The future of Entities, Fraggles, and API indexing w/ Cindy Krum
In this Tech Bound Conversation, I talk with the fantastic Cindy Krum about Google's Indexing API, Fraggles, and entity recognition with machine learning.
The future of Entities, Fraggles, and API indexing w/ Cindy Krum
1:23 How Cindy started Mobile Moxie
2:33 The challenge of writing a book
4:47 New Mobile Moxie tools
9:37 The implications of Apple’s App Store scandal
11:50 Why Cindy started Mobile Moxie
15:15 Fraggles and how Cindy discovered them
19:33 How to use Fraggles for SEO
24:04 Wordpress Gutenberg and Indexing APIs
38:55 Gary Illyes and Googler statements
44:37 NLP and entity recognition
50:35 Google’s recent indexing problems
- Mobile Moxie: https://mobilemoxie.com/
- Sign-up on Mobile Moxie: https://mobilemoxie.com/ with the promo code “KEVIN”!
- Cindy Krum’s book: https://www.amazon.com/Mobile-Marketing-Finding-Customers-Matter/dp/0789739763
- Ted Talk Eat Pray Love: https://www.ted.com/talks/elizabeth_gilbert_success_failure_and_the_drive_to_keep_creating?language=en
- Fraggles article Moz: https://moz.com/blog/the-importance-of-fraggles
- Jobs indexing app: https://developers.google.com/search/apis/indexing-api/v3/quickstart
- Gary Illyes Q&A: https://www.kevin-indig.com/my-notes-from-the-gary-illes-qa-bay-area-search/
- NLP API: https://cloud.google.com/natural-language/
Kevin: 00:00 Okay. Cindy, welcome to the show. Thank you so much... Show, it's not even a show yet. For all of those who don't know you, congratulations you're missing out big time, everybody should know you, but for the few folks who might be new to the game and have no idea who Cindy Krum is, please tell us who Cindy Krum is.
Cindy Krum: 00:20 Sure, hi everybody, my name is Cindy, and I'm the CEO of MobileMoxie, we do ASO and SEO and have awesome tools, and we've been around doing really technical SEO kind of projects for over 10 years.
Kevin: 00:36 And you're also a kick ass speaker and writer, and you're the proud owner of Barkley. Who's Barkley and how is he doing?
Cindy Krum: 00:46 He's good. We're getting ready for winter, all the leaves, he bring in so many leaves, because he's like velcro, everything just sticks to him.
Kevin: 00:56 To be fair, Barkley is a dog and not a human, but I think at this point it could of been either way.
Cindy Krum: 01:03 Yeah.
Kevin: 01:06 Cool. So, awesome - most productive member of MobileMoxie. While we talk about MobileMoxie, when did you start, and how did it develop? What was the reason for starting MobileMoxie and where are you at right now?
Cindy Krum: 01:24 Sure. So, I started over 10 years ago, I think 2008, and it was after working at an agency for a number of years, I had been speaking on behalf of the agency and I just wanted to focus on Mobile a lot more, so I left. I also wanted to write a books, so I left and I wrote my book, and I started my little consultancy and it's been a wild ride. But our consultancy it's stayed really small, for many years it was just me, but now we have three full time consultants and a bunch of part-time developers, and it's a fully remote company. Well, the consultants are all in Denver, but developers are all over the world. So yeah, it's fun, I get to travel, it's what I wanted.
Kevin: 02:20 It sounds like fun. What was the book real quick, just for the show notes?
Cindy Krum: 02:29 MobileMarketing: Finding Your Customers No Matter Where They Are.
Kevin: 02:31 Nice. What was the experience writing the book?
Cindy Krum: 02:33 I hated it.
Kevin: 02:34 Why?
Cindy Krum: 02:34 It was horrible, I don't recommend it, if you value your mental health it's something you should avoid.
Kevin: 02:42 What was the most dreading part about it?
Cindy Krum: 02:45 Well, half way through the book the publishing company that asked me to write the book decided that they wanted to totally reformat the book and put it under another series. So, I was writing for this one series that was a bunch of short chapters for introductory level. I had written a bunch of short chapters for introductory level and they were like, "Oh, nevermind, what we want is long deep chapters." So, I basically had to start over, and they said, "You can just cram everything together." But that's really not how it works, because there are intros and outros and stuff like that, so, I basically had to start over, and they didn't pay me anything extra. In fact, I wished I had stayed with the original series, but I didn't realize back then that I probably could have objected or that I had a choice, I just did what I was told.
Kevin: 03:31 Yeah, I probably would have done the same. It's probably a tough call, especially when you do this for the first time, but would you ever contemplate to write another book, maybe in a different way, maybe not with a publisher even?
Cindy Krum: 03:44 Yes, I would consider self publishing, and I would consider maybe writing another book with another company, yeah. But it was just like, there's a TED talk with the author of Eat, Pray, Love, and she talks about how sometimes you have to squeeze every single line out of your veins, and sometimes it felt like that.
Kevin: 04:11 I will certainly include a link to that in the short notes. So, if any of you a publisher out there listening, just send Cindy a fat check any maybe we can talk about a book. Because, I think it really shows, I love your writing, I've read many of the long series of articles on MobileMoxie about entities and your ambitions and [fraggles 00:04:36], and it's such a level up in SEO and understanding. So, that was hugely helpful, and so yeah, just want to congratulate you on that, and thank you very much, because it was amazingly insightful.
Kevin: 04:47 Before we jump into that, actually quickly want to point it out that in doing a bit of research for this episode, I looked at mobilemoxie.com and saw that you put a bigger emphasis on your tools. So, can you talk us a bit about what you got going there? I'd also be very curious about how you develop these tools, like what your goal is and maybe what your thinking behind these tools is?
Cindy Krum: 05:10 Yeah. So, actually MobileMoxie has had tools for many years. And I decided for my 2019 goal was to really do a better job of focusing on the tools, because I think they're really cool and they're really useful, but, they always sat in the background for consulting, consulting has been the main focus. But I wanted to really make sure that the tools got a fair shot, and I wasn't just sticking with what I was comfortable with. So, I switched and made the whole emphasis of the site driving sign ups for the tools. The tools are actually really cool, and there's not anything else out there that I know of that does what they do. We created them because I had a need for them with consulting clients, because SEO results change from desktop to phone, especially on the phone they're more location specific and aware, and they change and show different things related to your location. I had clients all over the world, and I wanted to be able to test and see what people in the locations of my clients or in the locations of my clients clients were seeing when they were doing searches.
Cindy Krum: 06:28 So, we made a tool that lets you locate yourself anywhere in the world, Choose from a bunch of phones, you can even choose two phones at a time if you want, and just put in a query and see what the live Google result looks like right now. It's even as accurate as if you're testing in the US somewhere in Europe and it's the middle of the night, all the map pack results will be there, but they'll all stay closed, because it's the middle of the night. So, it's a real time live testing for what a SERP looks like from anywhere.
Cindy Krum: 07:01 So it's caught on with the local SEOs, but anymore I think so many searches are mobile and customers are mobile, but everything is to some degree a local SEO play, because if you think you're Best Buy and you're working for Best Buy, well, Best Buy has lots of offline locations, that's really what they're going for. They're probably going for online sales too, but if they have any offline locations, or if you're doing anything where you think people might be searching from a mobile phone, which according to Google is more than half the searches, then you should know what your SERP looks like. And also, the point that I've been making most recently is a lot of the things that are happening in the search result aren't being reported to us anymore. Things like whether or not you have a featured snippet, whether or not there's these interesting finds or found or the web. Or, all of these things that could be pushing a ranking way down and causing traffic for a ranking to go down even though the numeric ranking that they show you on Search Consult stays the same. I think it's sneaky, don't you think it is?
Kevin: 08:14 It absolutely is.
Cindy Krum: 08:14 I think Google's trying to get away with something.
Kevin: 08:16 I feel like there's more and more of that kind of stuff happening, where, yeah, you just find yourself in a position where there's so many different factors that impact your rankings and the traffic that comes to you, and I feel like it's not always reported in the right way by Google. They do give us more data over time, but also they develop their SERPS or change the SERPS at such a faster pace, that it's almost impossible to keep up with. But, I love the tools and I think it's such a big important need that you will fill there, because it is so hard to see or to get a feeling for what the SERP looks like even at the other side of the coast. At G2 we're a bit more... we have offices on both sides of the US coast, and sometimes I get increase from sales people or other colleagues, who ask me, "Hey, why are we ranking, I don't know, in position seven here?" And I look it up and we're actually in position six. So, the SERP looks different, and it's getting to a level that is so difficult to grasp. And even for international research.
Kevin: 09:19 So, one thing that I'm doing as well is I look at global markets or local markets overseas, and see how we could expand there. And if you want to get a feeling for local competitors and what the search results there look like, it is tough sometimes, there are tools, but I feel that the ones that you provide are super helpful for that need.
Kevin: 09:37 Then there's also the App Store Rank Checker, which is another super handy tool. And one thing that I wrote about in Tech Boundary [inaudible 00:09:45] was that App Store scandal for the New York Times, where the accidentally, I assume, ranked, I think, tons of their apps, I think the outcome was that there were over six or seven hundred keywords where Apple had ranked number one with their own tools. Have you noticed that and what do you think about that?
Cindy Krum: 10:07 Yeah, of course I do. And Google ranks number one for their things as well, very often, but not as much, but something that you might find interesting is now, Apple is so conscious or self conscious of that, that they've stopped even having star rankings or download volume in the app packs. So, for instance if you search for podcast app, you might get an Apple podcast app or the Apple podcast player, and it has zero and zero. When all of the other apps have their star ratings overall and how many downloads they have, their Apple products have just stopped reporting on it, so you can't prove that algorithmically they shouldn't be there. Because what if in reality they have 2.5 out of four or out of five, then now you just won't know. So, there's no way to really give it a fair comparison.
Kevin: 11:06 Sneaky, sneaky. Yeah, I feel like all of these companies they pull such sneaky moves, what is up with them? Yeah, I feel the department of justice in the US might have a say or a word or two about that. And I feel tech companies got a bit more pushback, especially in Europe as well, so, yeah, lets see how that ends. But it is certainly sneaky, and it's tough to compete in those waters. So, one thing that I was curious about is, to maybe step back for a second and when you started MobileMoxie, you said you wanted to get into mobile marketing and focus on mobile just a bit more overall. What was your thinking back on the time if you recall? What triggered you to say, this is an opportunity I want to jump on?
Cindy Krum: 11:55 Well it seemed important and no one was talking about it. And I had just gotten my first smart phone and I was doing all these searches and seeing that the wrong sites were ranking, or sites were ranking with really bad experience, and I knew enough about SEO to look into it and figure out what those sites were doing wrong and how they could make it a better user experience, or how the right sites could do a better job of driving rankings in mobile. So, I just started looking to see that the results were in fact different, and that no one was talking to these people about what they needed to do to rank well in mobile. So I just started writing about it and researching it, because no one was talking about it.
Cindy Krum: 12:43 At the time, that was... Because I first started MobileSEO before I started my company, so, right about 2005, right when everyone still had Blackberries and before the first iPhone had come out. So web experiences were sometimes on mobile still text only, or very limited images. So, there was a lot that you could do to make your website better for mobile back then, because it was a vastly different experience. Now we have responsive design so it's very similar, but even back then I was talking about using style sheets to control the size and the rendering of things before it was called responsive design, it didn't even have a name back then. And back then people were still talking about .mobi, do you need a .mobi? So, I got up and explained for SEO that's a really bad play because it's splitting all the SEO value, and no ones going to link to your .mobi.
Cindy Krum: 13:46 So, it was the wild west, it was so new, and even Google, I mean Google has taken vast shifts on what they recommend for mobile over the years, and so, Google didn't know what to tell people. It was so long ago, they were transcribing certain webpages, they had a system that would just scrape all the text and show you only the text and a couple of images. There were settings that you could put in your http headers to turn that off, but you had to know that, and you had to know it was happening. It was totally new and also back then, everyone was just writing the same article over and over again about title tags and meta description, and I was like, "I'm not going to write another article like that just because all the SEOs write these articles. They're about title tags, I don't want to write about title tags.
Kevin: 14:44 Yeah, I feel like you have a really good eye for things that are happening everywhere but nobody talks about, you just keep on delivering on these things. And I think you certainly have a gift and a skill for that. So, I want to dive into how you got there in just a second, but something that fits perfectly into that series are fraggles. And you are the queen of fraggles, So, for anybody not knowing what fraggles are, can you please explain real quick? And also, explain how you discovered them, what was the thinking behind it, and maybe even the context or the situation when you realized what was going on?
Cindy Krum: 15:27 Sure. So, fraggles are a combination of a fragment and a handle. What happens when a fraggle ranks is that Google not only links you to the page, but links you to a section of the page and scrolls right to it. Often they're lifting that piece of the page into the search results. So, the piece that they lift we call the fragment, and the thing that makes it scroll down to the certain piece of the page is called a handle. What I saw was that, we can program handles into pages, they're also called bookmarks, named anchors, stuff like that, but what I was seeing was that this was happening without the jump link, that's another name for it, without the jump link being programmed in.
Cindy Krum: 16:14 So it seemed like Google is now able to not only find the most important part of the page that answered the query, but also create or overlay a locator onto it, so that it knew exactly where it was on the page. And I saw that, just because I do so much more testing on phones than most people do, and I was seeing a lot of jump links, a lot of jump to kind of anchor text in the SERP, and Google would turn the volume up on that to a lot of those and then turn it down, it was clear they were testing something. And then I just saw all of a sudden carousels of pieces of a page, where it would scroll you to this part, or if you clicked on this one in the carousel it would scroll you to the next one down. That was really common especially for Q & A and FAQ Schema, where there were multiple answers on a page that Google was trying to lift, and it would show you the best answer, and then other answers even though these were all on one page.
Cindy Krum: 17:22 And to me, I saw that as a fundamental shift, because we've always talked about one page, one URL, and Google can only rank one URL. Before I was talking about fraggles, I was talking about how URLs might be going away in the SERP, because Google can index so much more than just URL. So it kind of flowed really nicely from this idea that the URL isn't the only thing that matters, or having one page doesn't matter as much, because now Google can index apps and they can index non website content like TV shows and product feeds, and stuff like that. So, it's like those things don't need their own URL, and maybe these things on a URL don't need their own URL. So, that's kind of how it came up.
Cindy Krum: 18:11 So, I just had to name it, because it seemed, again, so important. But it also tells us things about how Google's crawling in indexing. And if you think about how bullish Google is on PWAs, but so many PWAs including demos that Google uses at their conferences, so many of them are single page apps. So, then you have a SPA on the web and you have a native app that has no URLs, or that you have to superimpose URLs onto with [EBLANKS 00:18:42] and app indexing. And so it's like maybe Google just can create locators, and that's how they're going to start indexing and servicing SPAs, and that's why they're casually callously using SPAs as a demo when they're a search engine, right? How could they lean so heavily on a single page app, and also be Google? It's like, they must have figured this out, they must have a way around it that we just don't know yet. So, that was the thought process. Sorry, I'm long winded, long winded answer there.
Kevin: 19:17 I want the long winded answer. I think that's super smart. And how do you think SEOs or webmasters in general should change their thinking keeping that in mind? How should they factor fraggles into their work?
Cindy Krum: 19:37 Right now what we're telling people is that they just need to lean really hard on all the speakable schemas, so how to FAQ and Q & A schema, because that seems like what Google is already putting into fraggles. But it also makes sense from a voice perspective, and fraggles overall make sense from a voice and a Google assistant perspective, because the worst experience you can imagine when you're talking to a Google assistant, is that reading the entirety of a webpage, right? And so fraggles allow it to find the right piece of the webpage and just read that to you, and the things that people interact with Google assistant on are usually quick answers. They're not looking to do a huge research project when they're talking to their assistant, they just want an answer, and they don't want a website, it's not the idea scenario to be like, "I have a website, it is https..." They just want answer.
Kevin: 20:38 Yeah, that's a good point. That kind of plays into the next question that I wanted to ask which is, what is Google's end game with fraggles? Or how can that fit into a greater strategy? And you already kind of explained that it is a reasonable component of voice search, which I completely agree with. Do you think there's more to that strategy, or do you think it helps Google in other ways outside of voice search?
Cindy Krum: 21:05 I think that there's more to it but it usually involves voice search. So, Google has made a huge media play, and many big companies are making huge media plays right now, because there is this power vacuum with all the cable cutters not spending their money on a monthly cable subscription, but now they can spend it on subscriptions with Google or Amazon or piece it out and spend it a bunch of different ways. But since those things, those kind of entertainment options are on demand, Google needs to be able to surface those well and surface those with information.
Cindy Krum: 21:42 So, two years ago at CES, Google had a really good example on how their connected TVs will allow you to not only search for media that's available in the Google platform with Google Play or YouTube or whatever any of their subscription services. But within a topic, so for instance if you're watching a biography of Alexander Graham Bell, and you're searching for the documentary, you can also have it bring up the Wikipedia and read you the Wikipedia. Or you can ask it questions about what year did this person die? And it can give you the answer to engage with you in a more multimedia way, while Google is presumably getting the money, or at least the traffic and the data for whatever it is that you're watching.
Cindy Krum: 22:36 So, we call that being a multifaceted brand. Which is like facets on a diamond, it's much harder to spam stuff like that. In the same way for years Google was like links are votes and it's hard to spam links. Well, turned out it wasn't hard to spam links. But having entities and concepts that are multifaceted where you have a documentary and you have a Wikipedia page and you have products and you have podcasts, that all takes work. And you can spam it for sure, but to do a good job and to have good engagement signals, which I think is what Google is trying to go for now in terms of spam blocking, it's much harder.
Kevin: 23:23 Yeah, absolutely. It gives me so many avenues to venture down, but yeah, I totally agree spam is certainly a huge problem that Google has been fighting since forever and will continue to fight forever, and then also more of an independency from links. Maybe the end game is not even to get rid of links all together, but just to add more layers to an understanding of what relevant results are. It also makes perfect sense to me that they want to get a better understanding of the different blocks of content that they could serve through voice search, and whatever comes after or whatever comes with it. So, when you create content, should you think in terms of where Press Gutenberg blocks, would you say that's an adequate way to look at content in the future, where you keep in mind how content could be decomposed and relevant just by... Content pieces could be relevant just by themselves?
Cindy Krum: 24:25 Yeah. So, Gutenberg, I know a lot of people really hate it. But the concept is spot on, in terms of what Google wants. Because if you think about also what Google has limited resources of, the web is getting bigger and bigger, and having to crawl the same pages over and over again with largely the same content is a waste of their time and energy. So Gutenberg allows them to crawl and find the ID of the module or the unit and say, "Okay, I've got that, don't crawl through that anymore." And then, when you piece meal it out, we have an index of this, this, this, and this, and even though they occur on 50,000 pages, we don't have to crawl them 50,000 times, we just have to crawl the unique piece, the 50,000 pieces that are unique, and not waste time, content, energy, or index space on the stuff that is replicated over and over again.
Cindy Krum: 25:22 I mean, it's such a smart thing, and I don't think it's an accident that a large portion of the web is using WordPress and that Google has changed how they index, and that this framework for a large part of the internet is also changing how they compile pages. I think that WordPress knows... It just makes sense that Google would get them on board, because they power so much of the internet to make it so that Google's job is easier. B think, do you see the conspiracy? I am there with the conspiracy, they all have some merit, and I think that's a good one.
Kevin: 26:13 I think you're spot on. If I recall correctly, I think they even have some sort of a partnership, WordPress and Google. And yes, it makes perfect sense I think to jump in on some of the conspiracies, I think that would make a ton of sense for Google to get a feed directly from WordPress to make indexing easier just like they get from Amazon. I have no way to prove this, it might as well just be in my head, so I want to get that caveat out, but why wouldn't they? It would make their life so much easier.
Cindy Krum: 26:49 Yeah. And it's not evil. I mean, even Google has backed off of their don't be evil mantra. But I think that it just makes business sense, and I don't think they're saying it gives you an unfair advantage anymore than doing a correct site map gives you an unfair advantage, it just is making it easier and less error prone to crawl.
Kevin: 27:11 Absolutely, why wouldn't they reach out to bigger sites? They wouldn't have to pay them, right? But they would reach out to Facebook, Amazon, WordPress, a couple other companies that have such a major stake in content creation, and just ask them for maybe an XML sitemap on steroids, and therefore provide them a faster indexing. And as you said, indexing is not ranking, right? It just allows them to have a bigger [corporate 00:27:37], but it does mean that any result is being prioritized. And then what we see with... I'm blanking out right now, but I know they have an API where you can index certain results super quickly, but-
Cindy Krum: 29:35 So to me we can't bank on that, it's just one dude at Google, but if every PWA creates a API that Google can just ingest the content, and the way the service worker works, is it does exactly what Gutenberg does, it separates out the pieces and it says, This is the most important piece on the page that you need to get, this is the unique content. Because remember PWAs have an app shell. And the service worker is meant to control what renders and what works even during an offline interaction. Well, that would be the most unique text and images of the page.
Kevin: 30:13 Right, absolutely. Yeah, it's funny how Google just becomes more and more of this meta-layer of the web, where they just do pretty much everything for you, right? They provide a better experience for you, they standardize experiences especially with something like AMP, and then they also standardize indexing. And as you said, why would they not? Why would they not become the Yahoo that's on steroids, where you come to them if you want to get your stuff indexed? It's just a much more sophisticated way. Why won't they even turn this around all together? And be like, Look, we're not going to crawl the web. They always will probably crawl the web, but they are in a position where they could simply turn around and say, "Look, if you want your stuff indexed, we need access to your API." And it would make their life so much easier, because it's structured, because it's controlled, they would understand when new content is uploaded, or when content changes all together. It would make their life easier and save them tons of dollars. I mean, all that stuff is super expensive at scale.
Kevin: 31:16 Couple of days ago we had Gary [Yeish 00:31:20], I'm trying to pronounce his name right. Yeish at the SEO meet up here in San Francisco and he talked about how just the localized versions of Facebook, how href rank calls this petabites of data. Again, Facebook is an incredibly large site, but it's not the only one of its scale. And if you have to fan [inaudible 00:31:47] that and crawl it around the globe, and index it quickly whenever somebody posts a status update or a business changes it's address, that adds up to a ton of dollars, and Google is under immense pressure from shareholders to deliver revenue, and one of the ways to make more money is to decrease cost.
Kevin: 32:08 So, anyways I think that's how it plays in, but I just find it so intriguing how APIs are becoming that kind of gateway into faster [nixing 00:32:17]. And I can totally see Google just basically forcing websites and businesses, because of their position to basically outsource indexing to sites via APIs. That's how I look at it.
Cindy Krum: 32:32 Exactly. Yeah. Well, and I think that that fits in really well with this concept that Rand Fishkin has been talking about a lot, which I feel like I've been talking about, but I never gave it a name. He calls it ON SERP SEO, and he's talking about all the zero click search results. This is something that we've been focusing on in mobile for a long time, and I've been talking about, what does your search result look like on the phone? Because the look of it will drag clicks or not, and that includes star ranking schemo, whether you have images, stuff like that. So, it's not just the text in the ranking, but what else comes with it. And a lot of SEOs are upset about the decrease in traffic, and understandably if that's the measurement of success. But it's an interesting conversation, is how do we talk those SEOs off the ledge, and what is the new advice? I'm interested to hear your thoughts on this, because I have thoughts, but you're really smart too. What do you think?
Kevin: 33:37 Thank you. I'm still trying to live up to your thoughts on this, so I'm trying to say something really smart. In all honesty and seriousness, I wholeheartedly agree, the mobile results are so different and Google is able to test so much more on mobile. So, to come back to that Gary Yeish, he also mentioned that when you would search for something, and I think he came up with this Deadpool trailer example. So, you search for Deadpool trailer, and you don't find a result that you are looking for, which would in the best format be a video. And then, you would click on the video tab, that that is a very strong signal to Google that this search or this query probably deserves some sort of video outreach media, Which makes perfect sense. And I think that is even easier on mobile, and I have my own theories about that, probably because people use mobile search results differently compared to desktop search results, and might even be less inclined to rephrase their query. There might be a steep theory, but that what it seems like to me.
Kevin: 34:51 Anyways, I've also noticed that Google has way more SERP features on mobile, where they just ask you, "Is this the results you were looking for?" Or, "Did you mean this and that?" So completely changes, and it's such an important thing to factor in, which once again, where your tools come in and are so useful. And again, yeah I think it really depends on how you look at the user journey, and how that applies to your specific business and vertical. And what I mean by that is that, I think we also have to forget about or throw this concept away of copying a desktop experience into mobile, and instead ask ourself, What part of user journey happens on mobile and what happens on desktop? Especially in the B2B world. B2C it might be a bit different, but in B2B I do see a lot of searches that are a bit more top of the funnel, or early on a user journey happening on mobile, with a completely different experience and then venture into desktop. So, there was kind of a bit of a ramble, but I'm curious about your thoughts on this.
Cindy Krum: 35:58 Well, I completely agree with you that people on mobile are less likely to rephrase. And I think it's partially because Google is doing things differently there. So, lets look at this query example that I like, if you ask Google who is the drummer for the band, The Cranberries? What it's going to show you is the knowledge graph, and you're going to feel like that was a good result, and Google didn't have to try very hard to get you that answer, that feeling of satisfaction, it just shows you the knowledge graph and then you can click around and see who are all the band members and what are all the names of the songs and the albums and stuff like that. And that's what you wanted even though The Cranberries didn't have a drummer. If Google still feels like it got it right, and Google didn't have to figure out that detail that there's no drummer, and the user is satisfied.
Cindy Krum: 36:54 And so algorithmically, it's a much faster query to say, "Can I exact match this with a knowledge graph, than to try and find the fraggle or find the detailed answer? So, Google's just getting it in the ballpark with knowledge graph, but users feel happy. And so, they're doing less work for a better result, and that's clever too.
Kevin: 37:22 Yeah, that's insanely clever, it gets me thinking really. That obviously also provides a lot of risk for false information, where they just try to algorithmically identify certain things, and then that could fire backwards really quickly, [crosstalk 00:37:40].
Cindy Krum: 37:41 Well, and it has with featured snippets, less with knowledge graph, because there's a higher bar to get into the knowledge graph, but I wonder, and I've been wondering this on stage for a while, do we think that featured snippets are part of the knowledge graphs, like they're sub nodes or they're nodes that are being tested for factual relationships. What do you think on that? Because to me there's some level of confidence to show something in a featured snippet. And they seem to know the topic, if you say, how to tie a tie, they understand what a tie is. So, I'm interested to hear people's thoughts on that and see where it goes, because I've always counted them as knowledge graph as the topic layer, but then there's this next layer that comes before bluelinks and it's featured snippets. And bluelinks are now the lowest quality result you can get, right?
Kevin: 38:43 Yeah.
Cindy Krum: 38:43 From Google's perspective and from the user's perspective. No one wants a bluelink.
Kevin: 38:50 Yeah, that's absolutely right, they certainly don't provide the same experience. There's something to it, I absolutely agree with you. So, what Gary Illyes also said. I'm trying to not always repeat what he said-
Cindy Krum: 39:03 He slipped!
Kevin: 39:06 Oh yeah.
Cindy Krum: 39:06 He says I think, sometimes more than he's supposed to.
Kevin: 39:12 Yes, absolutely. And I mean, I want to be respectful, I think what he does and maybe the paradigm shift a little bit, but I feel like he's trying to reveal more, or he does reveal more, which makes me anxious in two ways. So, one of them is, and maybe that's just my SEO suspicious mind at work, but I wonder if you give certain things away, does that mean that Google is so much further ahead that those things don't matter anymore? Honestly, that's my conspiracy theory here.
Cindy Krum: 39:49 I think that's true.
Kevin: 39:51 And it's a bit scary, because some of the things that he gives away, I didn't expect or didn't know, and they are very helpful for my work. But if I think that Google is so much further ahead that that doesn't matter, then that's nasty. Anyway-
Cindy Krum: 40:07 But also, I think sometimes he maybe is telling us what he suspects, with very educated background knowledge. But no one, there's not one person at the Googleplex anywhere, who knows the algorithm, right? It's too deep and too multifaceted, right? And I like explaining it, my new way of explaining it with all the, there's algorithms and there's sub-algorithms. And the algorithms already had industry specific or intent specific algorithms, but then they're stuck with that.
Kevin: 41:48 Yes, absolutely, I totally agree. He also said that there are probably millions of little algorithms at work, and that fits neatly into this understanding that ranking factors are getting more query specific or maybe category specific. So, there's just more customization even for the query, which makes perfect and total sense. So yeah, and then also, if you go to Google actually or to the offices you'll notice that the people who are working on search are not in the Googleplex, they're sitting somewhere completely else, they're completely shielded, Google is basically a company of, it's basically... I Mean, it's many companies, but in a certain way it's also two companies, as in that there's people who work very close to the core product, which is search, and they're completely shielded from the rest. Which is probably common practice, I would also suspect that Apple has designers or R&D people shielded off from everybody else, because it's such a scale of a company's size, you can just simply not guarantee that something gets out.
Kevin: 42:52 But yes I agree that Gary slips sometimes, and me as an SEO I celebrate that, I love that. And again, I have a lot of respect and gratefulness for him going out into the lion's den, and going to SEOs and kind of trying to help, and showing his opinion and his understanding of things. To be fair, he can only know so much, I know he worked on some search algorithms or search features himself in the past, but yeah, it's not that easy type of a deal where as you mentioned, Matt Cutts goes out and knows all the different things.
Kevin: 43:28 There's also an interesting New York Times article, or could have been diverge, where they had an interview I think with some Googlers and they mentioned that even kind of builds search after his intuitive understanding of what people like. And then completely shifted after him where algorithm people or machine learning people like Jeff Dean basically changed everything and now try to solve things at scale, which makes sense from their perspective as well. To tie the knot on this, one thing that Gary kind of reveals is that the algorithm that extracts the NTTs and work tokens from content to create featured snippets is relatively simple, and that there is a click component to that as well. But then he also mentioned that there is something that he cannot talk about that all of a sudden comes into play when they filter, for example, for a negative or bad tokens or spend me tokens to avoids the scandals that happened in the past with bad results that are fake or fake facts, and that there is probably a lot more that has also happened in the background.
Kevin: 44:37 And my theory is, and again I can only theorize here, that potentially there is a deeper connection between knowledge graph and the featured snippet algorithm where they might run them through the same filters and neural networks that basically are taken care of the entity understanding and the relationships. So I could see some bridge there. What do you think about that?
Cindy Krum: 45:02 Yeah, totally. I think that's right. We've been playing a lot with the Google cloud natural language API, which if you just search for cloud natural language API you can copy and paste text from your website and put it in and they'll show you what they understand from the text is an entity and what's not. So for instance, we've recently been working on a problem where a big brand came to us and said, "We obviously need to be ranking for this keyword, but we're just not." We looked and said, "Yeah, we agree, you probably should be there and your metadata looks right." We looked at who was ranking for the word, and popped in some of the texts from their website and their other assets, and the things we're getting identified as different kinds of entities.
Cindy Krum: 45:57 So all the ones that were ranking were this subclass this entity, but we were this subclass something else entity. And so our idea now is we need Google to really understand that we're the same as those guys, and they clearly misunderstood. How to go about changing the entity understanding, that's another thing, but I think we'll be able to do it.
Kevin: 46:23 Yeah, yeah, that's a really good point. I've noticed some of the same things where you have very subtle differences in we're tokens on the page and entities and I love the NLP API. I'll also add that to the show notes in case people can find it, but yes, I strongly recommend everybody to play around with that and see if there are certain differences in Google's understanding off your content. And I also think that Google punishes, not punishes, probably the wrong word. But Google has become way more aggressive of how they rank certain sites that do not fulfill that entity profile that they're looking for in a really aggressive way. And I think that some of the recent core algorithm updates have done exactly that.
Kevin: 47:09 So I've seen sites that have traffic and ranking drop, similar to Panda or Penguin in the past, we have a 50% traffic drop almost overnight. As much as technical issues can certainly be at play here, I'm not ruling that out, I think there is a chance that this comes back to the understanding, and Google is at a point where they are so good at it that they can just recklessly reshuffle the search results in some regards.
Cindy Krum: 47:40 Yeah. Recklessly reshuffle and revert, if the click data shows that it was a bad idea. That it didn't work out.
Kevin: 47:48 Yeah, yeah, you see that a lot, these kinds of dips and then comebacks with different algorithm updates, which is so crazy because it's kind of everything goes back to what it was in the past, right? I remember when you had to wait for the Panda algorithm to update or penguin even to kind of have a comeback, so we see that again. You had Yahoo, where you had to add your site to a directory and we almost kind of see that again happening. So it is funny how it goes back to that.
Cindy Krum: 49:28 So to me it's like, wow, they need to turn that machine on and off or something like... How is it that this keeps happening? But then what does this tell us about the way that the new crawling and indexing works, right? That they can have massive outages basically, that it takes a while to notice, and what does that look like on the other side of it? What are the Googlers that are running around with their hair on fire trying to fix it, what are they fixing? And how did they miss it?
Cindy Krum: 50:01 So that's all very interesting, because it's something that didn't happen before mobile first indexing they didn't say like, Oh we're sorry we have a problem in new content, we just figured it out, the bots would just roam randomly. But now Google seems to be efficient enough that they know when something's broken, and when it's their fault and not ours. Whereas I think before they just always, the bots would crawl randomly and assume that it was our fault or websites were bad.
Kevin: 50:32 Bet SEOs scan something and broke it. So what is the theory there real quick? How it works in the backend.
Cindy Krum: 50:39 I don't know. I honestly, I brought it up because it's something I've been thinking about that I really can't come up with an answer to. I'm sure they're smarter SEOs out there who have interesting theories. What do you think?
Cindy Krum: 51:48 Yeah, they haven't. They definitely haven't, and they're giving a lot of false positives. I think they think that they're better than they are. Because just the other day I was looking at a page where the whole homepage was covered by an interstitial which used to cause problems and probably still is at least not a good ranking signal, but it was like yeah, mobile friendly, this is mobile friendly. And for years they said that that's a big signal of not mobile friendly. So at least on their own evaluation, they're broken.
Kevin: 52:26 Yes, they absolutely are. And it's only agree to see these false positives where there are certain sites out there and I keep bashing them, and I will until I die because it's a horrendous experience. But there's certain sites out there who plaster ads everywhere and violates all the mobile friendliness guidelines. Then you have other sites where Google reports mobile-friendly problems in the search console and it tells you, your text is too close together, and I almost feel like that's a very standardized notification that they send to webmasters, whereas they should be able to much better understand how qualitative that mobile experience is. But anyway-
Kevin: 53:37 Yeah.
Cindy Krum: 53:39 It's really bad. Search for top 10 whatever, you'll find it I bet.
Cindy Krum: 53:58 Yup. Yup.
Kevin: 53:58 Okay. Okay. Sorry, just go ahead.
Cindy Krum: 54:01 It has been good talking to you. I know we're wrapping up.
Kevin: 54:04 Likewise, likewise. See, I really don't want to wrap up, but I feel like Barkley is going to be really mad at me and I kind of don't want to burn that relationship. So anyway, I could talk to you for hours and you're an insane gem to the SEO community and we're lucky to have you. So thank you so much for letting me plug into your brain API and extract some of those insane thoughts. It's been a pleasure talking to you Cindy.
Cindy Krum: 54:34 Thank you. And let me do a promo code for your listeners. If they want to try the tool, we want to make it Kevin all Caps?
Kevin: 54:37 Please, love it.
Cindy Krum: 54:38 All right, will do that [crosstalk 00:54:39].
Kevin: 54:39 And by the way, where can people find you before I forget, to plug everything else?
Cindy Krum: 54:45 I'm on Twitter @Suzzicks, S-U-Z-Z-I-C-K-S. And the site is mobilemoxie.com. So M-O-B-I-L-E-M-O-X-I-E.com. And we're also integrating the tools through APIs with all of the other SEO tools. That's really my goal, I'd rather integrate with the existing tool, so if you have a tool set that you want mobile search or mobile emulation, email me cindy@mobilemaxi.
Kevin: 55:13 Oh, love it, love it. I'll add the promo code and links to the show notes, and I'll connect you with a couple of people from different tool vendors, because everybody needs those tools in their other SEO tools. So thank you so much Cindy, this was awesome. Have a great weekend and a great rest of the day. Thank you.
Cindy Krum: 55:30 Thank you.