Google Announces 'Helpful Content Update'

MICHAEL:
Hi guys, Michael here. Do you want a second opinion on your SEO? Head to theseoshow.co and hit the link in the header. We'll take a look under the hood at your SEO, your competitors and your market and tell you how you can improve. All right, let's get into the show.

INTRO: It's time for the SEO show where a couple of nerds talk search engine optimization so you can learn to compete in Google and grow your business online. Now here's your hosts, Michael and Arthur.

MICHAEL: Hello, hello, hello. I am Michael Costin and I'm joined by Arthur Fabik. How are you going?

ARTHUR: Good morning. I'm doing good. How are you doing?

MICHAEL: I'm doing better than you. It's not even morning anymore. We've been trying to get this set up for about an hour now.

ARTHUR: It was morning when we started setting it up and now it's mid-afternoon.

MICHAEL: Yep. Yep. So basically I'm at home, Arthur's in the office. We tried recording Arthur in the podcast studio. There's all feedback. It's been a nightmare. It's been a tech disaster. But we've found a way to do this. Arthur's now talking into a little lapel mic.

ARTHUR: I'm sitting in one of the little booths that we have for meetings, holding a lapel mic up to my mouth. So apologies if the audio quality isn't as good as it normally is.

MICHAEL: But the reason we wanted to soldier on and have a chat is because something, some news dropped overnight in the SEO world. And that is that there's a new algorithm update rolling out next week. And for once, Google have given away quite a bit of info on it. So, you know, we woke up to this news, Arthur had just finished dreaming about the link he built the other day and SEO was on his mind and this big news dropped and we just wanted to get in here and talk about it. So we're going to run through the helpful content update. which is the funny name that Google have given this update. You think that's funny? You don't sound like you think it's funny.

ARTHUR: Any algorithm update isn't laughing matter to me because it means, you know, you know what can happen with algorithm updates, so I'm more concerned than I am humoured by it.

MICHAEL: Yeah, well at least this time they've given it an update, like normally they just go broad core algorithm update or something. This time the name is helpful content update. Might shorten it to HCU for the purposes of this podcast, but basically we've had a look at the release that Google put out and it really does just look like they're trying to target and clean up we think automated content, so stuff that's been done with AI writing tools and you know, there's been a lot of sites out there that have been getting huge amounts of traffic from just spinning up tons of what I would call garbage content with these tools. And it looks like Google are wired to that as well because in their words in the announcement, they've said they want to ensure people see more original helpful content written by people for people in the search results. So the TLDR of that is Google has seen everyone using Jarvis and GPT-3 tools to spin stuff. So now Google's going to crack down on it. What do you reckon? We haven't really been doing much with the AI though.

ARTHUR: We don't do it. And I don't think it's really going to target, I guess, the clients that we work on. That's just my wishful thinking maybe, but, you know, service-based businesses, e-commerce sites. I think this is going to be more focused on, like you mentioned, I don't know if you mentioned already, but like mummy blogs, recipe sites, sites that produce a lot of content and use that content to try to rank. Again, it's too early to tell. They only just announced it overnight. So I guess we'll have to wait and see.

MICHAEL: No, I think that's a fair bet. Mummy blogs, for example, that might have all sorts of really thin articles on nothing. Makes sense. I mean, you know, like recipe sites. You know, recipe sites where they just have… I hate it.

ARTHUR: You have to scroll to the bottom of the page to get to the recipe and they just fill it with… I don't know, I don't get it. I know why they do it because they want to rank but it's just… it's a pain in the ass having to scroll down and they just have the most annoying stuff anyway.

MICHAEL: What I think will be hit will be, you know, there's a trend at the moment with sites to spin up like tons of pages and then put ad networks like, you know, like media vines of the world on them and make money from page views. So they just try and get as much traffic as possible and make money off the page views from the ads. And the way they do that is by scaling up content creation using AI. So the pages are barely legible, just junk. But they've been ranking in Google and people have been making a lot of money off them. So that sort of stuff I think is going to be first in the firing line. Then maybe sites that programmatically create pages. So like they might use a feed or they aggregate content together and it's really thin. That will potentially be hit as well. So yeah, whether it's thin content, whether it's going to be the AI content, I think the hints in the The news update that Google gave is that it's more on that AI side of things that they're probably going to be looking. I found one thing in the update that they gave pretty interesting. They said, this update introduces a new site-wide signal. Our systems automatically identify content that seems to have little value, low added value, or is otherwise not particularly helpful to those doing searches. Right. So site-wide.

ARTHUR: So that's going to signalize the whole site if it finds that type of content?

MICHAEL: Yeah, like if you have a ton of really low value, like, you know, blog posts that don't really say anything or pages that have just junk content on it that's been created with AI, your whole domain is going to be sending a signal to Google that it's low quality.

ARTHUR: And I guess, yeah, one of the concerns would be if you do have, you know, programmatic pages rolled out, you know, it might, you know, like you said, penalize your whole domain. So something to keep in mind.

MICHAEL: Now, the good news is I also read that in that same update that they've given overnight, that a natural question some will have is how long will it take for a site to do better if they remove that content? And what Google is saying is that their classifier that looks at this stuff is running all the time, continuously. So it can monitor new sites and it can monitor old sites. And if it sees that you've removed this content and over time you're not sending that low quality signal, it will basically no longer apply to you. That site-wide penalty, let's call it a site-wide penalty, will no longer apply. So the good news, as I interpret it, is that it's It's not like other updates where something can happen and then you have to wait until Google does another update to fix it. You might be able to fix it by getting rid of the junk pages once it does. But I think we're gonna have to wait and see what happens, right? Like with all algorithm updates. Yeah. Don't jump the gun.

ARTHUR: Yeah, I was reading an article earlier today and essentially the way I kind of interpreted it was that it's going to target any SEO copy. So any copy that's written for the benefit of search engines to try to manipulate the search results rather than copy that's written for the end user. So I guess that's concerning to all SEOs because as you know, obviously we're writing copy not only to appease or to, I guess, resonate with the end user, but also to rank. So it'll be interesting to see how that plays out over the coming weeks. But it also made me start… I started to think a lot of the time for websites, you're writing copy and you're trying to rank a page essentially with the content and keywords. If this update is in fact targeting SEO content, then maybe in a lot of cases, rather than producing a lot of copy, you might scale back on the copy and use things like Natch. So we use natural language processing, Google's natural language processing and optimizing the content that's already there rather than adding additional content to try to manipulate the search results. Just a thought I had, maybe rambled there a little bit, maybe, but

MICHAEL: Well, the question to me is how does Google detect this is a big question. Because this release is saying it's targeting all sorts of stuff, but it's the general reading between the lines, it's like it's going after AI. Because in the release, it says it has little sort of questions that you can ask yourself about the types of content you're creating and, you know, Are you likely to be caught up in this? And the questions are things like, are you using extensive automation to produce content on many topics? So to me, that means using AI to spin up tons of pages on junk. They also are saying questions around user metrics to me. So one of them is like, does your content leave readers feeling like they need to search again to get better information from other sources? So to me, that's saying if people let go from Google over to your page and then back to Google because your page was like a junkie recipe article that has nonsense waffle before you get to the actual crux of it, then Google's probably tracking all of that stuff. Then after reading your content, will someone leave feeling they've learned enough about a topic to help achieve their goal? So to me, that means like if someone searched something, gone to your website, and then maybe going back to Google search again, they're not really getting, I guess, value out of your content. So Google's tracking that sort of stuff. Yeah. I feel. They also say they don't, but I reckon they definitely are. You know, how else are they analyzing those types of questions? No. Um, so you know, how, how, how does Google detect AI content?

ARTHUR: Well, that's a question that I've, we've talked off air over and over again. And it's something that does kind of blow my mind because you mentioned that they have all these algorithms and things in place that can detect AI content and you know, having used Jarvis and different, you know, tools that uses and incorporates AI and open AI. Is it open AI?

MICHAEL: Yeah. Yeah. Open AI.

ARTHUR: That's right. Yeah. And the stuff that it produces to me seems like something that a human would be able to write. And I kind of just think, you know, out of all the people all over the world, all the possible sentences that someone can produce on a particular topic, there is a likelihood that a human could have written or rewritten this sentence. So it's almost like how, how can it detect whether it was AI or a human that wrote it?

MICHAEL: Yeah. I'm going to give my best attempt at explaining it. I don't work at Google. I am not a AI guru by any stretch.

ARTHUR: Does that make sense though? Do you know where I'm coming from? Yeah, it makes sense. Because I read it. The stuff that it produces is good. I mean, I guess when it's writing content from scratch, it's hit and miss. But when it's rewriting content, for example, often I find it could be better than the original content that you've put into the tool. Yeah, it's just interesting.

MICHAEL: I would say that Google's looking at overwhelming use of it. So here and there, use of it, it's going to look like humans have read it. But if every page on a site is written in that tool, there are patterns and things that they look at to figure this stuff out. So we touched on OpenAI GPT-3. It is basically the the model, the AI behind all of these tools like Jasper and different AI tools that create content. And the way they work is you basically put in like a prompt, like you as a user, you put in maybe like keywords or the start of a sentence or something like that and then tell it to go and create content on the back of it. They create really legible copy, but they do tend to use a simple way of writing. And that simple way of writing, there's patterns in it that can be detected. Because there's actually a website, it's gltr.io. And you can go on there and it basically can detect automatically written text, or it says it can. So basically the way it works is it plugs into GPT-2, which is the older version of GPT-3. What you do is you put a passage of text into it and it will analyze what GPT-2 would have created at each word in that passage of text. And it will determine if it's from the top 10 most likely words that GPT would have picked through to the top 1,000 most likely words. What we're finding is like, well, not we, what this tool finds is that with AI tools, they tend to sit very much in a range of using like the top 10 to top 100 most likely words at each point in a sentence. Whereas if a human writes it, there's a lot more variance in like how likely the words are to show up in a sentence. So that's sort of like programmatically or sort of using computers can detect patterns that the AI tools are using that humans don't. Yeah, these detectors can be game though. You can put like a random word in here and there, so you could put in like sprinkles or cornflakes or something random in a passage of text and it will be enough to throw these detectors into thinking that it's been written by a human because you're using words that are unlikely to be in a passage of text generated by AI.

ARTHUR: Okay.

MICHAEL: Imagine what Google can find.

ARTHUR: Yeah, in tools like Jasper, you can set or select the tone that it produces the content in. So whether it's informative, formal, informal, funny, casual, and they all have slight variances in the way they rewrite the content. So yeah, look, I'm sure they have a way to do it. It's just, yeah. My small brain sometimes has a hard time comprehending some of the things that the algorithm can do.

MICHAEL: And look, I think with this, this is launching next week. No one knows what it's going to be like until it launches. First of all, wait and see what happens. And this AI detection I feel will be garbage at first and it will improve over time. Like all of this stuff when it gets released. There's going to be sites that are innocent that get caught up in this as collateral damage. So that always happens with algorithm updates. But I do kind of feel like there's sites out there where every single page on it is machine written. The whole thing just exists to get traffic from Google visibility and get page views for ad purposes. Yeah. I feel that they will already have it dialed in to go out and make sure those sites aren't getting visibility anymore.

ARTHUR: Yeah. I mean, fingers crossed, hopefully it just targets the big offenders and it actually does improve the search results rather than hitting innocent sites that might have used different tools to help write copy and they find themselves penalized, not only the content, but the whole domain, which seems like it's a bit… And I guess on that topic, I guess it makes sense what you said earlier, because if they're targeting the whole domain and they're looking for that that AI written copy, it's going to wipe out a lot of, you know, mummy blogs and general websites, which, you know, people use to sell links, which cover a whole variety of different topics, but very, very poorly using AI content. So maybe they are just after those types of sites and they're not going to, you know, target the innocent florists?

MICHAEL: No, they won't be targeting them. We didn't even touch on that. The sites that sell links, there's a lot of sites out there that exist more or less purely for link building purposes. And I would imagine a lot of them are just churning out AI content and will be smashed by this. So sites that previously would have… We spoke a couple of weeks ago about metrics you look at when deciding whether a site is a good linking partner. Yeah, organic visibility and organic traffic estimates were one of the two of the metrics. If they've been using totally automated junk content on there, they're going to be wiped out and a lot of sites that might have previously slipped through the cracks as a potential link will no longer be good for it.

ARTHUR: And what does that mean to all the sites that we're linking from that site?

MICHAEL: Exactly. So this has potential to cause massive upheaval in the search results over the next couple of weeks. Now I'm a little bit more worried. It's going to be very interesting to see. Yeah.

ARTHUR: But I mean, end of the day, just got to stay calm.

MICHAEL: Keep calm and carry on, as we always say. The cynic in me, the update is called the Helpful Content Update. And they want to make sure that people are seeing helpful content written by people for people in the search results. Unless, of course, it's Google's ads and all of their advertising stuff above the fold, they're going to be okay with that showing. They're not going to reduce any of that. And the cynic in me says like, what is this update really doing for Google's bottom line and then making more money? Because that's what it always comes back to at the end of the day. And that's going to be something to keep an eye on as this update rolls out. Who are the winners and losers? And how much more money do they make by Google Ads as a result of it? Time will tell. Time will tell, but for now, if you've been pumping out low-quality content with Jasper, you should probably be a bit nervous. For everyone else, sit on it, wait and see. It doesn't roll out until next week. Keep calm and carry on. There's going to be a lot more commentary and analysis that comes out, and we'll be sure to update this as more comes to light. But for now, happy SEOing. See you later.

ARTHUR: Keep calm and happy SEOing.

INTRO: Thanks for listening to the SEO Show. If you like what you heard, don't forget to subscribe and leave a review wherever you get your podcasts. It will really help the show. We'll see you in the next episode.

‍