Improve Your SEO with ChatGPT: Create Topical Clusters and Visualize them with single Prompt

hello everyone today we will talk about a check gbt and how to use its code interpreter for the search engine optimization before I share this specific tweet and this specific Post in our holistic SEO community group the public one and in this section many people ask us how we actually get this specific topical cluster so I personally don't create topical clusters from the queries that much but from time to time if I am creating a new project I am trying to just understand what is the overall structure of the competitors and since I am actually

doing research for the five years and Democratic democratizing topical Authority concept for the last three years and since we launched the the biggest course in the industry and redirect actually became nearly the 1200 people in our course program and shut down the gates I can tell that after a point you get some certain advantages and you don't actually need to know coding to be successful in semantics and you don't need to create topical clusters that much even sometimes you don't need queries but if you're a beginner you will need these things because they will be

helping you to understand the overall structure in a visualized way and today lucky that actually chat gbt and the code interpreter is really helpful to do these type of simple stuff before starting I must tell that if you see this type of a blue hat in our social media posts or these videos in the left bottom kernel corner you should be seeing it means that this is very simple video and many people follow us over three years and I'm very happy for that but I can tell that I might be having the the Nerdist SEO

Community ever and I am happy for that because my community loves reading doing research and they love doing the hard work and this is the separator between unfortunately someone who will be losing in the age of llms and agis and someone who will be winning as well as we have said in the Saigon or the Birmingham speeches that we have given and in our private Community which is consists of consists of the course owners here there too actually we help each other to learn these things better too so before also starting as well I might

stop I must tell you that you should be learning the coding called interpreter is for interpreting the code not for writing the code it will be really hard for you if you try to write the code with jgbt it will always give a different result and you will be asking many things you can end up doing let's say half hour of work in three hours even because it doesn't have to understand also it doesn't allow you to download any Library there so that's why my suggestion to is that open a folder in your desktop give

a name like code or python then open another subfolder inside that and give that give a name like topical visualization and there whatever you see in this specific jgbt screen try to give that these calls into the F code file in inside that once you are able to do that you will be ending up maybe 200 different secrets for 200 different let's say purposes and then you can use code interpreter while learning the coding as well this was How I Learned actually so let's start examining our prompt here and I can tell it in this

specific sample I used a simple sentence for as the prompt but since this is a video I want to make it a little more developed and usually in my prompts if I want to just use a single prompt I mean you I can actually use chain of the prompts but usually I try to spend less time for these things and this is something that I can actually register or give one of my teammates or you can also register this prompt and use it as you wish I mean record so here I directly tell that there

are four section sections in this prompt and I give every prompt a section number every prompt has these quotation marks as the border so this is the section number this is the let's say purpose of the section and this is the content of the section we directly to tell that our purpose is creating a topical cluster and visualizing it and Please assign every query to a specific type of let's say topic according to its relevance and here we start to actually imply that there will be a data set in the form of date let's say

data frame then we tell that make the topical cluster actual bigger if the number of the queries in the cluster is higher than the others the second section here it directly defines our data set we tell that only focus on the keywords volume and the keyword intent the purpose here is actually preventing code interpreter to look at all the trends CPC or other types of uh let's say the columns because if you decrease the dimensionality it will be working faster too I have given this I did not use it but it will be your homework

I will explain it to in this part we directly use these specific terms so if you want to create topical cluster that are actually three sorry two methods for that one is supervised learning the second one is unsupervised learning so using actual unsupervised learning is better for creating topical clusters because it will be creating way much bigger clusters and the reason for that s is that since there is no cluster name in this case they the sir the machine actually creates all these clusters and they give just numbers so the cluster names but it is

not that much useful for you because you need a representative word or a name or a label for it so that's why here we will be using this supervised learning for topical cluster creation which means we are giving topical clusters a name directly before creating them and then we tell the machine that according to the name of the topical cluster Please assign every every keyword in the keyword columns to a specific type of cluster and in this case we will be having higher precision and more presentable clusters but the problem is that some of the

queries won't be seen relevant to the any of these and you will be ending up creating more cluster names to be able to have overall structure and it might force you to understand query semantics and do a really good amount of extensive research in this specific industry the industry that we have chosen here is Beauty and the makeup and this is the Sephora so this section is actually important also for giving a few other type of search engine optimization tips please do not use this type of Mega menus as much as possible if you link

everywhere from everywhere it means that you linked nowhere and at the end of the day this is a really large actually menu and it will be making every page especially in the mobile a little more let's say slower and it will be harder to actually reflect your main topic on this specific page and the search engine might start to ignore these boilerplate since it is too large and it might let's say the search engine might try to configure the importance of the main content further but since this website is very old and already alternative it

means that it is worth to run and Trigger some certain type of improved and costlier ranking and understanding algorithms and you might not be that much lucky especially at the beginning and with that side the second specific tip from me is if you want to understand a website or if you want to create a topical map I even don't use as I say the queries anymore I do it in a reflexive way since I am the structure or let's say the founder of the specific concept and the methodology in this case I don't use always

these things they just happen in my Tulsa stream but if you have hard time to create a topical map please use the menu items you can create a topical map based on these specific sections and in the case study that I am writing in this area I will be showing some examples for doing it and we will be sharing it probably in one month it is also one of the reasons that I am I started to publish these blue hat tricks or let's say SEO suggestions with the blue hat because writing my SEO case studies

usually I put juventy websites 10 websites into a single SEO case study and it has it takes over one year to write it and also Gathering the data making the extensive amount of research in the patterns research papers and again getting the first hand data comparing it to the ranking algorithms defining the concepts visualization of the concepts and democratization of all these methods is not an easy thing I am doing it over three years now and still it's hard I have written over half million words and it takes really a good amount of time that's

why to break the silence in let's say our audience and also help every level of people I just wanted to do this in this Vibe because I must tell that I have the nerdiest community in the SEO industry and they love it but still it will be helpful to give these simple things from time to time so with that site you should be checking these Concepts to be able to create topical clusters I just scrape these terms in this area I did not take all of them but if you want to create a let's say

more comprehensive topical cluster you can do that you are free to do that and I have taken the queries from scmrush these are the USA keywords or the queries that this sf4 actually is ranking for I have taken only these ones then this is the last section of our prompt we tell that do not use bat and body as a single term because this is a representative for other multi-word phrases and the concepts that you see in this area if the chair gbt code interpreter uses the regex you will need to actually tell directly that

these are not single verse be careful for your regex and when it comes to here create a line and a bar plot here we tell in this end example we just actually have the bar plot but here I just wanted to improve it further and I told that user bar and I use a line plot if you use two different visualization in the same graphic you will need a multi-wire access and you can assign one of the y-axis's numbers to let's say um to the query count and another one might be the search volume if

you look at the search console you will see actually similar stuff there too because every search console graphic is actually a multi-wire access depends on how many metrics you are choosing and here I tell barbalot is for topical cluster size based on query account which means keyword count the line plot is for the total search volume inside a topical clustered queries or the keywords and then I tell write the total query count to the top of the bars in the bar plot because it will be helping you to actually see the exact number inside the

topical cluster and remove the topical cluster from visualization if the total query account is lower than 20. so you can remove this section if you want but it is important to know note that if you don't remove that the plot will be too complicated to use it or making it readable and because of that I have put it here because I don't want to include everything and it is also because of restrictions in the code interpreter if you use your local machine you can create bigger plots then you can remove this part As You Wish

itel use warm colors in the visualization because it's my preference and when we come here as you see they directly actually skip reading other columns it directly starts with this one it says there are variate of the columns rated to search analytics however as specified in the prompt we will only focus on this thanks then these are the three steps clean the data set basically and then perform keyword clustering generate the visualization if you would not create this type of prompt the steps will be more complicated sometimes it can even try to use word to

vac and then they can it can tell that computation cost is too high let's use fast text than if they use the facts fast tags in a successful way then they will try to download a few other things then it won't be downloading or they might not be running the script as you want so that's why I try to be more specific as much as possible and this is how they are being clustered and this is how they are being actually let's say shaped because basically here we are creating a certain type of dictionary to

be able to visual visualize it then this is the visualization very simple so I did not state in my prompt whether I want actually a vertical or horizontal bar plot and here actually it created directly a horizontal one and in this case you can you can if you want you can change it and I see that actually this is not a warm color but it might be relative it might inform for a machine so red what is warm for a human I guess this is blue I am a little bit colorblind a warm for the

machine or maybe I should be sectioning my prompt further or I can actually run a second part as well too here also you should read these sections as well because here as you can see actually it uses the warm color for line plot so I can't blame it I guess he not he or she looking at the plot we can see that higher cluster has the highest keyword count while the foundation cluster has the highest total search volume this section is important for you to actually focus on your content Network creation process a little bit

further and sometimes it can also tell you that some certain type of queries might be belonging to the multiple other clusters for example here if you look at carefully most of the I related things are not reflected in the our visualization it is happening because eyeliner eyebrow bro eyelashes eyeshadow eyelash serums eyebrows serums eye primer so let's say we have a query with the eyebrow and it doesn't it mentions the serum but also it mentions the cream let's say or it's mentions eyelash and also the let's say eyebrow my main point here is that if

you have two similar cluster names some of the queries won't B clustered properly and the cluster size will be lowered and since we tell that remove every cluster which is smaller than 20 you don't see that much I related things in this area but we have lip oil lip liner for example we have eyebrow actually here and eyeshadow is here too but still actually if you check these specific sections one by one it will be easier for you to understand the similarity issues or distance issues there so that's it we will be adding very much

more simple tricks like that I hope you will be enjoying it one more time I am telling that if you are following me over three years it means that this video is really really simple for you but there is always something that I am proud of my community and my follower base it consists of from beginners to the 15 years of experts I even have followers and networks with the seos who has seen Alta Vista and they still read my case studies and they learn new things and at the same time beginners of course they

learn new things too so imagine that beginners and very Advanced seos and Engineers they also learn things together in the same channel it makes me happy and if it is too simple for you you can watch the case studies or you can wait the one that I am writing and that's it we will stop being silent and we'll be adding way much more things like that too thanks for the listening love you all [Music]