You could use it… I don’t know that I’d want to use it for image generation just yet. I think for recipe blogging, I want to actually take pictures of the food itself.
(00:25:33):
But you can use it to maybe verify facts in your article. That’s something that Google says there were some new things added in terms of accuracy, and Google traditionally hasn’t been able to tell if your content’s actually accurate, and so you could use AI to summarize your comments. I think comments are very important, and that’s actually listed in the guidelines as well. I would do a search for comments, if you’re a recipe blogger. Comments can tell whether people are using the recipe, whether they’re finding it helpful, whether they’re suggesting changes, and you can use AI to take all of your comments, put them into a tool that has a large context window. I like using Gemini and AI Studio. And just say, “What are the complaints about this recipe?” Or “What are people really pointing out in this recipe?” And then you know to put that near the top. If people are like, “This is the creamiest mac and cheese I’ve ever made”, then that’s a feature that you want to put at the top. You want to put that in your title tag, that this is one of the reasons why this recipe stands out, why people use it. So yeah, you can use AI.
Arsen Rabinovich (00:26:48):
You can also do this For your competitor’s comments. You can take a look at, you can grab their comments and see, because if you’re writing a new post, and you’re competing against someone who has maybe 200, 300 comments, you can take a look at what the user feedback is there, and that could be another opportunity to establish a [inaudible 00:27:01].
Dr. Marie Haynes (00:27:01):
And you can use it to help with originality, which sounds like a contradiction because AI can only really generate stuff that it’s seen before, but then it also knows what it’s seen before, and it can put information together. So you could take this page that you’re about to publish and ask. Again, I like using Gemini. The thinking models that you can get in AI Studio, and they’re totally free to use, are the best ones right now. And ask it, “What aspects of this page are original to this page and valuable to a searcher and could not be found elsewhere?” And if the system can’t determine anything, then you can ask for hints like, “What could I do? What could I add to this page to make it uniquely original and helpful for people?”
(00:27:49):
Sometimes it’s going to give you suggestions that maybe don’t make sense. I always think of when I’m using AI that it’s like an assistant who’s sitting next to me who has all these great ideas, but isn’t always perfect all the time, just like a human. So it’s like having somebody who’s seen every recipe in the world look at your recipe and give their opinion on whether it’s a good one, whether it’s unique and original and valuable to people.
Arsen Rabinovich (00:28:18):
We talked about brand reputation earlier, and I want to come back to that, because I feel like this is super important. And how it applies to EAT, and not so much the offsite aspect of it, but how do we signal that reputation? How do we hint towards it onsite? Can we talk about that? And that wasn’t one of the questions, but I want to come back to this.
Dr. Marie Haynes (00:28:44):
A good chunk of your on-site EAT is the quality of your content. So having… I can’t prove this, but I really believe that one of the factors of quality for a recipe website is whether people make the recipe, and whether people come back to make it. Again, the Helpful Content guidance talks about, “Is this the type of content that people tend to bookmark and share?” Well, so those are examples. It’s not like there’s a direct flag that’s like, “Oh, this post’s been bookmarked 800 times, therefore it’s good.” Posts that have been bookmarked a lot are examples of potentially helpful content, content that people have found helpful enough to save for another time, or to share with somebody else.
(00:29:26):
So one of the top things in EEAT is to just genuinely have content that people are finding helpful. Now, you also can say stuff about yourself, and we’ve experimented with using verbiage like, “I’ve been cooking these recipes for 30 years” or whatever. However, that’s actually directly spoken about in the guidelines too, that it’s not enough to just say, “I’m an expert.” There needs to be external evidence of that.
Arsen Rabinovich (00:29:56):
Right. Third-party validation for it.
Dr. Marie Haynes (00:29:57):
Yes, third-party validation. Now, that validation could come from the comments, and I think Google probably is pretty good at detecting whether those are legitimate comments or pattern. I think it’d be pretty easy to see patterns of self-made comments. So the comments are, and that’s discussed in the guidelines, that they’re a part of your reputation that contribute to… I worked with one site owner who was strongly impacted by the Helpful Content System. And she used to rank for this one, like a substitute for something, and her recipe, and the comments on her recipe were like, “I tried this and it didn’t work.” “This recipe was not good.” And so those speak to your reputation.
(00:30:43):
For a lot of topics, I don’t know if I want to say not so much recipe, but it would still help, the guidelines talk about what others have to say. And so, being mentioned in 2017 at Pubcon in Austin, I asked Gary Eash from Google, “How does Google determine EAT?” I didn’t think he would answer me, but he gave a very, very specific answer. He said, “It’s primarily based on links and mentions around the web.” He said, “For example, if the Washington Post mentions you, that’s a good thing.” So now we’re not all going to get our recipes published or linked to from the Washington Post, but there has to be some type of external evidence that you are known for those topics. And that might be like, right now, what we’re doing right now, this builds my EEAT for talking about Google algorithms, for talking about the quality raters. You wouldn’t have me on if I didn’t have something to share on that topic.
(00:31:42):
Going on podcasts, doing interviews, really if you can get links, I’m not talking about the page rank, but if you have content that’s good enough that people legitimately want to mention it, then that speaks to your EEAT. But it’s very hard to just get people the link because you have an awesome recipe, out of the trillions of recipes that exist. And another way to build your EEAT is to write a book. It’s not hard to do. Publish a recipe book of your favorite recipes. I made it a goal for myself one day to figure out in one day how to get a book on Amazon and self-publish. This was when Bard first came out, so I used Bard, and we wrote a book about how to prompt Bard. And did it in one day, got it up, published on Amazon, and now I’m a published author on using language models. So that’s something that can speak to your EAT as well. So yeah, lots of things.
Arsen Rabinovich (00:32:43):
Right. And we talk about this, about connecting the dots from an entity perspective for Google, and how you should mention, “I’ve been cited here”, or “I’ve been interviewed here” on your About page. And even on your blog post, on your recipe post, or even the blog post, we talk about how the top section, the above the fold that’s right under your title, that area should be used for two things. Should be informing the user what they’re about to get into. What’s in this recipe? Is there a specific preparation method that they need to know?
(00:33:21):
And then, second thing should be why they should listen to you. And if you can cover that properly, and again, I don’t think these are algorithmic factors. These are secondary signals, where by doing this, we’re going to engage the user and keep the user on the page, prevent the click back to the results, and have them dwell on this page because now they like, “Oh, I have all these ingredients”, or “I don’t have all these ingredients” and “I should listen to this person about this recipe because of XYZ.”
(00:33:55):
With that, so we talked about comments, and with comments come ratings, and Google seems to be cracking down on expired domains and manipulated user rating, and that’s kind of in one section. Is Google treating manipulated user ratings as a part of the expired domains update, or is that a broader initiative to improve content trustworthiness?
Dr. Marie Haynes (00:34:17):
Now that might be a part that I missed in the updates, about ratings. I know the expired… Let me find the expired domains section they added. Which surprised me, because I thought expired domains was something Google has been on top of for quite some time.
(00:34:35):
Expired domains… And then they do cite reputation abuse. Yeah. I mean, there is information in the guidelines about the legitimacy of reviews, and so I think that that’s important. I really think that, again, Google can see the patterns. So if that is in there and I missed it, then Google’s looking for patterns of these are legitimate user comments. And then all they need to do to train a machine learning system to determine whether they’re legitimate or made up is just show them a bunch of examples of each. And then the systems can predict.
(00:35:16):
That’s what SpamBrain does, Google’s AI brain for spam, mostly for links. They can see which links are legitimately like somebody wanted to recommend this page, and which ones were made just for SEO purposes, and find the patterns that connect there. So yeah, I certainly wouldn’t recommend faking reviews. I can see the-
Arsen Rabinovich (00:35:41):
That’s why I brought it up. There was a question when people registered for the webinar, there was a few questions that people were concerned about anonymous reviews on their sites.
Dr. Marie Haynes (00:35:53):
Oh, yes.
Arsen Rabinovich (00:35:53):
I don’t have any insights into that, and I was hoping that one of us could come up with something, but if not, then that’s fine. So with all of this, and let’s get more into how should this be applied. With all of this knowledge, and with all of these changes, let’s talk about recipe bloggers. Should bloggers be rethinking post structure? And we talked about this. Recipe card, and what’s above the recipe card, and the jump to recipe. But from a post structure perspective, keeping in mind that the way they make money is through the scroll and the exposure to ads and all of that, how should they take these changes to the quality rater guidelines? Understanding that this is just the guidelines for the raters right now, and that we assume that maybe down the line this is going to be a part of the algorithmic evaluation. What can we safely recommend from, like, let’s implement this. Let’s start preparing for this.
Dr. Marie Haynes (00:36:51):
I think it’s a really difficult place to be in, because if you look at the guidelines as what Google is aiming to reward, and if I’m right and there are going to be algorithmic changes that go into effect because of these changes that are put in the rater guidelines, then Google wants to see, probably, the recipe card really high up on the page. And I would imagine for most recipe bloggers, that’s going to cut your monetization tremendously.
(00:37:21):
I would say, perhaps wait and see. Because right now, it’s just a change that’s in the guidelines, and I would expect that at some point in the future, if this happens, if Google starts to reward pages where the recipe card is up higher, well then we’re going to see that, and the whole world is going to be talking about how this change is happening, and how Google is making it difficult. It seems to get harder and harder to monetize informational websites, whether that’s recipes or knowledge or whatever. But I would at least make sure you have a prominent, like you said, a prominent “Jump to recipe” button where people can find it. I don’t know. Do you use any tools like Clarity, Microsoft Clarity?
Arsen Rabinovich (00:38:10):
Yeah, a lot of use bloggers use Clarity. Yeah.
Dr. Marie Haynes (00:38:12):
I would be looking at whether people are clicking on your “Jump to recipe” button. And if not, then maybe it’s not in the best place, or maybe it’s hard to find, maybe it’s hard to tap. Maybe I’d also look at your ad experience. It’s totally okay to have ads. Google knows that people need to make money from their websites, but where it becomes a problem is when the ads are really distracting the searcher. So some recipes, you can’t see the recipe card because there’s an ad that you have to close. And I know that that’s on purpose, because those ads tend to get clicked by accident. But just kind of looking at your… Some sites need to reduce the number of ads, and there’s really no actual guidelines for how to do that. It’s more from the perspective of a user. Here’s something that you could try. I don’t know how many of the people in the audience have this, but Arson, have you used ChatGPT’s Operator yet?
Arsen Rabinovich (00:39:07):
No, not yet.
Dr. Marie Haynes (00:39:09):
And most people haven’t, because it’s part of the $200 a month plan, which you have to be crazy, like me, to pay $200 a month to use a language model. But I tried it out, and it is wild. So what it does is it does browsing for you. It opens up a browser, and it’s like an assistant that browses. I said, “Go to my client’s website and try to purchase…” Whatever their product was. And it couldn’t open. It had a hard time opening this one accordion, and it gave me all this UX or user experience information. And so that’s kind of a cool thing to do, is to see, like, can this browser actually make its way to where the main content is?
(00:39:56):
Another thing you can do is put your entire page, take a screenshot, like a full page screenshot. I use a tool, a Chrome extension called Awesome screenshot to do that. Put that into Gemini and AI studio and ask, “Where is the main content on this page?” And you can even give Gemini a copy of the rater guidelines. Say, first of all, “Tell me what you know about main content.” Then ask, “Where is the main content on this page? And is it easy for a searcher to find?” And that can give you some hints as to whether people are finding it.
(00:40:32):
Then really, it sounds kind of trite to say, but I really would focus on your users first. I know it’s hard to wrap our heads around, well, how could Google know if we’re satisfying our users? The DOJ versus Google trial told us a lot about the importance of click signals, and I don’t think it’s direct. I don’t think it’s like, “Oh, your recipe got clicked on 10 times more than this recipe, therefore it must be better.”
(00:40:57):
I think again, the pages that get… There’s three things that are measured in terms of clicks. There’s the first click, whether you’re the one that sites choose, and I’ve seen some of the people who have dissected the API docs that got leaked, said that Google actually predicts what percentage of the clicks you’re going to get for a query. So if they have you ranked at number three, and you’re actually outperforming the prediction of clicks, well then there’s a good chance that you’re going to jump up to number two or number one.
(00:41:35):
And so the first click, the the longest click, which I think is super important for recipe sites, is to keep people on the page, because you’ve got it open and you’re making the recipe. And then the last longest click, meaning that there’s a good chance that you’ve satisfied the search. So if we aim to be the result that people find the most satisfying, then that should be our ultimate goal.
Arsen Rabinovich (00:42:01):
Right. And we talk about this, we talk about last meaningful click. That not all queries are going to have hundreds of thousands of monthly searches. They could have 20, but that’s still a topic that interests someone, and the content needs to help with that. And that as long as there’s a click, and it’s a meaningful click, we can say that this content is helpful. You published a quality rater guidelines workbook that I’m going to link.
Dr. Marie Haynes (00:42:28):
I did, yes.
Arsen Rabinovich (00:42:28):
Can you quickly tell us what’s that all about?
Dr. Marie Haynes (00:42:30):
Yeah, sure. So I first wrote this guideline book in 2018, and that’s back when I had a team working with me. And we created this whole checklist that we would assess websites, trying to look like a quality rater, basically. And so it talked about how to assess EEAT, how to assess your reputation, and then I rewrote that book in, I think, 2020. And then I’ve been meaning to rewrite it for years, and then when I saw this update that happened in 2025, I knew I needed to rewrite it. So it basically is a workbook. It’s 88 pages, where you can go and answer all these questions about where is the main content, look at your competitors and see where their main content is. And then there’s also a bunch of prompts you can use in language models and even some GPTs that I created with a bunch of my other books on helpful content and stuff in there, that will help you brainstorm ideas on, how can I improve my reputation, and how can I improve helpfulness of each page? So you got the link there, and-
Arsen Rabinovich (00:43:37):
Got the link, it’s in the comments, it’s in the chat.
Dr. Marie Haynes (00:43:39):
Yeah. Hundreds of people have used it over the years, and a lot of SEOs use it as part of their audit process. It takes quite a long time to go through one page and analyze it with all of the questions in there, but you can get tons of insight into how you could improve your site. And then Google actually says in their documentation on helpful content that if we read the rater guidelines, it can help us. The words they use are “Align conceptually with what their algorithms aim to reward”. And so that’s the goal in this book is to help people do that.
Arsen Rabinovich (00:44:19):
I love that. That’s really helpful. Okay, so we’re almost at our 45-minute mark. I want to get through some of these questions. We do have a lot of questions. I’m going through the starred questions. I’m going to pick out the ones with the most thumbs up. We’ve answered some of these already. Okay, so let’s go. We talked about placing the recipe cards higher. “How does Google view block groupings? Is this significant?” What do you mean by block groupings?
(00:44:56):
I think the question is blocks inside of WordPress. Content blocks. I’m assuming this is the question.
Dr. Marie Haynes (00:45:02):
I don’t think it really matters. I mean, I think as long as it’s not done in a way that’s distracting users, but I don’t think that… Well, let me take that back a little bit. Those API files that were leaked, I don’t know how many have heard of this. Last year, there were a number of people that found this leaked version of these API documents from Google. And the documents, they don’t tell us exactly how search works, but they tell us a lot of the attributes that Google can use in their search algorithms. And that’s where we learned a bunch of stuff about Navboost, and the first click, and all that stuff.
(00:45:44):
Those documents updated a couple of times since everybody was going crazy looking at them. And the latest version, I think it’s like 0.6.0, we were looking at 0.4. They added a section called “Chunks”, and it talks about how Google can see the chunks that are on your page. And I actually believe that when you get a page indexed, Google tokenizes that page for use with their language models, meaning that they convert it into numbers that are used for vector search. And so Google can break your page down into chunks. I don’t know if this even relates to the question-
Arsen Rabinovich (00:46:21):
[inaudible 00:46:22].
Dr. Marie Haynes (00:46:22):
… but it makes sense to me to have very good use of headings. So if you have a section where you’re like, “Here are the ingredients”, I don’t know, that doesn’t really make sense. But if there’s something specific, “Oh, okay, yesterday I made these oysters”, and if you had a section on shucking oysters that wasn’t exactly the recipe, I would give that its own heading, so that when Google is using vector search to find content that matches the searcher intent, it might pull from that chunk of content, basically.
Arsen Rabinovich (00:47:00):
Right, right. That makes sense. We have another question here from Chrissy. “What about adding additional ‘jump to recipe’ links every so often throughout the post, making it clear that it’s an option for someone missed the initial jump?” And I suggest this, I think we had a checklist at one of the events that we recently did, that your ingredients section should not be a rewording of what’s in the recipe card section for the ingredients. It should be a unique perspective around the ingredients. So you can say, “Hey, if you don’t have fresh strawberries this time of year, wherever you live, you can use frozen ones” and all of that. So that’s the unique part.
Dr. Marie Haynes (00:47:35):
Great. Yeah.
Arsen Rabinovich (00:47:36):
But the opening paragraph, the opening to that section, should be “All of the ingredients and measurements are listed in the recipe card below”, and you can jump link that to help with user experience. And Google does see those links and will travel down to that section.
Dr. Marie Haynes (00:47:52):
I can’t see there being any negative in that whatsoever. I just thought of another tip that would be helpful for this type of situation. For the content that’s at the top of the page, you can use either ChatGPT, or again Gemini and give it your entire page and say, “What are the three to five most interesting things about this page?” Then, “Write me a paragraph that encapsulates those things.” And so if your page talks specifically about a certain technique, or something that that’s what’s unique to the page, then that will be mentioned at the top of the page. I really believe that the first paragraph scroll, perhaps, is super important to Google understanding the context of what your page is about. So that’s something that I’ve found helpful.
Arsen Rabinovich (00:48:43):
And there’s multiple places where context is derived from, but we know that where that content is positioned is also very important. We have another question from Laura. “What about main content versus main entity? For a recipe site, isn’t the recipe card considered a main entity?”
Dr. Marie Haynes (00:49:03):
I mean, to me, an entity is more of a person or a specific thing. And I know it’s kind of a popular thing right now to talk about entity search. I don’t know that I pay as much attention to that. I really don’t know how to answer this question. I think, I mean the recipe is really the main content. The main content, and it says in the guidelines, the content can be text, it can be images, it could be a video is part of your main content. And then for recipe sites, the main content is the part that helps people actually make the recipe. But yeah, I don’t know about entity.
Arsen Rabinovich (00:49:49):
We answered this in our chat. Here’s a good one. Tammy’s asking. “Are backlinks considered a third party validation?” Marie, I’ll let you.
Dr. Marie Haynes (00:49:58):
I get into trouble talking about links, because I have differing views. Let’s talk first about if you are a black hat SEO, which I know none of you are here, then links are really important. I was being facetious, maybe there are a couple. To me, links are important for a few things. One is for Google discovering new content. And if you have a website that doesn’t have an existing audience, that isn’t naturally getting linked to from other places, then in order for Google systems to find it, you need to link to it. And that’s one of the reasons why, if you’re not in the top 20 to 30 results consistently, then you’re going to need links in order to get there. And so there’s a whole… Some people have found systems to make your own links to help Google discover stuff, and then Google is continually trying to fight against that.
(00:51:02):
Links, though, are an important part of your EEAT. They can help. In the very original PageRank patent that was written, it was really interesting that they talk about every document had two things. One was the doc ID, so it was like an identifier that could label that page on your website. And then also, there was a component of the anchor text, that what were people linking to you for? And if you look at my anchor text, if you go to Ahrefs and open up, or Semrush or whatever, and open up my site, you’ll see people link to talk about me, about EEAT, or about quality raters, or Google algorithms. The anchor text of the links that point to your site actually help Google understand what your topics are about. So you can see why people would want to go off and create links.
(00:51:56):
So some of you may not know that for many years my main, the way I got into SEO was being interested in Google penalties. And so, I’ve probably removed more unnatural links, manual actions, than anybody. And I’ve seen millions of examples of links that were made for SEO purposes. And so if I can see them, if I can open up a page and go, “Okay, this was not a natural link, this was just somebody linked for the main purpose of SEO, of manipulating rankings”, then Google systems that look for patterns can see that too. So backlinks are great. You should get backlinks, but backlinks that you have to make, and that you have made because of relationships that wouldn’t have existed other than SEO, can trigger things for Google and may not necessarily be a good thing.
Arsen Rabinovich (00:52:47):
And when we talk about third-party validation, we’re talking about where, think of it from, I’m going to use the word entity, think of it from an entity perspective. So just a backlink is not counted towards, I wouldn’t say any regular backlink as third-party validation, at least in my eyes. I think if you have a backlink from a food-related news publication where you were interviewed, or where you were cited or quoted, or a podcast that you were on, or brand that you’ve worked with, or if you published content on the third-party website, or you have a author bio on foodnetwork.com, I don’t know, right? Those are the third-party validation signals that I think are meaningful, because Google has an entry for that website, if it’s a popular website and it’s been around, and understands what that website is, the entity around that website, and the topics associated with that entity. And if you are closely related to that, I think that’s a very, very strong third-party validation signal.
Dr. Marie Haynes (00:53:56):
I want to add, because I see a lot of recipe bloggers do this, where it becomes difficult is where we often create these link wheels. And sometimes it can be okay, like Arsen, if you wrote an SEO piece and I link to you and then you link back to me, that happens because we know each other. There’s relationship there. That does happen. However, you can often clearly see the patterns that, where at the end of a recipe, there will be links to four other recipes you might like, and then those have links, like it’s a network.
(00:54:32):
And I can’t remember which update it was, but in the last couple of years there’s been a couple of updates where a number of sites that use that type of reciprocal linking suffered losses. I don’t think that they were penalized for the links. I think it’s just that they actually were helping, and then Google figured out how to neutralize them. So really if you can get a link that truly is somebody mentioning you, then that contributes to your EEAT.
(00:54:59):
A good way to check, I wrote a post recently on using Gemini to get an idea of your EEAT. And it’s not a hundred percent perfect, but you can go to the language models and ask, “What is…?” And put your name in, or if there’s multiple people with your name, “What is your name recipe blogger known for?” And if the language models don’t have information on you, then that’s a sign that you need to be building your reputation. And it’s not because of links, but it’s because of where you’ve been mentioned, which kind of is related.
Arsen Rabinovich (00:55:32):
You can also search, you can use the minus operator. You can say, “My name minus my website”, and all of that. We are about five minutes away from our finish line. I have one more question that I want to bring up here, which I think is important, and we did kind of talk about this. Kathleen is asking, “I’ve been seeing more bloggers placing the recipe cards up higher in the post under ingredients and under how-to, then having a supplemental below that information. Thoughts on that?” So I see, and I’m going to hide this question, because eating up a lot of the screen. But I do see this happening and I do see this happening frequently in the top 10 results. And I think that’s an interesting cue for us.
(00:56:10):
I think Pioneer Woman does a really good job at this, where they do a really good job at right above the fold, as soon as you land, they tell you, “This is how long it’s going to take to make, this is what’s included”, kind of prep you for what you’re about to get into. Then there’s a video, a short video that plays without sound for people who don’t want to read, they want to see, let me quickly see this 32nd video of what I’m about to get myself into. Then it’s a bunch of questions, it’s a recipe card, and then under the recipe card, they have their process photography and all of that.
(00:56:40):
So it is worth… Now, again, we can’t say correlation causation. We can’t say that this is the reason why, but it does kind of speak towards prioritization of content, giving the user what the user is there for, and keeping it more focused towards that primary topic towards the top. Your thoughts on this?
Dr. Marie Haynes (00:56:59):
Yeah, I think, one of the things I really like to do is to search for my own site or my client’s site on my phone, or you can emulate it on a desktop. And then just see how many scrolls it takes to get to the part that they wanted to get to. So if you move the recipe card up, but it’s still behind five scrolls of content, that’s probably not so good. As long as users can easily find what they’re looking for, then that’s a good thing.
(00:57:27):
You can look at your GA4 information, which I know it’s difficult to get into GA4, but if you look at engagement metrics, I did some work that really showed that session duration correlated with rankings improving. I think it’s much more than that, but session duration is a good measure of whether something’s helpful. And then look at which of your posts have longer session duration, and which of your ones people are clicking away from. And if they’re clicking away, it’s probably because they didn’t find the answer they’re looking for. And sometimes you can see, “Okay, I put way too much up above where the card was supposed to be.”
(00:58:06):
And I think this remains to be seen. I do think Google mentioned it enough in the rater guidelines that we’re going to see a change in the future, and then we’ll have to see. It may be devastating, that a lot of sites that have their recipe at the bottom don’t do well. Or it might be like when they went all out about Core Web Vitals, and it was a minuscule change. Core Web Vitals are still important, but-
Arsen Rabinovich (00:58:32):
Not as a big deal.
Dr. Marie Haynes (00:58:34):
… but not to the extent that Google kind of threatened, not threatened us, but made us feel threatened about.
Arsen Rabinovich (00:58:41):
It’s interesting, because when you organize your page, you kind of want to give it a little bit of your own spice to it, for lack of a better word. And it’s easy to start moving away from the focus of the topic. One thing that I preach is, you want to ask yourself, “Is this content, is this piece of content really helpful to the recipe, for the reason the user is here?” If you can honestly say that this is not going to make this recipe any better, I don’t think that you should prioritize it higher towards the top.
(00:59:24):
There’s so much more that we could be talking about. We can go another hour on this, but we’re about 30 seconds away. I want to thank you for joining us. Melissa, I think you have an ending slide that we’re going to put up here in a second. I’m going to one more time, link out to Marie’s QRG workbook for everyone. If you didn’t catch that, definitely check it out, and subscribe to us through our website, through YouTube, and all of that. Subscribe, like, and comment stuff. Marie, thank you so much.
Dr. Marie Haynes (00:59:56):
Yeah, thanks for having me.
Arsen Rabinovich (00:59:57):
How can people find you? I mean, we’ll have a slide, but can you tell us how can people can find you?
Dr. Marie Haynes (01:00:00):
Yeah, I saw a ton of questions that we didn’t get answered. I have a community called the Search Bar. If you go to community.mariehaynes.com, there’s a ton of great people in there. And you can message me through that as well, and I’m happy to answer those questions. And then you can email me at help at mariehaynes.com as well. I often take quite a bit of time to reply to my emails because I have a lot, but I do reply to all of the questions that I get. So yeah, I’m quite happy to answer questions.
Arsen Rabinovich (01:00:34):
Thanks for coming and talking to us today. You’re amazing. We love you so much. Thanks, everyone.
