Introduction to which will be...
That the gathering of the semantic core is just the beginning and that the most difficult is to the next article on clustering and determining the user's intent (intention). This initial stage can also be described as the most free and creative throughout the work of the SEO. After all, further, everything will be done strictly based on the gathered semantic core.
Further in the article, I can and will use the abbreviation of the K. for the keyword.
I foresee some objections that they don't need a semantic core. It is difficult, uninteresting and long to do. Here I agree, this is not the most interesting stage in creating a site. And most likely, the guys in my field, that is, the developers and designers do not consider it necessary.
I did a little poll on Reddit, here and here to find out who and how uses the semantic core. Here are a couple of quotes:
Of course, I'm not talking about everyone. Professional teams, and especially professional teams in highly competitive areas, use semantic cores and use them to the fullest. But this article is not about them and not for them. so, according to a study of December 1, 2023, conducted by the Ahref team, 96.55% of all content, do not receive traffic from Google at all.
I will not retell the study verbatim, but here are the main reasons for such ... spread:
- The topic for which the web page is written is not in demand among users.
- Nobody links to pages
- The keyword on which the page is promoted does not match the user intent.
Given today's realities, I would add a 4th reason from myself - the appearance of AI. I wonder how many pages get traffic now?
Wait, wait, but what does the semantic core have to do with it? Not a word about him in the study. Indeed. But the 1st and 3rd reason are a direct consequence of the absence of a semantic core - the strategy for the development of the site as such.
In this article, I would like to convince you, my reader, that it is not so difficult to assemble a core, and most importantly, a useful activity even in the era of AI. This is only the first part of the series, about gathering the semantic core, it is directly about the collection of keywords and phrases that one way or another relate to the theme of your site. The second part will be about clustering and definition of user intent. And without further ranting, let's continue.
Bulding the semantic core
The collection of the semantic core can be divided into three stages:
- In the first, most complex, it is necessary to firmly and clearly define the niche of your site. In the future, it will be possible to expand the sphere of its influence, within its niche, of course.
- In the second, we will use special web resources to collect keywords. It is important for us to work with the original source, that is, if you optimize the kernel for Google, then you need to work with the Keyword Planner, and if you plan to optimize your site for other search engines, then for Yandex this is Yandex Webmaster, and for Bing it is Bing Webmaster Tools.
- And in the third stage, you will need to clean your core of any debris. Garbage refers to navigation, brand and any information or commercial keywords that do not belong to the chosen topic and niche of the site. More about keywords types in "Clean out the semantic core".
Further on each stage in more detail.
Determine the niche
You need to decide on a niche. How to do this and how to find such a niche so that there is a lot of traffic and low competition? This is actually the Holy Grail for any webmaster, having received which, any doors and opportunities on the Internet will open in front of you. Alas, no one has yet found it, including me.
But there are general recommendations, at least for those who just want to start. And, in fact, that's how I started making this site. The recommendations are as follows:
- Decide what you like to do.
- Choose from this what you can do best and you are an expert in this field.
- It is desirable at the same time that this niche is also profitable. One enthusiasm is not enough for a long time. Although this is debatable, I have been doing this site for 2 years and 3 months, but this is probably just me.
There is a more difficult option where you choose from already known niches on special sites. So I can note the following sites:
- Similarweb- Here you can look at the full breakdown of sites into categories, it is quite convenient to determine at least the search direction.
- Liveinternet- Similar to SimilarWeb, can also help with the choice of the direction of the site, but for the Russian-speaking segment.
- Listofniches- a large list of niches, with a superficial opportunity to study a niche.
- Nicheslist- Just a large list of niches. See if you can find something for yourself
Let's say you decided on a niche, went at the call of the heart, or chose a niche from the lists presented above, it doesn't matter. Let's say you decided to make a site about telegram bots. You understand this, you have already made two bots of varying complexity and functionality for hundreds, and are ready to create a separate site for this. What's next?
Next, you need to "assume" basic keywords for your site. Or just googling and see what keywords are used on the main pages of your competitors. So, for a site about the creation of telegram bots, we can assume that the main phrases will be:
- telegram bot- The first thing that comes to mind
- aiogram- the main library where you write your bots
- tg bots- Common reduction in competitors
- telegram Bot- if you plan to write to an English-speaking audience
Gather the keywords
Having decided on the basic keywords, you need to find the rest, not so obvious. For this, special services from search engines are used:
- For Google - Keyword Planner.
- For Yandex - Wordstat, or "Selection of queries and market analysis" in Yandex Webmaster.
- for Bing - Keyword Research on the corresponding tab.
They are free and indispensable at work. If only because they are the primary source of incoming traffic.
On the semantics keyword collection pages, you will need to enter the basic keywords and unload the received lists to the tables.

Google(Keyword Planner) 1) to download 2) to enter the basic keywords 3) Keyword Planner allows you to filter all brand requests at this stage, do not ignore this possibility and uncheck from here

yandex(wordstat) I'm not sure, but perhaps in the future this tool will cease to exist due to the appearance of an analogue in Yandex.Webmaster

Yandex (selection of requests and market analysis)
In general, I recommend using the selection of queries instead of Wordstat. He is much more informative and useful than his successor.

You are not required to use all three search engines, only the one that brings or can bring traffic. It's just that in my case, all three search engines bring me traffic, so I try to please all three.
After downloading the tables, you can immediately go to cleaning the semantic core. And you can collect additional semantics of search tips. Usually I do it manually and selectively, because not everything that is there is useful.
And the "People also ask" section can be easily assembled using the extension for Firefox, or for Chrome. This is how it's done:

Although there is such a nuance that this extension works only with Google.
Clean out the semantic core
After we gathered a bunch of keywords for our core time to clean it up. What does it mean to clean? This means removing keywords that are not relevant to our site. as well as certain types of queries (keywords). Speaking of types of queries. Traditionally, there are 4 types:
- Informational - Such requests imply that the user is looking for certain information. For example: 'How to make a bot', 'why is the bot not working'.
- Commercial - The user intends to spend money. For example: 'tg bots not expensive', 'Buy bot'.
- Navigational - The user wants to get to a specific page/website. For example: https://google.com or 'Telegram Org Bot'
- Branded- The user wants to find a specific site, but does not know its address. e.g.'missrose_bot' or 'Omad Shou Bot Telegram'
Immediately, we delete navigation queries, they are of no use to us. We don’t need branded ones, although if it’s your brand in requests, in my case it would be 'Tim the Webmaster' or 'TimTheWebMaster', then we leave, because these people are looking specifically for you. We definitely leave these.
With commercial ones, it is more difficult and depends on the chosen niche and theme. So if you are going to sell tg bots, leave it. If not, then there is no need for such requests.
We remove information queries depending on the relevance and the implied intent. What do I mean by this? For example, here is the query: 'telegram apo' people entering this query are looking for a specific documentation page. It makes no sense for us to rewrite the telegram documentation, because we are not their creators, so we delete this query from semantic core.
If there is doubt about the intent, then the easiest way to find out for sure is to google it for sure. In short, check the SERP page for a specific query. If information pages predominate there, then this is informational keyword.
Conclusion
As you could notice from the article, the first stage, about the choosing the right niche for the site. It is the most difficult. The second, about collecting the core itself, is the easiest. And the third, cleaning the semantic core, can also be easy if you arm yourself with the right tools.
Alas, there is no single solution to collect 'search tips', 'frequently asked questions', and 'People are also looking for'(Block below). At least not on all browsers and not for all search engines. I promise to fix this and develop such an extension for browsers like Firefox and Chrome. Well, for now, we have what we have.
Although we have collected all the necessary keywords for the semantic core, that's not all. In next article we have to cluster this kernel, that is, to group keywords by intent, because in fact, many keywords lead to the same results in search results and although the phrases can be very different from each other, the intent will be the same.