LIVEstream Workshop CMSEO2019: Content Vs SERPs - Rad Paluszak

Key Takeaways

[Music] [Music] ladies and gentlemen wear sunscreen if I could offer you only one tip for the future sunscreen would be it the long-term benefits of sunscreen have been proved by scientists or as the rest of my advice has no basis more reliable than my own meandering experience I will dispense this advice now [Music] beauty of your youth until they've faded but trust me in 20 years [Music] photos of yourself crap now how much possibility lay before you and how fabulous really [Music] don't worry [Music] hello testing alright everybody we're probably gonna get started in about three minutes so grab your drinks get your stuff find a seat and yes get ready for the show Brad you good [Music] [Music] [Music] [Music] [Music] [Music] [Music] alright alright is everyone getting settled in good good good whenever you guys are all sad just go ahead and find a seat no matter what we're just gonna get started anyways first wanted to just say big thank you for everyone coming out it means a lot for all of you guys to pack up thank you thank you thanks so much for coming out yeah like I was saying those means a lot for a lot of people to come out here to chiang mai thailand this obscure little place in the north of thailand just a geek out about SEO i love this stuff so it means a lot that you came out also thanks a lot to everyone tuning in on the Facebook or the YouTube stream so hello everyone thanks for everyone for tuning in so in a moment I'm gonna pass things over to rad I suppose a lot of you guys had the app installed right let's get a show of hands who has the app installed good ok so the way we're gonna do Q&A questions is not receive a phone call right now but go to the app and scroll over to where you see this presentation content versus SERPs and what you're gonna do is you should see and ask the speaker button at the bottom now when you click that you can ask the speaker a question or if you don't have a question you can see who's asked questions already an update upload them and download them and then we'll go from the top at the end so that's how we're gonna handle the questions this is a beta test before the real conference so hopefully it doesn't blow up all right right so I'm about to hand it over to rad rad is the CTO at the search initiative the agency that that I'm a part of and hands down he's one of the best SEO sino why is that because he's a technical genius he's a master at the technical he's a master at the nuances and he understands algorithms and patents and what he's about to talk about today is how to apply these kind of his brain his thoughts to content so if you were here last year you probably remember rad was the first speaker and he killed it hit it out of the park with technical SEO so let's give her a big round of applause to rad and kill it man lovely thank you very much Matt really appreciate it so okay guys first of all what we're gonna be talking about today obviously content versus herbs what Matt already said we're gonna indulge in this with a short introduction then I will present to you stuff about intent and sentiment I will introduce a few Google patents that I had a look at and I selected specifically to kind of like show you and maybe get you're interesting in this stuff because they're worth analyzing then we will dive into rankbrain quickly a little bit of a different look at rank brain actually maybe something that not every one of you already know so so that should be interesting natural language processing I will talk a little bit about neural matching and recently introduced or announced Google Bird then we'll have a quick look at search quality rater guidelines because that's a document that also gives us a lot of inputs on intent that Google wants you to optimize for then we'll have a quick look at sentiment it since herbs because I believe this is a lot for the SEO so actually analyzed in terms of the sentiment of the Serbs and the language of the Serbs and then we will go back to mainly user intent I'll talk about the classification of the user intent that Google suggests and not only that were Google suggests and then we'll go through a few methods of the discovery of of the user intent then we'll dive into the latent intent which is like a delight intent I will talk about it in in more details there and do a quick summary and that's really all now raise your hand who saw me a Chiang Mai last year ok quite a few people so I'll do the full range show them and for all all the other ones as much said I haven't actually updated director of SEO informally a director of SEO but also acting as a CTO chief technology officer the search initiative I started doing SEO back in 20 2010 it was a round caffeine update so basically all the update all the Google updates are basically surrounding me from the very beginning I started my IT adventure as a programmer as a web developer so I'm very technical I like algorithm I like machine learning I basically really like geeking out about the tech stuff and data mining and all this and overall I I believe that saying technical SEO is boring it's not really the thing because if you put technical SEO and the high end strategies you actually need to build up to rank the site you know that that allows me to be a self-proclaimed technical SEO artist because I think there is a big artistry element in that if any one of you have any questions obviously outside of the app outside of the talk drop me an email through the TSI email address and we'll be happy to answer now why do we care about Internet sentiment basically I need to I need to tell you something guys I always do this I always make those presentations and I stuff them with so much stuff that I really need to run through all of this and you know speak quickly talk quickly through the slides and all that and this time I was like hmm how would that be a fight is basically gave you of the resources the main resources I read you know Google patents that I had a look at and then I'll be like yeah that's it five slides in thank you very much let's all go grab a beer right whoo who wouldn't be who wouldn't be a hot forgetting semester s just right now raise your hand who wouldn't be okay a few people good good so anyway but yeah I did it again so I packed it up with a lot of stuff now why do we care about Internet sentiment overall we can see mentions of this everywhere right main thing if you're watching any Google page and Google has like thousands of them to relate to the to the user intent obviously not all of them are applied in the search and stuff but it's worth knowing how they perceive that how they look at these things obviously with good rankbrain that is basically the whole engine built around content and around user intent as well right and then we've got the natural language processing which again user intent is a big big big part of now let's have a look at those those four elements as I said starting with Google patents and as I said I selected four of them not all four of them apply to search or have any relation to search this one for example is the predicted travel intent right so I wanted to basically bring your attention a little bit to the fact that through reading your calendar data your messages search queries social network data and other things that you're doing on your phone Google is basically trying to predict if you're going to travel and I'm sure pretty I'm sure mostly like probably all of you or a lot of you who are using Android phone or are using Google services have seen this are you at mellow ship do you know this place or oh you're today's commute is going to be like five minutes longer because there's traffic right this is where Google actually predicts your behavior and like predict your travel intent right so overall Google is spying on you right nothing else it's just spying on you right but I'm not a big fan of conspiracy theories and stuff like that I'll let that sink in though I've had a chat with a few people about that and they were like yeah tell me something I don't know why Google is spying on us everywhere basically right and quite a few people that I actually talked to are okay with that because that brings a better service from Google because Google knows where they're going perhaps you know they will save them some time or find a shortcut on their way but yeah let's just let's have it I think in now another patent that Google has this one is related to the search they're actually determining user intent from the query patterns right so considering how you search and how what things you search and how your queries are changing there try to predict what it is that you're exactly one would it what is it that you're actually looking for if you're doing some research or something else and I'll actually start adjusting the search results based on your behavior and based on the queries that you that you that you ask Google but this in this patent I brought it mainly because there's a really cool indication of what elements are actually inside the main search engine right and we can see that this indexing engine ranking engine but also rank modifier engine which probably or most likely applies the filters to the searches search results and there is also the intent identifier so I ought to say that intent is a really important thing in Google algorithm if someone had any doubts over it then that's that's just there right now another another one that is strictly related to search and we probably use it and see it every day so Google is clustering query we find by inferred user intern whatever that means I don't know but basically Google builds a graph based on what you're searching for and your main query is the initial node of the graph and then sub queries are subsequent nodes but based on those subsequent nodes and the refinements in what you're searching for Google is actually building the related searches right so if anyone was wondering how the related searches are actually coming a coming alone along it's not just you you know asking different things about different things it's actually how users are refining that stuff right and if you if you if you if you say that here right based this is this would really this would really create a related search for those two refinements right it's it's topical its semantic so this more than just you know what words what search volumes users are asking and also in the same pattern we see in that a similar graph diagram so again indexing engine ranking engine and search suggestion engine which is also apparently a part of the whole sales search system so the base would be the core I'll go now that patent I just got it up because it's really easy to decipher obviously and it makes sense a lot of sense what that stuff says that I just copy pasted it from the abstract it actually relates to the paid ads so it's mostly most probably not related to the search however it's quite interesting because that Katyn tells you that Google is building your own individual behavioral profile and based on that is trying to monetize on you pretty much so it shows you better ads it shows you more relevant ads and it's through that increases conversion right but there's a few things that brought my attention specifically so obviously there's a big user understanding unit so Google tries to understand you try to understand your interests and the information that you're analyzing and and stuff and your events as well so I would go back to the travel intent thing because it's again about your own events then we've got social media and third-party content analyzer and I again don't wanna like you know make you hate Google or think or what is is Google saying or doing so I'll just let it sink in again that they're using you know different third-party social media platforms to actually analyze your interests in your behavior right and the one in the middle this one is specifically close to me because whenever I hear or read something about taxonomy I'm thinking rankbrain because taxonomies and the universe of vectors and stuff like that this is very related rankbrain so might be something that might be really interesting and inapplicable for said now obviously there's a lot of patents I encourage people to read through them I need to say out loud that not all of them are in use so that the fact that Google is actually granted the patent doesn't mean it's in using search or in any other service but from reading this stuff you get to better understand the technology that they have at their disposal and obviously it allows you to better understand and better decipher what they can actually do with that technology and how they can evolve in with the algorithm right now let's look at rankbrain so obviously most of you know that rankbrain is is the third most important ranking factor it's apparently an AI algorithm that was confirmed back in 2015 and its main job is to grab to match your query into a concept and then even if it hasn't seen that query before based on the concept that it market to it can actually pretty relevant results now this is what a clink Oh Brian Dean says about rankbrain so he actually presents it as as an adaptive algorithm that measures user satisfaction first and then it tests like small changes to the algorithm like in this case for example increasing the relevance the importance of backlinks for that query and then measuring the user satisfaction again and that and then leaving that amendment in the algo for that query in the I personally don't quite agree with that however if you've seen my presentation last year this is what I showed so there is a big part that is about the users query or a result result satisfaction but this is more to do with the actual relevance of an rankbrain itself I do agree however that Google's algorithm especially the core part of the algorithm is basically self adapting so it's an adaptive algorithm however not necessary I wouldn't really attribute that to rank brain itself so going back to another another decision tree that Brian Dean suggested I would say that obviously it does work this way so there is a user query but then rank brain as I said also and I agree with that map's the calmed the query to the concept but this is pretty much where rank brain where rank brain ends its main job right so all the ranking or the relevance for that concept happened in the core algorithm right it's I would say this is actually outside of rank brain it's actually deeper in the in the core Corral go so you've got obviously we've got the results for that concept and then the user satisfaction measurement and if the user is not satisfied the algorithm will probably try to get to show you another page or user for the same query or for the same concept another page next time and if the user is satisfied it would probably try to up rank the page that they link that they click through to and now I know you what your Dodge SEO or so thinking you're probably thinking how do i how do i trick it how do I make it for clicking through the results who actually uh prank my page so I would say you could but you can't for practical reasons and this is because Google has too much data for you to actually play with that with that element right so if you think about it it wouldn't be sustainable it it wouldn't be worth it you'd have to pay more for all these micro workers and stuff like that by the way friend told me done actually doing it do it doing it doing it yourself and having it sustainable now in all this however this common denominator and the common factor and this is CDR right so again I based that research on a study that's Brian Dean published together with word stream they found a correlation with increasing CTR in your search results mainly meaning that you basically increase the satisfaction over the users by only three percent and having all going up by one organic position right which is which is pretty good right so again if SEO somehow friend told me would have found a way to trick that that would probably be that will probably be effective and this is another thing that is correlated with higher CTR this is also higher conversion right but then it's not directly related to the search itself but it's related to user satisfaction which is related to user intent which is related to the search itself right so in turn + CTR is your conversion issue rankings is your money right and this is an example of one site of of our clients where I would lie if I said we only optimized for the intent Oh city are here we also did a little bit of of conversion rate optimization but overall as you can see that was quarter-on-quarter we've got only 14% increase in the number of sessions but actually 56% increase of the monetary value that we brought to the client and we actually achieved that mainly through designing people also buy that product with suggestions so we helped the client to build that suggestion engine for the products related to the one that client is currently buying or adding to the to the basket and looking at that just think about what would plus 50 56 percent of your income mean to your business probably a lot right now let's have a look at natural language processing so natural language processing is pretty much everywhere from Google Translate and machine translation through intern discovery sentiment analysis all the content accountant extraction stuff like that because even if you think about it good for Google to to get those search snippets out of your page and form of meta descriptions they also need to analyze the content right now my favorite tool to play around with with natural language processing is IBM's Watson there's a free demo you can use you just plug in the URL if you want to and first thing you see is the overall sentiment so obviously if you compare your site with let's say 10 other competitors that are currently ranking in serbs and your sentiment asserts the 50% level they're like around 90 then obviously maybe you're just you need to be more optimistic in your writing and this is also that that slide is related is relates natural language processing to rank because obviously rankbrain as I said based on the entities and I really like how it breaks down all the entities that it can find in the context I was really chat when I found myself there but I wasn't too happy when I actually discovered it classed me as a location so I'm just hoping that Google is Google sentiment and natural language processing analysis is better than IBM's or at least knows a little bit more about me but when we're talking about natural language processing we need to mention rule matching and this is something that Google has actually confirmed or introducer announced last year and neural matching if you think about it it's kind of like a turbocharged turbocharged synonym system right so in this case this is an example that Google gave all those sentences have a word change but through neural matching it Google actually can understand that the intent for that one word is different in the sentence the sentences intent is different for each one of those examples but the most flagship example that Google always shows is that one where when you query why does my TV look strange I would say maybe buy a new one but Google can actually connect that with the soap opera effect which I had no clue about so and basically even without mentioning those words in the query it will show you the relevant results now the difference between rankbrain and rule matching I would say is neural matching is turbocharging rankbrain so ranking Maps your query into the concepts and rheumatic helps rank brain to better understand the words in your query this is this is the example of neuron matching that I experienced so recently I had to jump on the call with with someone and we did it for Google mate and I actually wanted to record that call but there was no but I was like what the heck why is there no record button so I started searching and this is like basically what it showed me that's my search history as you can see I've gone through quite a bit of queries to actually find it and it didn't actually show me what I really wanted to see so I would say with all of these with rankbrain and neural matching this one really important thing that you need to remember Google works at the huge scale right so it doesn't really care it doesn't really always pay attention to which you're doing individually it actually pays a pension what people are doing right so if there was like 50,000 other queries similar to mine where people are struggling with something that would probably show more refined and better and more adjusted with results to that actual issue than you know just one-off guy who's who can't verify in the bottom right now I think it was last month when Google announced that they're actually using bird in the search in their search and bird is actually natural language processing pre-training mechanism to pre train the the NLP models and this is something that is actually open sourced is it is available in Google Cloud and Google call up platform set for tensorflow so if you want to play around with that then you can but overall the main job of this of this algorithm of this system is to better actually understand the query and understand it more as a human would so the best example that they give you is in stateless mechanisms like for example word to vac word Bank when you're saying bank card would mean the same what were to like would find in a statement we rank you Bank for example forward to that there will be no difference between Bank here and Bank here Bank heart and your rank we Bank right where is bad will actually find that that means something else right I wish I could tell you more about it but it's fairly new and I haven't tested a lot with it working currently on the on the model that I'm gonna use it on but this is to come now the next bit that I would want to show you is the famous google search quality evaluator guidelines document and obviously it's a pretty thick document I'm not gonna tell you to go around and do old and all the eat stuff right now but you should do actually but I want to focus on the actual is user intern that they show you so they they divided it into four main groups it's no websites and visiting person so if we look at the examples that they show you the no query is basically an informational research one and they divided into no and no simple for no query you're obviously searching for something that because you want to know about that thing for the no simple query Google will most likely show you the instant answer to whatever you're searching for now the two intent that they described you want to do something yourself but they also have a sub intent of that which is the device intent that you want your phone for example to do for year for example call someone will find a number of something etc etc so this is a device square live device an intent related query and just to stir it up a little bit they also showed queries that could potentially be taken as a device query my favorite one is phone number Empire State Building which if without all the understanding behind the words and the intent here Gugu might want to call the Empire State Building number directly but obviously since they understand that they would know it's not a device where it's not an actionable query for the device is actually actionable query for you because you're looking for that number right and the last in turn related one sorry the almost last interrelated one is a website so obviously anything that you want to do that she sided to see the website I like the last example when they say canon comm iOS digital camera obviously if you didn't mention Canada Canon that come they would take that as a do query that is transactional and show you all the oldest stuff and all the ads and and products and stuff like that whereas with this one they would just take you to to the Canon dot-com website and visit in person which is quite interesting obviously there are some simple examples here you know you search for Chinese restaurant or gas station obviously you really want to go to those places they also give examples based on at different locations sorry that's Walmart so they also give an example where this is a mixed intent so by searching for Walmart you might want to go to the shop or visit the website but a more interesting example is here visit in person so you're searching for turmeric which is obviously a spice but also if you're a sunny if you're in Sunnyvale California there's a restaurant called turmeric so they would also try to supply with that intent and satisfy that intent in case you wanted to just go to that restaurant where is if you're in in a different location they would just show you the the actual spice of the help or whatever turmeric is now looking at sentiment in Serbs I've got my favorite keyword by Instagram followers we've had a client for that keyword the ages ago and I'm actually tracking that keyword for the last two years or something like that and currently there's only these four results with strictly negative intent in the Serbs whereas the other results are mainly transactional because you actually might want to buy the Instagram lights but it wasn't always the sentiment wasn't always distributed like this so if we look at the example earlier this year there were four negative results just as now for negative results but the difference here is I think that stage Google was testing what which sentiment was more relevant because we was actually shuffling the sentiment in between the results and it was showing one positive one negative one positive one negative where is now it basically clustered them together so it thinks that you know people who would be looking certain with the informational research in turn would want to have those results together but if we look at the graph of the positions or of the sentiment of that keyword of what was really ranking there this is by the way H Rev so we can see that you know the sites were taking in but most of them were we're staying in the search results unfortunately a attracts couldn't give me the the data for like a year or so for that because it only shows five currently ranking sides but I actually kept the track of that throughout goes through out over 16 or 18 months and as you can see the set that the positive versus negative ratios we're changing over the time so yeah it basically shows you that Google is testing what sentiments what sentiment is more applicable what uses my wand and etc etc so two takeaways from that obviously currently you've only got six spots if you want to rank for that keyword with buying intent or positive sentiment right there's only four spots if you want to slam anyone who buys the Instagram land Instagram followers Google diversifies the search results this is a clear example because as you could see in the previous slide early 2018 they had like one or two resource with negative sentiment all and all the other ones were positive obviously that sentiment has changed over time and now the most important one relating to rank drain again Google will test the CTR and the user satisfaction around us around this diversification and again mmm adaptive algorithm right so don't forget about it now another keyword that I actually had a look at with that clearly shows the change in sentiment over time it's TSLA it's a symbol as the stock symbol of Tesla and that also shows that the sentiment around the brand can actually completely reshuffle the search results so I've got the results from September 2018 the syrup has actually changed slightly because back Vanda was there was also twitter and and top stories but you can see that I highly I highlighted here clearly negative sentiment it was around the time where everyone was shouting hey mask is a fraud and obviously that was reflected in the search for that you know strictly brand related query where is at the moment there are still top stories no tweets however all the all the things shown in the results are actually more positive or actually in the people also asked the they toned it down they no longer like oh is it the fraud is losing my beloved is it going up and are they actually losing money right so it brings me back a year ago where petcube IT said on during one of the workshops and then I also said something similar on the stage that it's not actually Google that ranks your site it's your competitors who rank your site so obviously changes in the sentiment is coming up in the search results what's being said about your business what's being said about your keywords around your keywords this is actually where you would you would see the change and you would see the results actually changing and and getting a different different results now let's go back into the intent classification so as I said in one of the previous slides Google divides it into know the website and visit but this classification is actually too broad right if you think about it I was mentioning transactional intent where is you know kind of like a sub part of do but that's that's basically not enough so this is a different classification model that has like nine different intents first one is the research intent that you can probably spot in the search results through featured snippets showing up people also ask questions and stuff like that now the second intent in this classification is the answer intent so again this is similar to know and know simple query and you would see that with the instant answer like insert calculators and stuff like that the third one oh sorry forgot that this is for me like a special kind of special kind of the answer intent because obviously it is also an instant answer that Google shows but this is related to a to an event or something like that so obviously it also satisfies your answer intent with the answer however it relates to something something else right third one is transactional intent which I mentioned a couple of times before so obviously you can recognize them by having a lot of ads in the Serbs and having made mostly Google shopping results on there as well now the local intent which is very easy one you've got the map and the visual intent and here obviously we have a big images part but I would like you to remember that query it says Jessica BL now right I'm gonna be getting back to that query in a second when I'm showing a completely different intent but I want to I want you to just remember it for a second okay so obviously this one has an image pack video intent where when we're searching for a video don't know if you guys know this guy is actually quite hilarious so you should check him out obviously all the results or most of the results will be showing the videos or there will be a video carousel now fresh news intent obviously indicated by the top stories and results that are mostly time-sensitive so you were to get a lot of these little nine minutes ago and stuff like that in the serbs brandon in branded intent when you're searching for a brand so obviously this this is classified by large site links usually there's only seven results on the first page and also only like the whole the brand owns the whole page one right and the mixed intent and this is where i would like to go back to the to the query that i that i mentioned earlier so the previous query was jessica BL now that query is jessica biel 2019 right for us it would mean pretty much the same thing right however for google that other query was strictly image related whereas this one is a mixed content because you've got top stories here that indicate fresh news you've got videos that indicate video intent obviously you've got you've still got the images up here you've got the knowledge graph but the information is actually completely random I don't know why this got into the knowledge graph you've got the related searches so her movies and TV shows and because I was squaring from Europe where is this law or you're allowed to be forgotten the right to be forgotten I think it's called it also indicated that it's a it's a name related query indicating a person and some results might be removed because of that right to be forgotten and another example of mixed query again this is English rugby team so we've had run Rugby World Championships recently obviously there are games up here with good top stories and the actual big brand related knowledge graph here as well now that classification was suggested but by content harmony but came Jemison this is just to summarize what Serb elements are actually and what sirs indicate a given intent which we'll use in just a second to actually through the kind of like automatic intent discovery but I would like you to have a look at two specific keywords that are really interesting to me and by the way this part is influenced by Rory Truesdale who had a great talk about user internet search lead so I'm giving the credit where the credit is due keyword co-working space London we've got none of the ten results that are strictly research that indicates research intent there's no transactional results because the last one that is out that isn't marked there is a directory of co-working spaces so you might think again that you're actually searching and researching if you're looking through the directory and we and basically if you have a transactional page there's no way you're going to rank in those search results right this is a specific example we we'll also revisit on the same pay on the same site in a different keyword which is we work so if we look at the page that Google ranked from we work domain we've got again very informational page with a map down the bottom there are the amenities and everything else that they offer in their co-working space now the second keyword is office space London that is for us again fairly similar keyword right but if we look at the search results that should actually say 10 out of 11 including maps so 10 out of your 11 results are actually transactional right so this is where you should aim to rank your transactional page not not in the not under the our keyword obviously even though it suggests you know a similar infant from the understanding of that keyword now looking at the we work here where they also rank if we look at that page Google actually selected a page that is 100% transactional we got the offers here okay we've got the content but this is probably an SEO SDS EOS job oops yeah there we go we back nice so obviously this indicates that Google not only will pick the the different size that it thinks will satisfy the income but it will also try to get through your website and select that page that satisfies that satisfies that intent in the search results right that's why the intern is really important so to recap all this every search every search has some intent behind it and Google will always try to predict and adjust the results based on that intent and for Google the intent is actually relevant right because if you think about it actually is right now optimizing for the dominant user intent gives you a better chance to rank obviously right and as I said before intent will intensive this faction will influence your CTR your engagement and eventually to your conversions and sometimes even rankings to go up right so this is your money now the intent discovery there's four methods that I want to show you this is a more practical part of this of this presentation so obviously the first method is through the Serb features that I actually showed you on the other slide you can basically look at the search snippets and based on which elements are coming up in the serve you can call almost automatically this thing that is you can always automatically and scale and in a scalable way decipher the intent of your of your keywords right now method number two this is one that I'm using quite frequently yeah shook it off should get it soon hopefully didn't find out that oh let's couldn't do it okay so by analyzing people also ask the questions you can actually decipher the intent behind that search query as well but it's actually a bit more tricky than just finding one action word like rent or offer or by etc because you have to actually understand that the question and never ever ever try to scrape faq schemas because you know that SEO czar doing that stuff now I'm not sure if you can see those questions but who thinks that the top one is research related or informational so anyone who thinks that okay one it's not if you read the questions they actually indicate more transactional intent because how much it costs how to how do you rent etc etcetera what's included that's more like indicating that you actually want to rent that thing right now you can do that also obviously through the VA through VA but you can also plug it into an natural language processing Meccan system there's quite useful walkthrough on search engine journal obviously those links will probably be shared with you or the presentation you can rewatch it this is something I'm playing around with haven't got many results yet but I know it's doable right and then it's it's really scalable method number three this is actually a really interesting one and I'm glad Rory truesdell actually shared that one with me so you can use Google search I'm sorry custom search engine to get to get the user intent so you for that you have to actually create your custom search engine 400 results add them to your Google to your custom search engine will see you go to set up and this is how you how you can scrape its easy and by the way if you or people who actually are more familiar with the code there's a little pink element here which is used by Google to actually measure the user satisfaction with the search results so it's basically pings itself and it knows what results you clicked and when you came back to the search results but once you once you scrape that you plug it into your custom search engine select the include just the specific pages that you have entered obviously including most probably your website then you need to ensure that she sorted by relevance and now the cherry I did that for keyword affiliate SEO course to see how affiliate lab is doing so in the custom search engine Matt's coming up number 10 can see people probably can't see the number here but it's coming up number 10 in the search results let's just get a beer no I'm kidding that's that's not true got you I got you right fine that's actually Mac not window domain so in the life's herb Matt it Matt's website is ranking or affiliates the affiliate lab website is ranking number 12 and now what does it tell us if your site if your page ranks higher in custom search engine and lowering life serve it means that you either almost probably have a technical issue or an authority issue right because the relevance is there that's why you're ranking higher in the custom search engine but there's something wrong with your website right or with your links right because the custom search engine takes less importance in terms in terms of links especially when you when you say it by when you studied by the relevance now if it's the opposite so your site ranks lower in custom search and higher in this in the live Serbs that means that most likely you're being pushed but by your link juice and you don't really deserve you know considering the the relevancy signals rankbrain and natural language processing all these you probably don't really don't really deserve to rank that high but the authority is pushing you higher and the best thing about the custom search engine for me is that it's actually got the API which you can use to pull all the results from it and the method number for you obviously can scrape the serbs through screaming frog to get page titles and meta descriptions and I'm sure none of you are doing it because it's a volley violation of Google Terms and Conditions and just remember that John Miller has a history of kicking off when these things happen so we're not allowed to do it and I'm sure no one no one is doing that but the scrapes herbs appearing on you just list them enter manually you go google.com slash search Q equals search query you can do however many you want obviously you'll face some issues but I'll talk about them in a second now then you have to go to customers extraction we've got an XPath code here you can use that for now I actually checked it yesterday is the works for the desktop search results but just remember that Google's HTML in surp is changing and it's different for desktop and mobile so whatever you're scraping you may want to adjust it once you have usurps there by the way you may get into some issues you may face some problems with that because you start getting captures as you do that as you do the scraping you you have to set your custom user agents to the chrome desk if you're obviously scraping the desktop version - a mobile version if you're scraping the mobile one and you need to switch on JavaScript crawler as well because Google is doing a meta refresh or some JavaScript refreshes so you have to set that on - but then once you have that you actually have all the meta descriptions then you can plug them into a spreadsheet ask your VA to classify the intent find the dominant one and then you can obviously optimize for that or you can use machine learning for that as well but obviously that's a bit that's a bit more difficult now why is that even important so through scraping through analyzing all this you can actually analyze the SERP language well it's not just the keywords it's not just the queries but a sexual actual language that you get to understand and obviously you have to do that having the predicted and dominant user intent in mind while you're scraping page titles and meta description so the descriptions in the Serbs you can also use that to better optimize your meta tags all right which is really simple but you just need to think what the users want and what your competitors are doing as well alright and finally having all that serve information also allows you to create silos or topical classes because it's not just about as I showed you in the patent relating to related searches it's not just about you know one query it's about all the queries so you can build the whole side structure around that as well say you have a transactional page and you want to support it with some informational content so through that through understanding the SERP language you can actually build that much easier and finally the semantic relevance intent and desire so this is actually you know coming up and meeting your potential visitors in turn and what they really want and what they desire from from beside this is a short little results page results case study not really a case study but I wanted to show results we got that's actually as you can see 28 days by on 28 days and then again we focused the optimization of the user intent however it's really hard to optimize just for the user intent right because what you're gonna change all the words find into by not really it doesn't work like that you actually have to reword the content you actually need to implement some other changes but overall focusing on the intent and having the all these optimizations that we did based on the intent research we actually managed to achieve that month four month plus nine percent of clicks that was a fairly small site by the way plus 32 percent in the number of impressions so obviously that would probably indicate that these results are gonna kick in more and more because booboo is so adjusting the changes that we've made because that's that's a very fairly fresh case here as well obviously average position slightly app City are slightly up now the latent in turnin that's the last part of the presentation so around Christmas like Friday the valley or Diwali matures or if I mess it up any other festive seasons Google will try to change the intent that they have that they are showing in the search results results and they will twist tweak it into more transactional and more positive with with more positive sentiments well this is something that people on the internet gr Oaks have shared he found that he mentioned that there is there is a clear clear trajectory that Google is actually adjusting the the intent behind the serbs depending on the season depending on the time of the year I've observed that as well so this is a slide that I took last year black game console November 2018 that was just before Black Friday obviously the Serbs were mainly transactional only to research related or informational results now this is the same keyword taking screenshot taken a few days ago you know you can you may argue you know they've been a lot of cool changes along the way you know bird neural matching all these things but overall if he only if we only take into consideration into consideration the this the intent obviously we've got the top stories here we've got more informational results some videos here so for me that clearly indicates that the intent is mix here so I'll be really curious to observe that keyword for the next couple of weeks or three weeks as we're getting closer and closer towards Black Friday if that intent is going to change again as we get closer but now this is a really tricky part with latent intent it's really difficult and it's very niche specific right because they can't apply it for all of the keywords you can't really you can really predict it that much now guessing the future interest or desire of your users is possible based on trends based on the data this is by the way a trend over a keyword that we've been observing for a couple of months now we got the client to it's it's a it's a t-shirt printing business we got the client to start printing it in early October and it's actually picking out that one keyword we'll see how it goes I'm not gonna say I predicted that keyword predicted the latent intent or anything like that maybe slightly predicted the interest but overall I'm not we haven't managed to create any content for that we're just trying to help the client to monetize on the interests on that as well so you might argue again it's not really a latent intern it's more of a what trending right now but yeah we'll see we'll see how it goes but the bottom line is it's still guessing right so the latent intent is still guessing and even though there are strong case studies around latent intent and this is shout-out rain there anomic s-- who's got a really strong case study around that I think it's still more of a proof of concept actually them you know something scalable something you can do for your big website for example or even in your industry because as I said not all industries can use this latent intent and this is all folks it's time this time thank you thank you know Easter eggs now I'm gonna finish things off with a couple questions we're a little bit short on time so we're gonna take the top two most voted questions starting with Marty's question can you recommend three or so actionable items we can do for improving rankings Thanks a toast Adam probably is a great one ion I suppose you mean like the browser extensions actually so Surfer has released one recently it will be a really good one use a choice one no product placement here but they're just really good products okay next question how does Google know if a user is satisfied with the results oh that's a really great question if you go through my slides again there's like a like a thing that I pulled from buckling goes website and then showed you from my last year's presentation it basically and then I think I've referred to that with that little pink there so it basically tracks what results you click and then because obviously it has your cookie so it can identify you it knows if you're going back to the same search results or not if you're buying if you're going there and back so if you're pogo-sticking as some say as some call it and also because it as I said it looks at the patterns in your searches it will try to identify whether or not you're actually doing a research so you're going and checking a lot of websites and going back and forth or you're actually a user that isn't satisfied with the search results because then that pattern would look completely different right so it has mechanisms in desert in the Serbs that allow it to see what you're doing that basically I see what you did there all right how about one more question this one just got voted number one what is the best way to figure out if your website has the right search intent for the query okay that's a great one so obviously you should as I said my my best thing best bet would be to look at that query and look at the other sites analyze the sentiment through the tools see what other people are doing because as I said it's not Google that ranks your site it's your competitors that - right and I wouldn't actually just limit myself to just one query I would actually have a look at more queries that have all kind of like suggests the same intent in the SERPs and then you know just see what your competitors are doing good all right how about another hand for rad thank you [Applause] all right thanks everyone for tuning in on the YouTube livestream we're all gonna head over to Neiman social at this point it's I'd say it's a little bit much for a walk but you could walk if you want a little bit of exercise otherwise grab is going on you might find it took took on the side but it's a block over it's on Neiman we've rented out the whole place we're just gonna have a networking event and yeah chill enjoy the conference take care guys [Applause] [Music] [Music] [Music] [Music]