Skip to main content

Semantic core

A lot of professional literature has been written about how to correctly assemble a semantic core. Our task is to tell people who are not immersed in the topic how and why the semantic core is being collected for the site. You will find step-by-step instructions on how to make the semantic core of the site, with examples, listing programs/services that will help with this.

The SEO specialist should study the algorithms and rules of search engines as much as possible in order to get good positions on the promoted queries. The promotion of a web resource is influenced by many factors. But if the collection of semantics for the site is carried out incorrectly, the framework and architecture of the resource will be assembled incorrectly, which is why further promotion will be ineffective.

The creation of a semantic core is the foundation on which all the rest of the SEO promotion of the site is built.

The semantic core(s) are grouped queries that characterize the type of activity, specifics and everything related to your site. In simple words, this is a selection of queries that users write to the search box of the PS in order to get an answer to their query.

During the collection of phrases, a specialist can immediately find out:

  • What exactly are people looking for
  • How information is requested
  • Why do they need it – to buy, find out, order, get a price
  • Where can I use XIA
  • Where can the semantic core be used

The core is used in SEO-promotion of websites, setting up contextual advertising, and drawing up a content plan. With proper use of semantics, you can achieve your marketing goals. But you need to start from how a person requests information and whether your site (current, future) meets this request.

Let’s say you have an office repair company, respectively, you are not interested in providing mobile phone repair services. Also with the collection of requests, your target request will be “office repair”, not “smartphone repair”.

In addition to organic search, the semantics of the site solves 2 more tasks:

  • Competent collection of semantics will help you predict when and how much traffic you will receive from the search engine. This will help to calculate the potential income of the business
  • The second task is not about technology, but about marketing. Before launching the site, the entrepreneur and the hired specialist need to decide what information we broadcast. A good SEO specialist thoroughly studies the target audience (CA) or works in conjunction with a businessman. To compile the semantic core of the site, you need to know what gender, age people will come to you, how they think and their price segment.

A business should decide what to tell customers, independently forming a trend, and not just create pages to organize website optimization. There can be many tactical options for solving the two tasks. But there are only 2 strategies. The rest is derived from them.

What is important to know when collecting the semantic core

  • The better you know the project, the business and the target audience, the more competently you can assemble it.
  • The core should be focused on the keys – words or phrases that your potential customers use. SEO specialists divide keywords by various parameters, but the basic indicators that even a novice specialist should pay attention to.


  • Frequency of request
  • Request type
  • Binding a request to a location
  • Below we will tell you more about each type of request.

When collecting key queries, it is necessary to take into account the main classifications

How to choose the right keywords

There are 2 cases of collecting semantics: collecting semantics for the current site and collecting semantics for a future (new) site. In the first case, we analyze what is there, add the missing, expand the semantics. In the second case, we do all the work from the very beginning, we assemble the maximum volume core.

But in both cases, the keys are distributed into groups during analysis and cleaning (clustering). Groups (clusters) of queries indicate the number of required pages on the site because each group must be “linked” to a specific page.

Queries that may relate to your topic and respond to a number of factors that are beneficial to you, not all may be useful.

Depending on whether your site has been developed or will only be developed after the collection, the method of collecting keys and the number of necessary tools will be determined

The amateurish (classical) way

The simplest method of forming a core is to pull high–frequency queries from WordStat that meet general business criteria. Let’s say you have an office repair business in Moscow. In the WordStat service, the query “office repair” will have a good frequency and any specialist will definitely add it, but there will also be less frequent queries derived from the main query: “cosmetic repair of offices”, “turnkey office repair” or “urgent office repair”.

An amateur way of collecting semantics for SEO

If you collect only fat keys hoping that the main request will gain more traffic, this will lead to the fact that you will not receive customers who request a service with clarification or write more complex requests.

In the case of the amateurish method, such a method is also used when a person takes key phrases “out of his head”, we invent them ourselves based on how it seems to him that other people are requesting. Simply put, he does not use tools at all, writes keys based on his representation of search queries. 

It is clear that, having no suitable programs and services, requests are not subsequently clustered, analysis is not carried out. And if programs are used, for example, the same WordStat, then the final semantics is usually limited to collecting key queries for the most popular query and only for it.

In most cases, with this collection, requests of incompatible types will be contained in the same cluster, and this is a big mistake.

The above is not the best way to form a semantic core, how to compose it correctly?

Professional (marketing) method

This method takes into account the anatomy of search queries. Each, as a rule, consists of:

  • The body (in the example “office renovation”)
  • The Specifier (defines the user’s intention – “order”)
  • Tail (details and explains the user’s interest – “turnkey”)
  • Professional way of collecting semantic core
  • Analysis of the request “Order turnkey office repairs”
  • Analysis of “tailed” phrases is a trademark feature of a good SEO specialist, which allows to reduce the competitiveness of the core and increase coverage

In fact, the query types of the semantic core are a filter through which we pass the available keys.

Keyword classifications

The varieties of key phrases are divided into 3 main types, below we will consider each of them.

Common. These include requests for which it is impossible to determine what the user’s intention is. “Repair” – how to carry out, what stages, what will be needed, how much does it cost?

Transactional. Commercial direction, they can be accurately identified by the words, “order”, “find a company in Moscow”, “leave a request”

Informational. The user is looking for the necessary information: “what are the stages of repair in the office”

Navigation. Requests related to the search for an address, route, point of sale

Multimedia. In addition to the text, there is additional content – a video sequence, music. This can be an informational request or a commercial one

Keywords by frequency (search queries per month)

High-frequency queries include phrases that have at least 1,000 requests per month

Average frequency phrases are from 100 to 1,000 requests per month

Low-frequency contain up to 100 requests per month

Varieties depending on the user’s location

Geo-dependent. For such queries, the search engine output depends on the user’s location at the time of the search. 

Non-volatile. The output of search engines for such queries in different cities is often not particularly different. An example of such a query is ”how to make pepperoni pizza”

The task of an SEO specialist is to work with general and multimedia in order to sift through the unnecessary and select the necessary. With transactional, informational and navigational, we understand the frequency and profile of the site.

Instructions for collecting the semantic core in 5 stages

To create a competent semantic core, it is not enough just to collect a lot of queries, you need to perform a number of works.

Stage 1 – Competitor Analysis

The analysis of the competitors’ top will help to see the amount of semantics desired to achieve such a result. You can also understand how the structure of the site will look like.

Stage 2 – Collecting markers or basic keywords

We can’t hammer everything that comes to mind into the parsing tools. We need to start developing a semantic map first. We have the website of a large construction and repair team. Probably, we will divide the services into two large groups: construction and repair. In the “repair” we will divide the services for the repair of apartments, country houses, garages, industrial and retail areas, offices. 

Let’s take the office “branch”. This is a cosmetic repair major, designer. Accordingly, in the capital, these are services: turnkey office repairs, project development for office repairs, purchase of materials for office repairs. This is the base we will work with. It will be necessary to “throw” primary words and phrases across all “branches”.

List of phrases that should be in the semantic map:

  • General: “pet store“
  • Transactional (commercial): “purchase pet supplies”
  • Information: “what pet food and supplies we offer”, “pet goods selection”
  • Navigation (geographical): “pet stores in Estonia”, “pet stores in Tallinn”
  • Multimedia: “photos of pet toys and accessories”
  • You can also combine (intersect phrases),using various services.

Stage 3 – Parsing and expanding key queries

When creating a semantic core, it is worth considering promising keys. You currently sell pizza, but you don’t sell sushi and rolls. And people often look for them together. For the future, when the business will expand, it will come in handy. To expand the resulting pool, we take tools for selecting keywords and phrases. We set the settings, analyze. What clearly does not fit – we immediately filter out. You can use the following tools:

Google Ads Scheduler

Google and YouTube search suggestions – you will see them in the corresponding fields when you type as a continuation of your phrase

Collecting associations (together with the search query) under google search results

In addition to tools from search engines, we also use programs:

Google Search Console – ”Report” – ”Efficiency” – ”Queries”

There are a huge number of additional services for query parsing, here are a few of them:

  • Moab
  • Megaindex
  • Key collector
  • A-parser
  • Rush analytics
  • SE Ranking

Most of the keys will be repeated, but there will also be those that did not exist before. Everything collected will require combining and further cleaning.

Stage 4 – Cleaning and normalization of phrases

After evaluating what people are looking for, and collecting hundreds or even thousands of keys, we start cleaning. If we understand that they are not looking for what we need, we need to get rid of such requests. It’s long, tedious, but you need to do it manually. There is no faith in automation here.

Cleaning and normalization of phrases for semantics

Cleaning of unnecessary requests. The basket will include requests mentioning direct competitors, with those regions and cities where we do not work

Removing duplicates. There will be synonymous queries in the basket. If there is a request to “order repairs”, then we no longer need to “order repairs”. We also do not need those queries that are a refinement of the main key.

Spell check (without typos). Queries that are typed for some reason with errors are demolished. We delete queries whose meaning we do not understand

Removing unnecessary characters and queries with a frequency of 0. We safely remove microqueries with a frequency below 10

To determine the exact frequency of requests, you can use one of the programs:

  • Key collector
  • A-parser
  • Spyserp
  • Rush analytics

We upload all this to a file and save it.

Stage 5 – Clustering of the semantic core

The result of the work should be a competent distribution of key queries by type and groups.

Determining the type of request and matching the page

For example, commercial, informational and page types: home and internal – we send information to the “Articles” or “Blog” section, and leave commercial requests on the “Services” or “Goods” pages.

We determine the compatibility of query promotion on the same page

“How much does the repair cost”, “The cost of repair services” will go together perfectly on the price list page. But “dismantling works” and “roof installation” will not “fall” on the page dedicated to the office repair service.

As a result, you will get a branched “tree” with all the collected, cleaned keys distributed throughout the site map. For post-processing of clustered semantics, you can use a free add-in for Excel. It will be convenient, clear and understandable.

Semantic core analysis or how to determine quality

A good option is to order an express semantics audit from KeyClient. Have you learned how to make the correct collection of the semantic core, have collected it, but are not sure of its quality? Get a semantics audit from Grentana.


What to do with the semantic core? Semantics is the foundation, the “house” of the site is being built on it. Having a core, we make a content plan. After the analysis, there is already an understanding of how many phrases, how many pages and which ones need to be made on the site. The main needs of Central Asia have been identified. We decide which phrase is taken as the central one. This is far from being the most popular key in 100% of cases. The main thing is that it fully reflects the needs of the resource audience. We distribute other keys:

  • Information material, a page where a product description is given
  • Question and answer
  • A special blog where the nuances of using products are considered
  • Product or service pages
  • Order page

Don’t run into information needs. The keys are important. Are they inscribed in the text? Great. But you are not writing for a search engine, but for people. The information that is given should be useful and easy to understand. If this is not done, there will never be success and a TOP. What not to do with the semantic core:

Ignore the “super competitive” keys. In order to acquire traffic and sell a product, you don’t have to be in the top. 

The reverse situation – the rejection of low–frequency phrases – is also undesirable. We recommend leaving as many ultra–low frequencies as possible – later they can be deployed to the content idea

Focus only on formulas and coefficients when compiling the semantics of the site. They don’t treat tests, they treat patients. KEI is important, but we don’t have a math school, don’t pray for it

Don’t create pages solely for the sake of queries. There is no need to make two pages “buy pies” and “order pies”. It’s the same thing here. 

Do not try to shift everything to automation. The programs and services are not bad, it is a good working tool. But by themselves, without manual analysis, they are worthless. The algorithm knows how to systematize, but it does not know how to think – it needs a person for this

Focus on collecting key phrases. Don’t turn into a private eye, who endlessly monitors competitors, collects keys day and night. This is redundant. At the start, Google statistics are enough. It is worth expanding the number of sources if the core requires expansion, or the team has a shortage of ideas.

It is not worth building a semantic core for the sake of building a semantic core. The procedure is carried out for a purpose. What is yours?