From The Vault – Natural Language Processing for SEO

From The Vault – Natural Language Processing for SEO

From The Vault – Natural Language Processing for SEO

Episode 065

While we’re taking some time off over Xmas we’re publishing some of our most popular or interesting episodes from our archive. Enjoy this chat with Rob Hoang from Local Digital about using Google’s natural language processing in SEO and see you with new episodes when we return some point late January 2023.

Hey, Check These Guys Out

The SEO Show is brought to you by Local Digital – need more customers? That’s where Local Digital comes in.

Stuff You Need To Know

The SEO Show is released once a week so subscribe now wherever you get your podcasts and if you’re feeling extra kind we’d love it if you leave us a review.

Learn more about us at
Check out our YouTube content at

Follow Michael on Twitter @servicescaling or check out TikTok @servicescaling

Follow our agency Local Digital on Instagram @localdigitalco

Check out our content on Youtube



Michael 0:00
Hi, Guys, we don’t run ads on this show, we don’t try and make money from it, we don’t even really promote our agency on it. So we’re not asking anything of you normally. But I do have a little ask now, if you’ve enjoyed the show, if you’ve got value from it, if you could please go leave a review, wherever you get your podcast, it’ll really help us get this show in the hands of more listeners and help more business owners. All right, let’s get into the show. Hi, guys, we don’t run ads on this show. We don’t try and make money from it. We don’t even really promote our agency on it. So we’re not asking anything of you normally, but I do have a little ask now. If you’ve enjoyed the show, if you’ve got value from it, if you could please go leave a review wherever you get your podcast. It’ll really help us get this show in the hands of more listeners and help more business owners. Alright, let’s get into the show.

Local Digital 0:46
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 host, Michael and Arthur.

Michael 1:05
Welcome to a very, very special episode of the SEO show. Now I know I sort of say that every week, but this week, it really is special because after 23 episodes, we’ve got all three in the house. Hello. We’ve got bill in the House. And we’ve got Robin the house.

Rob 1:19
Yes, that’s me. Rob is FF guest. Hello? Yep. Oh, yeah, I’m good. I’m good. I’m good. Yeah, I think he asked you.

Michael 1:31
I’m glad to hear you. Good. Maybe I should explain a bit more about who Rob is minds here. Yeah, it’s a good idea. Well, today we’re going to talk about natural language processing. And Rob is a guru when it comes to natural language processing. Right. So maybe tell us who you are and what you do.

Rob 1:47
Yeah. I’m Rob, and the SEO specialist. Local, digital, and can confirm Yeah, confirm the man have any nicknames? Oh, yeah. Yeah, yeah. Yeah. I go by many names, including spade Devon. They call me defence. ridiculous, ridiculous and out of context

Arthur 2:08
would be good. Maybe? Maybe not. Let’s not go into

Michael 2:11
a bit of mystery.

Rob 2:13
Exactly, exactly. And most recently, they call me anihilation. nation.

Arthur 2:17
We still don’t know who they are, they will find

Michael 2:20
I have a feeling they. But anyway, the reason why bringing him on but you know, when he’s got that many nicknames, he must know a thing or two about language, right. And we’re going to be talking natural language processing, because this is a little bit of a cool thing in the SEO world. And it’s something that we’ve been having, you know, quite a bit of impact with when it comes to optimising our client websites. So much so that we built a whole tool based on natural language processing to make our jobs easier, which is where Rob comes in, he was the guru behind that. So if you want to check that tool out, quick plug, go to, and your favourite web browser. But anyway, Rob, let’s talk natural language processing cool names. Sounds fancy, but what is it?

Rob 3:03
So what is natural language processing? It’s actually it’s like a sub branch of artificial intelligence and computer science. The main use of it, I guess, will Google is understand, I guess, human language in regards to content. So traditionally, Google would look more like the keywords to understand and rank content, but now it’s lost model. And it looks at the user’s manual pay to understand the intent behind sentences the time behind, so whether or not it’s positive or negative, basically, and you can identify keywords and phrases in and, and assign what’s called entities like whether or not it’s a person. It’s an event based thing. Last thing, your organisation

Michael 3:55
things I think the big thing with that is, it’s Google going beyond keywords you mentioned he was before Google evolved, basically. Yeah, yeah, it’s part of them becoming Skynet. And soon there’ll be just talking to us as an, you know, totally natural conversation with Google. Sounds like our dream is SCI. Fi but like, basically, if you think about, like, we’ve spoken in the past about Google’s algorithm, or its crawlers reading content, and looking at keywords to figure out what that content is about. Yeah. NLP. Is that on steroids? Yeah.

Arthur 4:26
So taking the next step away from just stuffing keywords on a page, and actually making sure that it makes sense. And if there’s a standard Yeah, yeah.

Michael 4:35
So bit of history. This really became a factor in about 2019 When Google released Bert. So, SEO, the SEO world does what the SEO world does when that happened, panic. Have a mild panic whinge about it on Twitter, and then a whole bunch of different SEOs around the world will reverse engineer what happened and share that knowledge. And it’s pretty much widely agreed When that happened that was looking at content like, you know, the depth of content, the quality of the content, the matching sentiment to the search intent, you know, you know, the content matching that search intent. And basically, that’s NLP at play. So when we’re talking about it in a, you know, Google SEO sense, really NLP is in that whole Bert wheelhouse. Does that all make sense? right in saying that, yeah.

Arthur 5:26
Yeah. I mean, it makes sense to me, hopefully, listeners, but I think

Michael 5:30
so. Yeah. That’s what NLP represents basically. So you mentioned entities before, places, things, that sort of stuff. NLP is job is to select and evaluate these entities in your content. So like, let’s say, you put up an article about a movie, it’s going to read all of that, and then try and extract the entities related to that movie. So it might be hora, or category directors, that sort of stuff, right?

Rob 5:58
Yeah. I also just realised one thing, the but algorithm update. It’s part of my name. It’s part of my now. Wow, Robert, there you go.

Michael 6:10
It’s meant to be a new nickname has been born. They call me. I call it Bob. Yeah, Bob’s Good. Well, there we go. That’s the groundbreaking element of this episode. And

Rob 6:22
that’s it. It was it’s actually one of the ones.

Michael 6:25
So let’s come back to the concept of entities and salience right. Tell us about like, how that works. You know, what is salience? You know entities, other people, places, things like it’s just categorising the content. But what’s the salience aspect to that?

Rob 6:41
Yep. So there’s a few scores that Google assigns to that series. So the standards is, is essentially how important that entity is, in the grand scheme of things. When compared to the rest of the text?

Arthur 6:55
How does it work on a scale of zero?

Rob 6:59
It goes from a scale of zero to

Michael 7:02
0.0 to 1.0. Yeah,

Arthur 7:04
that’s obviously the higher the better.

Michael 7:08
Yeah. The higher the higher, the more salient it is.

Rob 7:11
Yes, that’s it. There’s also the sentiment score as well. And there’s also like a score overall, which kind of represents the tone of the contrast sentence. And the key word itself, of its entities itself. There’s also the magnitude, which is how important overall that that entity is. But yeah, so generally good was pretty smart pick up like how positive or negative a turn, that’s what this is. So if I say, input, something that says like, this, Michael is a is very happy. Amazing, amazing. Well, Mike was amazing. Yeah, Mark is amazing.

Arthur 7:58
Some sort of positive sentiment within that sentence, that gives it a highest score. Yeah.

Rob 8:02
And you will actually see like the overall sentence have quite a highest estimate. But on the contrary, if you do something that Google might not understand, and in terms of like context, maybe like Michael is not an ice on top for the moon, I actually Googled poems out there as well, like, I guess the more abstract, more abstract, and the Google might not be able to pinpoint that that’s a sign to that. And then we just give it like a more like, either neutral 00, or like, could potentially go negative, which, depending on your depending on your content, you probably don’t want that.

Michael 8:42
Alright, so basically, with NLP, it’s looking at the entities, it will categorise the entities, it will give you salient scores, magnitude scores, sentiment scores, all that stuff, which is all well and good. But I guess let’s maybe take a step back, right? Because this this happened a couple of years ago, this update. The reason we’re looking at Google’s NLP is it’s basically a tool or artificial intelligence that Google made valid makes available to developers, right, like you can pay Google to use this tool. So you know, back in the day when we sort of learned about this, but algorithm update, and then we’re seeing all this chat about NLP, we did a bit of digging and investigation ourselves, you know, let’s have a look at this tool. And we jumped on the NLP website. And the website itself was saying things like in the marketing material and the thought that like, natural language processing helps Google products like search,

Arthur 9:36
what does it say it plugs into the algorithm basically, in a way in a way

Michael 9:39
it says it says NLP sort of helps search like products. So it was vague but it was also saying NLP is involved in search. Yeah. So what do we do then? Like we’ve started testing, right? This is where speed demon bird came up with. Basically the I the way we use this This NLP in optimising content right? So maybe if you can walk us through the early days of us testing and playing around with NLP, how would we go about it?

Rob 10:07
Right? So the early days or having fun in Google’s little demo, tedious little demo, very tedious little demo, it was very tiny. They had a small preview window. The demo was on Wednesday, and then obvious was that, and basically, you will just import your what, in our case, the clients, the clients content from this side. And we just vary in terms of the process, it was very, very

Arthur 10:41
well, it would analyse the content and spit out all the recommendations. But you were dealing with a tiny little content box, probably 300 by 200. Pixel. Yeah, yeah. And it was extremely hard to edit within that box. So it was just a regular text editor. And then every time you wanted to rerun the content, you have to go back, and it was just a very manual process. Yeah. So you had to remember all the salient scores for each run? Yeah. What else? I remember that was a preview tool. So Yeah, makes sense. But there was also also you only had a limited run. So you could do it, maybe 1015, whatever, how many times before you got locked out. And then you just got to use a VPN, if you wanted to, can you continue doing it? So it worked by it was just limited in what you could do, basically.

Michael 11:26
So let’s say we might, we might take 500 words, a copy, put it in this box, where you can only see one paragraph of that copy in the box yet, run it through the schools, that’s an ending, well, maybe we want to change like a couple of paragraphs here to try and improve our schools influence the scores, and then you’d have to go back and redo it. I remember

Arthur 11:44
Rob’s notebook that he had, every time he would sit there and you know, write down the changes that he made and how much it impacted. What keyword Yeah, so it was super manual to the point where it was, you know, pen and paper basically, yeah, times,

Rob 11:57
I found, like Mortal ways of trying to do it and trying to find which way it works on. But sometimes I took screenshots, and then I’ll print it out, and then I’ll mock it all. Yeah, but it got confusing. Yeah, sometimes I’m putting the notebook, right data to see which one’s changed. I’ll put a plus one here, a plus to the this one shifted this way, I’ve made these changes.

Michael 12:17
Like analogue if Yeah. If we, if we ever go into like a no, actually, I was gonna say if we ever sort of go into a dystopian future with no internet, you could still do SEO on pen and paper.

Arthur 12:29
Do you still have that notebook? I do. Maybe one day, or something?

Michael 12:36
Maybe we can share a photo of, of the notebook on how Firewatch has evolved from going

Rob 12:41
from, from paper to now, digital masters.

Michael 12:45
So the big thing I’m hearing there, the pain points was the tools call where we’re using it to understand sentiment and salient schools and entities. But it was an absolute pain in the ass to use in any meaningful way, in a scalable way. So because let’s let’s, what we found when we were running, copy through this tool is that if we tweaked the copy and influenced the entity schools to be similar to like the top rank sites, and in a spacer, you know, for keywords, we were going after we were getting results, the rankings would improve. So that’s why we were going through this laborious process because you know, we’re able to get results for our clients. But it was a pain. So we wanted to be able to easily compare things and keep track of what’s going on without having to resort to printing out worksheets and writing on them and all the like. So we built that to address these pain points, right? Because no tool existed. Not yes, no tool existed that we could find that did that right?

Rob 13:42
Yep. After extensive research will stall just doesn’t exist.

Michael 13:48
So well, maybe maybe we can give a little example of rankings improving or you know, you know, I think you had an example of a, I won’t say the client’s name, but there was that client. Yeah. Maybe maybe give a walkthrough of the process of a process. Yeah. And the results that we saw, and why we felt that justified us investing as much time and effort and expenses we have in creating this tool, and then we’ll talk about the tool maybe?

Rob 14:14
Yeah, so I guess this particular client, can we say the niche doesn’t matter?

Michael 14:19
Yeah, they’re a lead generation business. Like a hire, they do hire, they do hire.

Arthur 14:26
They hire products. Yeah, that’s all you need to know.

Rob 14:30
Yeah, essentially, it’s the the website and I think in particular, as well, the client didn’t want to make too many visible changes in terms of the content that the user will see. So a lot of SEO content is placed beyond like a random walk or accordion. And, yeah, we spent a lot of time using the natural process. The match process to do to optimise his content, because the client had previous experience in SEO, or another agency, and was pretty turned off by the factor of like keyword stuffing and stuff. So their

Michael 15:11
copy in the past would have just been outsourced overseas and just read like absolute rubbish shoved on the site. Yeah, there you go. Your SEO is done.

Rob 15:17
Yeah, yeah. So I guess our way around that was saying, we’re using Google’s natural language. So we don’t have to resort to any like, you know, all this keyword stuffing. And so what happened really was around, I don’t know, probably hundreds of tests on the coffee as a whole. And the manual way, the manual way. And, yeah, I was comparing it to talking about as it turns out, the FCS could do that their current their previous copy, in terms of FCS was all over the place. And a lot of it was irrelevant to what the content should be about. So if this case, they’re a bit high, let’s say they had something completely irrelevant, like mobile phone design.

Arthur 16:01
So the the competitors, the I guess the entities were more related to the keyword that you wanted to rank for? Yeah, that’s

Rob 16:07
good. So yeah, for example, let’s say, since there’s the service is a product, then the consumer good entity type was what the competitive side. So we would had to the reasoning behind it, the logic behind it was that if we could get the entities, the entities out there, in that sent out like carry type, ft type, then they should rank well. And yeah, especially after we rolled it out. After various ad changes, we saw the rankings go up quite a bit. So two weeks after we bought it out, they were on average, on the sixth and seventh position. And two weeks after they won the Fourth Division, on average. And then two weeks after that, we brought out another change. And they’ve been steadily going up ever since. And now they’re in position to on average is doing very, very well. And

Michael 17:09
that’s weird, like, maybe a little bit of link building and the lag. But obviously, we’d roll these changes out in the site, you get Google to come back and re crawl it and rankings instantly improve.

Rob 17:19
Yep. So the funny thing is, is that this client has been with us for a while. So like, they’ve been continuing to do and like building stuff. So to isolate, I guess, yeah, that was one of the main, the main changes we did. And it was after the January core algorithm update as well. So I guess that kind of plays into the fact that, you know, Google was massive algorithm updates are likely going to effect holiday read content. So the timing of it just worked out, as well as

Michael 17:48
Yeah. And so with this latch, it’s very hard to test this stuff in isolation, because with SEO campaigns, there’s always other stuff going on, you know, technical optimization, your link building and the like. Were working, we just have no hypothesis, you know, you’ve run this content through the tool, you get the entities and family and schools and stuff in the same ballpark as a top rank sites. And you should improve rankings. And we’ve just seen time and time again, that that works. It does work. So that’s generally the reason we created the tool we did and the process that we went through to get here. So maybe we should talk about nuts and how it solved all of your woes when it comes to things like you know, very small preview windows and not being able to track changes and not being able to compare and sort of easily review what’s going on. So um, yeah, walk us through nach give us a little brief history of nach.

Rob 18:42
So nuts. Yeah, it’s, uh, yeah, it’s mentioned, there’s quite a few pain points. One of the one of the main things that was addressed with that is that it now has a huge, maybe window. So you can actually say, like, five or 10 paragraphs when you saw something in. And it just makes things a lot easier when you need to go back and you want to change something. Another cool thing that was added was the comparison function. So now we can actually you can actually run a test. And then you can go, you can, let’s say, Run Test B. And then you can go test it back to the original test, and you can just keep doing that.

Arthur 19:20
I guess. I was gonna say, I guess how it all started was we were just brainstorming. Like, imagine, imagine we had this tool that could do all these things that you know, we can’t do now. So basically, we rob started putting together a wish list of all the things that he would love, nach to be NLP testing to me. And that’s where it all started.

Michael 19:41
Yeah. So there’s tools built on real world SEO agency process being improved and optimised, I guess. Yeah, maybe we should say the name NAT obviously, is a slang term for natural natural language processing. We just don’t match with short and sweet. We can say get rid of that. content being not as yet. Yeah, or this kind is going to come on board. Let’s match them. So you know, works pretty well. Yeah, I feel

Rob 20:06
it does. So. Yeah.

Arthur 20:08
Anyway, it all it all started from a conversation really? Yeah.

Rob 20:13
Yeah, I think yeah, we brainstormed as a team as well like, what we would want, and how to make the process a lot easier. And yeah, it just came to fruition.

Arthur 20:23
Yeah, one day, I think he just said, Let’s build it. Yeah, pretty

Michael 20:26
much, pretty much. So let’s, let’s talk about what we’ve built, like he put the content in, you run it through notch. And then it will spit out all of the entities and the salient scores and all that good stuff. Yeah. Then you can maybe tweak the copy based on what you saw there and run it again. And it will show you what all the new scores are. But then you can compare the two and it will show comparison. So like next to each of them what the scores are and how they’ve changed. And you can segment it down by different entities and see where they are in the text and the like, and very visual and quick process to get a feel for what’s going on with that content. As you change things.

Arthur 21:07
That’s a really cool feature of it. So being able to click on an entity, and then it highlighting it in the text. So you can see where they are on the page and how they’re grouped in Yeah, you know, that makes a massive difference, because you can then try to group different, I guess, clusters of keywords together, or different entities together, which you couldn’t do before, it was just impossible to kind of maybe not impossible, there would have been super hard to do that. Yeah, without it

Rob 21:29
like before, like the game because Google likely classes a lot of the entities together. And yeah, basically be able to click on it, and highlighting the ones that are in that particular group makes a lot easier to even optimise for that entity. So you can go through and see what individual sentences are impacted. Yeah.

Michael 21:50
I found that really helpful. Yeah, even just the ability to leave, like each round, you can put notes on it. So you can sort of keep track of what you were thinking at the time or, you know, yeah, probably that’s that’s sort of how I view that just dropping out there, what I was thinking or what I was planning to test at the time, very visual, it’s all stored there. And you can go back and look at it in the future, if things change,

Rob 22:12
or the author can keep track of the changes made. So if you’ve made a change to the copy of content, and you compare it to the original, it will actually show and highlight the text changes. So

Michael 22:26
right, yeah, little green highlights the new stuff red for stuff has been deleted. That’s it. Very visual. Very cool.

Rob 22:32
Exactly. So like, because sometimes you’re not going to be doing all the tests in one day, like you’ll be doing it across multiple days. It just makes things a lot easier to keep track of, and then we’ll move on to the client site, once you’re happy with it.

Michael 22:45
Maybe we should talk about some of the interesting things we’ve seen when changing tax because it’s, you know, it can be as little of a capital letter or really changes, yeah, the scores or the entities and has an impact on rankings. And that’s why we like doing it the way you were doing before with notes and all that. Yeah, that’s why this new version with like a sort of visual tracking of what’s been changed is so important, because it can be such a minor thing that’s changed, right? So maybe, yeah, what are some examples of that? You know, that goes on? Or?

Rob 23:18
Um, yeah, let’s see. Yeah, like you mentioned, capitalization actually goes a long way, when it comes to even Google understanding what type of Etsy it is. So for example, let’s say I do flowers, for example. I do like owl flowers, lowercase, lowercase f, but then it was, I’ve seen that show up as just a regular other entity. But then if you capitalise the F, it will show as a consumer, good, et Cie. So even that, like it changes quite a bit. Some other ones could be like putting like an N dash or an M dash, instead of full stop. So you’re kind of extending the sentence. So it’s another I guess, little tip that you can sort combine citizens together and allow kind of change quite a bit of how Google reads. So there’s a lot of really small things like I think I’ve even seen, like, it’s not fun and code. It’s at the semicolon, but it’s got a little comma at the bottom.

Arthur 24:20
Ah, the semicolon with a little.

Rob 24:23
Yeah, it’s basically on a naked Pay button. Yeah, I don’t know. Basically, yeah. So there’s a lot of these different little symbols that you can use, and it would just change how Google reads it.

Michael 24:34
Which is funny, right? You wouldn’t think that those things have the impact that they do. But they do they do. So what we should make clear, I guess is, you know, there’s tools out there in the SEO world where you can check in a URL, and the tool crawls that and then says, Go change all of this and it should get you in the ballpark. This tool is much more manual like it. It’s part art part science, you know, this tool makes it easy to see what NLP is saying, the entities and the salience and all that are easy to track everything. But it relies on you as the SEO to use your thinking and logic and I guess creativity to change the copy and change things up in a way that you think is going to impact things and test it and keep trying. And it’s not just a, you know, one button and done process, right. Yeah,

Rob 25:23
a shelf process. I think like experience using NLP plays a big part of it, because the more you use it, the more nuances you pick up. And yeah, I think that will really differentiate between the people that use it all, because obviously, it doesn’t spill out into recommendations. But yeah, I think the more he is that, the more you’ll kind of have a little bit of an understanding of how Google is interpreting content, but you also understand you also find things that are gonna be like, what,

Arthur 25:50
why? Well, you’ve got your little hints and tips cheat. Yeah, sure. Not gonna give away. Yeah.

Michael 25:56
So we’ve got all internal processes that Rob has built based on his experience, doing a lot of this stuff that is easy to use. But it’s never going to be a tool where you just hit a button, and it gives you recommendations, like it’s a tool for you to get in there and grind it out and find that really fine tune piece of copy that is, you can Google all it needs to see to be the same as the top rank sites. But yeah,

Rob 26:21
I think it also, it sounds like explain to clients.

Arthur 26:24
Yeah, I was gonna say that, yeah. For me personally, being able to show little changes and how much the impact of certain scores of different entities has been massive, because a lot of the time without being able to share that to a client and you send them some copy with some minor changes. And they’ll look at it and be like, Well, you’ve just barely changed anything. And they might think you’re crazy. Yeah, if you can kind of demonstrate that the changes you have made to the copy, have, you know, influenced the scores and these keywords and the magnitudes and all that, then it starts to make a lot of sense. And every client that I’ve shown it to so far have had been really impressed. Yeah, it has to go well, like

Rob 27:01
Yeah. Especially when, like you can they can see how many test runs? Yeah. Then. Yeah, the client, it will help the client understand how much time goes into that. Yeah. And optimise. I

Arthur 27:12
think that’s a big thing as well, because they don’t realise how much time actually goes into it. You know? How did we actually get to this point?

Michael 27:18
Yeah, you just read an article? Well, that should have taken 20 minutes.

Arthur 27:21
Yeah, really? We spent two hours? Yeah. busting our brain trying to try to get these entities to push up. So yeah, it’s for that data learn. It’s amazing.

Michael 27:31
Yeah, so it is a, it’s still a manual process. But this tool, it just makes it much more pleasant to do. Well, maybe that’s a good segue into talking about the change monitoring aspect of the tool, because like, what we’ve found with clients is they are prone to just adding things to sites or changing things, or deleting things, or all sorts of different modifications to a site. And that really hurts when we’ve been notching the copy for hours and getting it perfect to have things changed, they might go in there make wholesale changes to the copy. Yep. So another feature we added to the tool is change monitoring, where once we have fine tune, copy, and put it up on the client site, we enter the URL of that page into the change monitor, and then natural go and crawl that page and keep a log of it in the database, then it goes back daily, and crawls it again. And it’s looking at things like meta title tags and description tags, and page copy and heading tags and robots tags and XML sitemap, whether it has HTTPS, the download speed, like all sorts of key SEO factors, and it keeps it all in the database. And then if anything changes, it will alert us so it alerts us via email. We’ve got integration with Slack, so we’ll get pinged on Slack, we’ve also got a Zapier web hook, so you can build all sorts of automations on the back of it. But we sort of think of this as like an insurance policy for your website, right? Because if you’re investing all this time, particularly as a client, you might you’re obviously not always going to be aware that all this effort has gone into it. And so sort of finely tuned. So you’re gonna change things and undo it. And it could be weeks or months before it’s picked up, then I can tell you some stories about Oh, yeah. Oh, yeah. So and as an SEO you, you want to be proactive, like you want to be on top of it. Definitely knowing. And so you can reach out and say, Hey, we’ve noticed you change this? Yeah. Well,

Arthur 29:29
it looks great. Yeah. And like, in reality, I look at client sites every day, but I don’t look at every single page on this site every day. So it could take me upwards of a week to or longer depending on the site. Yeah, to know that they’ve changed something. And by that point, it could be too late.

Michael 29:44
Yeah. And when we’re talking like full stops and capital letters, having impacts on these salient scores and entities and the like, if the client makes a change when you’re not going to pick up on it, and it could, it could be until their rankings go backwards. And then you got to try and reverse engineer what’s happened?

Arthur 30:00
And by that point, they’re mad at you.

Michael 30:02
Yeah. What have you done? Yeah, I don’t know. And then it turns out, they’ve changed. So that now makes sure that that doesn’t happen. So I guess that we’re sort of trying to look at the whole lifecycle of this process, right? We get the client on board, we’re looking at the competition, we’re trying to improve things, get it in the same ballpark, get it live, and then make sure it stays that way. And things don’t change. Yeah, that’s just taking care of all of that,

Arthur 30:25
because you mentioned that it shows you what they’ve changed as well, yeah, kinda Yeah.

Michael 30:30
It’s, it’s sort of the same process as the, the NLP tool in that it will highlight you can, it has the original snapshot, then it keeps a log of all the snapshots of changes, and you can easily compare them. Yep. And it will show you what’s changed, has little red, and green indicators. And it just makes it easy to jump in there. You can even just click View All the copy on the page and see what the copy is on the page and what it was and what’s changed. So it makes it really good to quickly fix things to be proactive to reach out to clients. If you get pinged on Slack, that something’s changed, you go and see it, and then quickly call the client and say, we’ve seen this has changed. They’ll be like, wow, like, you’re really on top of things. So just from a client services point of view, as well. It’s a really cool feature.

Arthur 31:16
It is. It is sounds like you’ll love it more than match. Or NLP,

Michael 31:22
it’s all match. It is not she’s a multi faceted FCA suite. So um, I guess, what else is there to chat about? Really? Any any takeaways you want to talk about when it comes to natural NLP?

Rob 31:37
Yeah, good was very, very interesting. It sounds like it’s Evans concepts. You’ll say things like, Why? Why? Why? Why is it? Yeah, definitely like that.

Arthur 31:49
Sometimes you scratch your head. Why?

Michael 31:51
They’re not as smart as Yeah, might be made out to be? Yeah, well, it’s

Arthur 31:54
a bit of both, you know, you look at any like, wow, the algorithms really evolved. And then you do things like add a random force stock to a sentence? And you’d be like, Okay, maybe not. Yeah.

Rob 32:02
So, yeah, it’s pretty. Yeah, I mean, it’s still in, it’s still evolving. So I mean, it’s very interesting to see that AI has gotten, like so far that it can understand the tone behind it, because they’re listening

Arthur 32:15
to us all the time. They can, you know, capture all that data and analyse and get smarter. Yeah.

Michael 32:22
Well, I think that’s been a good intro to natural language processing, where things about like, it’s a bit of a, as we said, part art part science, like a lot of things in SEO, like we haven’t done hardcore scientific testing to prove anything in this. It’s more, we’ve had a hypothesis, we’ve done a bit of testing, and we’ve found results are positive, but you can’t put your finger on exactly what’s going on with it when you when you do it. But we have found the processes, that if you can make your copy, similar in terms of entities and salience and magnitude and all the rest to the top rank sites, generally, you’ll see a ranking improvement, so just worth the time and effort to do it. Right. Correct. And it’s, it’s where Google’s headed, moving away from keywords, keywords are still an important concept, but so too, is NLP and understanding sentiment and all the rest of it, and it’s only going to get more important. So if you are wanting to be on the cutting edge of SEO, then playing around with NLP playing around with the preview tool or going to and having a look at the tool we’ve got there is well worth it. Right.

Rob 33:28
That’s it. That’s so nice today

Michael 33:32
for you there as well. Rob, how was it? How was your first time on a podcast?

Rob 33:37
Yeah, pretty good. Yeah, I mean, this is very, very fun. I’m the first guest should be back.

Michael 33:50
You might become you might become a repeat guests all the time. Yeah. Well, anyway, guys, thank you very much for listening as always. And if you like that, like Rob, Rob’s crew that are tuning in, give us a give us a subscribe, give us a like, give us a review all that good stuff. Yeah. And until next time, Rob.

Rob 34:11
Until next time, happy see you guys next time. Definitely see you guys next time.

Michael 34:14
We’ll see you back at the office. All right, thanks.

Transcribed by

Meet your hosts:

Arthur Fabik


Michael Costin


Type at least 1 character to search
Everywhere You Listen: