How to collect a semantic core for the site: a detailed guide

If you have decided to develop a website on your own, this article is what you need, because search engine promotion will need to start exactly with the construction of the semantic core.

5. How to choose keywords for the semantic core
5.1 Determining basic keywords
5.2 Expand the semantic core
5.3 Removing inappropriate search phrases
6. How to group keywords and build a relevance map
7. What to do with the semantic core
8. What not to do with the semantic core, or how not to turn into an old-school SEO
9. Semantic kernel: marketing or SEO

The most effective source of attracting traffic is natural search. You need to make the site interesting and visible to Yandex and Google users. It is enough to determine what your audience is interested in and how they search for information. This problem is solved by building a semantic kernel.

Now let’s understand what a semantic core is.

Semantic kernel – a set of words and phrases that reflect the subject and structure of the site. Semantics is a branch of linguistics that studies the semantic content of language units. Therefore, the terms “semantic core” and “semantic core” are identical.

What information can be found on your site, and most importantly, with what search queries users are looking for information to be published on the site – these questions need to be taken seriously when compiling the semantic core, because one of the main principles of business and marketing is considered to be customer centric.

Distributing search phrases across the pages of a resource is another of the tasks that semantic kernel building solves. You determine which page most accurately responds to a particular search query or group of queries.

You can approach this task from different angles.

Creating a site structure based on the results of the analysis of user search queries. The semantic core will determine the framework and architecture of the resource.
Pre-planning the resource structure prior to the search query analysis. Semantic core will be distributed on the ready framework.

Of course, it is more logical to plan the structure of the site first, and then determine the queries by which users will be able to find this or that page. In this case, you choose what you want to tell potential customers, that is, you remain proactive. If you tailor the structure of the resource to the keys, then you remain an object and respond to the environment, rather than actively changing it. So both approaches work, it’s just a question of efficiency and rationality.

“Seoshnic” and marketing approaches to building a core are drastically different. Here’s the logic of the “seoshnik”: to create a site, you need to find keywords and choose phrases that are easy to get to the top of extradition. After that, you need to create the structure of the site and distribute the keys on pages. Page content must be optimized for key phrases.

A businessman or marketer will reason this way: you need to decide what information to broadcast to the audience with the website. This requires a good knowledge of your industry and business. First you need to plan the approximate structure of the site and a preliminary list of pages. After that, when building the semantic core, you need to know how the audience is looking for information. With content, you need to answer the questions the audience is asking.

In the first case, the information value of the resource drops. SEOs unreasonably screen out some of the promising search queries. This is due to high competitiveness or low KEI, low frequency, “junk” keys, etc. And businesses have to shape trends and choose what to tell customers. Businesses should not limit themselves to reactions to search phrase statistics and create pages just for the sake of optimizing the site for some key.

The list of key queries distributed on the pages of the site, which is actually the result of building a semantic core, contains the URLs of pages, search queries and an indication of their frequency.

The structure of the site is a hierarchy of pages. With the help of a hierarchical scheme solves several problems. You plan the information policy and the logic of presenting information, ensure the usability of the resource, ensure that the site meets the requirements of search engines.

In order to build the structure, use a tool that is convenient for you: table editors, MS Word, other software, or take a sheet of paper and draw the structure on it.

Answer two questions:

What information do you want users to know?
Where should this or that information block be published?

For example, that you are planning the structure of the website of a small confectionery. The resource includes information pages, a publications section and a product showcase or catalog. Visually, the structure may look like this:

Next, make the structure of the site in the form of a table. Specify the names of pages and indicate their subordination. Also include in the table:

columns for specifying page URLs,
keywords
their frequencies.

The table might look like this:

You will fill in the URL, Keyword, and Frequency columns later. Now move on to the keyword search.

You need to understand what keywords are and what keys your audience uses. With this knowledge you will be able to correctly use one of the keyword selection tools.

What kind of keys does the audience use.

Potential customers use certain words and phrases to find relevant information. These are the keys.

For example, to make a cake, a user enters “napoleon recipe with photo” into the search box.

Keywords are classified in several ways. According to various data, search phrases are grouped as follows:

К low-frequency are queries with a frequency of up to 100 impressions per month. Some experts include in the group requests with a frequency of up to 1000 impressions.
К mid-frequency are queries with a frequency of up to 1000 impressions. Sometimes experts increase the threshold to 5000 impressions.
К high-frequency The phrases with a frequency of 1,000 or more impressions are considered to be high-frequency queries. Some authors consider high-frequency keywords with 5,000 or even 10,000 queries.

Well, according to popularity there are high-frequency, medium-frequency and low-frequency queries.

The difference in frequency estimates is related to the different popularity of topics. If you create a core for the online store, which sells laptops, the phrase “buy samsung laptop” with a frequency of about 6 thousand a month will be the average frequency. If you create a core for the site of the sports club, the query “aikido section” with a frequency of impressions of about 1000 requests will be high-frequency.

According to various data, two-thirds to four-fifths of all user queries are low-frequency. So you need to build the broadest possible semantic core. In practice, it should be constantly expanding at the expense of low-frequency phrases.

However, without the high and medium-frequency queries, you can not do without. But as the main resource for attracting targeted visitors, consider low-frequency keys.

According to the needs of users keys are grouped into such groups:

Informational. Such keys are used by the audience to find some information.

Examples: “how to properly store baked goods,” “how to separate the yolk from the white.”

Transactional. When the user plans to perform some action, he enters the keys of this group.

Example: “buy a bread machine”, “download a recipe book”, “order a pizza with delivery”.

Other inquiries. Key phrases by which it is difficult to determine the user’s intent. When a person uses, for example, the key “cake”, he may be planning to buy a culinary product, cook it himself, or just interested in the definition, features, etc.

Some experts identify navigational queries as a separate group. With their help, the audience is looking for information on specific sites. Here are some examples: “laptops svyaznoy”, “city express track delivery”, “register on LinkedIn”. Non-specific navigation queries for your business can be ignored when compiling the semantic core.

How can this help when building a semantic core?

You must consider the needs of your audience when assigning keys to pages and creating a content plan.Information section publications must respond to informational queries. There should be a large portion of key phrases without expressed intent. Transactional queries should be answered by pages from the “Store” or “Showcase” sections.
To attract natural traffic for the query “buy a Samsung smartphone,” you have to compete with Euroset, Eldorado and other business heavyweights. Use the recommendation above to avoid unequal competition. Maximize the core and reduce the frequency of queries. For example, the frequency of the query “buy Samsung Galaxy s6 smartphone” is an order of magnitude lower than the frequency of the key “Buy Samsung Galaxy smartphone.

What you need to know about the anatomy of search queries

The body, specifier, and tail are the parts that make up search phrases.

Example, “cake”.

you can’t determine the user’s intent from this query,
The query is high-frequency, which determines the high competition in the results.

Using this query for promotion will bring a large share of untargeted traffic, which has a negative effect on behavioral metrics. The query consists only of the body, which determines its high-frequency and non-specificity.

Example, “buy a cake.” A query consists of the body “cake” and the specifier “buy”. It is the specifiers that indicate whether the key belongs to transactional or informational ones.

“Buy cake with delivery”. “Buy” is a specifier, “cake” is the body, “with delivery” is the tail.

In each case, one can see the person’s intention to purchase the cake. And the tail of the key phrase details that need.

Knowing the anatomy of search phrases allows you to derive a conditional formula for keyword selection for your semantic core. Define basic terms related to your business, product, and user needs.

Then find the tails and specifiers that the project audience uses with the base terms. “Tailed” phrases will increase your reach and reduce core competitiveness.

There are many keyword selection tools available, both paid and free. Choose a particular tool depending on the tasks you are facing.

Key Collector

This tool is essential if you are engaged in internet marketing professionally, developing several sites or composing the core for a large site.

Tasks the program solves:

Keyword Selection. Key Collector collects queries through Wordstat “Yandex”.
Parsing search cues.
Cutting off inappropriate search phrases with stop words.
Filtering queries by frequency.
Search for implicit duplicate queries.
Identifying seasonal queries.
Gathering statistics from third-party services and platforms: Liveinternet.ru, Metrika, Google Analytics, Google AdWords, Direct, Vkontakte, and others.
Search for relevant pages.

Key Collector is a multifunctional tool that automates the operations required to build a semantic core. This program is paid and if you want to avoid additional costs, with the help of alternative free tools you can perform all the actions that Key Collector “knows how”. But you will have to use several services and programs.

SlovoEB

This is a free tool from the creators of Key Collector. The program collects keywords through Wordstat, determines the frequency of queries, parses search cues.

To use it, specify your username and password from your “Direct” account in the settings. Do not use the main account, as Yandex can block it for automatic queries.

Create a new project.
On the “Data” tab, select the “Add Phrases” option.
Specify the search phrases that the project audience is expected to use to search for product information.

Under the Keyword & Statistics collection menu, select the desired option and run the program. For example, determine the frequency of keyword phrases.

SlovoEB allows you to pick keywords and perform some tasks connected with analysis and categorization of queries automatically.

Yandex keyword matching service

Wordstat is a service through which you can select and analyze keywords. The tool is free and has the following functionality:

You can see the statistics of impressions by keyword, search phrases that include the specified keyword. And you can also analyze general data or data on requests from the mobile audience.

The service shows data about queries by region.

Displays data about the popularity of the request in dynamics. The corresponding information is available in the “Query history” section.

Displays statistics for the specified phrase and its forms only. To get the data, put the key in quotes.

Without the forms, displays statistics for the key. Add an exclamation point before the search phrase to get the data.

Displays data for the key excluding minus-words. To get statistics for the query “kill a mockingbird” without the word “movie”, you need to put a minus in front of the last one.

Using the selected preposition shows query statistics. To get the data, put a “+” sign before the preposition.

To get statistics for a group of queries, specify groups of queries in parentheses, and separate key variants with “|” sign. For example, to get statistics for the queries “buy a cake,” “order a cupcake,” and “order a cake,” and do it quickly, use the construction shown in the illustration.

Shows query statistics linked to the specified regions.

With the help of this service you can pick up the key phrases you need to build a semantic core.

Use Wordstat if you’re ready to analyze and group queries manually.

Google Keyword Planner

To build a semantic core taking into account the needs of users of the Google search engine, use a keyword planner. For example, it makes sense to use Google statistics in regions where this search engine dominates.

The keyword planner allows you to search for new queries by query, site or topic. You can use it to collect statistics on search queries. And you can also use the scheduler to create new query combinations and predict traffic.

Select the appropriate option on the main page of the service to get statistics on your existing queries.

Enter the phrases of interest to you;
upload the file in CSV format;
select the query region;
if necessary, specify the minus-words;
click the “Learn the number of queries” button.

Mail.ru Search query statistics

The service shows the statistics of queries in the third most popular search engine in Runet. The information is presented by gender and age of users. The statistics for the past calendar month is available.

Serpstat

A multifunctional SEO service. With its help you can analyze key phrases and select medium and low-frequency queries. The service is paid and works with Google and Yandex. The user specifies the geography of the project.

Serpstat’s analysis produces the following data:

Key Frequency.
The level of competition on a scale of 1 to 100.
Approximate cost per click on the contextual ad.
The dynamics of the popularity of the query in the form of a graph.
List of competitors in the search engine and contextual advertising systems.

You can pick up additional phrases for the base key, this is done in the SEO analysis section. Serpstat gives you similar phrases and search cues. And you can also see a list of pages from the top output for the selected query and appreciate the dynamics of the visibility of competing sites in the output for a given key.

The Content Marketing section provides “long-tail” keyword phrases and content ideas. The service collects groups of search queries based on a given base key. For example, the “how to build websites” group contains popular queries like “how to start building a website” or “how to make money building websites.” That is, Serpstat actually generates ready clusters of key phrases.

“Mutagen.”

The “Mutagen” service allows you to search for keyword phrases with high frequency and low competition. The tool is paid, but existing users have access to 10 free keyword checks per day. However, if you are a new user, you need to top up at least once to access the free checks.

SemRush

A paid comprehensive search analytics tool. In beta-testing mode, the Keyword Magic Tool is available, which selects keyword phrase ideas for the base key, evaluates the frequency of phrases, the cost of switching in Google AdWords, the level of competition, the Google output for the queries at the moment. The collected keys can be exported to a table or added to Keyword Analyzer for further analysis.

The data on the selected keys can be obtained from the Keyword Analyzer section:

Keyword frequency.
Regional distribution.
Dynamics.
Cost in Google AdWords. SERP status.

This data can be exported in PDF format.

SemRush for the base keyword will allow you to get related queries and view ad history in AdWords. You can assess the level of competition for the keyword in the Keyword Difficulty section. How difficult it is to take a place in the top of search results for the chosen key, the system will estimate on a scale from 1 to 100.

Keyword Tool

A tool for working with keyword queries. Also paid. This service searches for search phrase ideas for Google, YouTube, Bing, Amazon, App Store and eBay. The paid subscription gives access to frequency, dynamics, cost in AdWords and the level of competition for the selected keys. The free version, on the other hand, displays only a list of keys.

Both versions support exporting data in .xls or .csv formats.

Rush Analytics.

Rush Analytics picks and clusters keyword phrases. Automatically analyzes keywords from Yandex’s “Matching Words” service. The tool is paid, a test period of 14 days is available.

In the settings of the project you can specify the parameters of key evaluation:

add operators
fix the word order
collected keys and statistics can be exported

The program collects cues from Google, YouTube and Yandex. These cues provide an opportunity to expand the semantic core and they can be loaded into a table.

With Rush Analytics, you can cluster your clues. The data can be uploaded to a table.

There are also features for generating TORs for creating texts with the collected keys in mind.

SEMparser

Key clustering service. Paid, but after registration a free test package is available. In order to cluster keys, you need to load a table and run the project.

You can manually edit keyword groups and export them to a table. With SEMparser you can get parameters of the reference texts from the top search results for a selected query. With SEMparser you can obtain the parameters of the reference texts from the top of search results for the selected query and the service will automatically generate a ToR for the creation of texts based on the results of clustering and analysis of search results.

Analytics Systems

Metrica or Google Analytics systems can help you build your site’s semantic core. Using them, you can determine what search phrases are used by the existing audience of the project.

The Yandex and Google webmaster accounts provide data on popular search queries. In Search Console, the information can be found under “Search traffic – Search query analysis”. In Webmaster, under “Search Traffic – Popular Queries”.

Competitor site analysis services

A great source of keyword ideas are competitor sites.

You can manually search for and identify the search phrases of the specific page of interest that the page is optimized for. To find the main keywords, you can usually just read the material or check the contents of the keywords meta tag in the page’s code. Or you can use semantic text analysis services such as Istio or Advego.

If you need to analyze the whole site, use comprehensive competitive analysis services:

A few examples of other tools for gathering key phrases:

There is no point in mastering all services and programs at once. If you have your own small site, use a free tool to compile a semantic core, such as Yandex keyword selection service or Google Planner.

Several steps in the keyphrase selection process:

Determining the basic keywords with which your audience searches for your product or business
Expanding the semantic core
Removing inappropriate search phrases

Defining basic keys

Put common search phrases related to your business and products into a spreadsheet or write them down on paper. Have a brainstorming session and record all the ideas.

Your list will look something like this:

Most of the keys in the picture are high frequency and low specificity. To get medium- and low-frequency search phrases with high specificity, you need to expand the core as much as possible.

Expanding the semantic core

You can solve this problem by using, for example, the Wordstat tool. Select the appropriate region in the settings, if your business has a regional affiliation.

Analyze all the keys recorded in the previous step. Do this with the help of the keyphrase selection service.

From the left column of Wordstat, copy the phrases and paste them into the table. In the right column, “Yandex” offers phrases that people used along with the main query. Depending on the content, you can immediately select the appropriate keys from the right column or copy the entire list. In case you copy the whole list, inappropriate queries in it will be filtered out in the next step.

In the end, this step will produce a list of search phrases for each basic key that you have fixed in your brainstorming session. The lists may contain hundreds or thousands of queries.

Removing inappropriate search phrases

This step is the most time consuming. You will need to manually remove from the core inappropriate search phrases.

Do not use frequency, competitiveness or other purely “seosh” metrics as key evaluation criteria.

Let’s look at why old-school optimizers don’t think certain search phrases are good? For example, take the keyword “diet cake.” Wordstat service predicts it will have 3 hits per month in the region of Cherepovets.

To promote pages for specific keys, old-school SEOs bought or rented links. This approach is still used by some professionals. Obviously, low-frequency search phrases do not, in most cases, pay for the money spent on buying links.

Now look at the phrase “diet cakes” through the eyes of the average marketer. Some members of the CA of the confectionery firm are really interested in such products, so the key should be included in the semantic core. The phrase will come in handy in the product descriptions section, if the confectionery prepares the corresponding products. Well, if the company for some reason does not work with diet cakes, the key can be used as a content idea for the information section.

Examples of phrases that can be safely excluded from the list:

Keys mentioning competing brands.
Keys mentioning products or services that you do not sell and do not plan to sell.
Keys that include the words “inexpensive,” “cheap,” “discounted.” If you’re not dumping, cut out those who like cheap, so as not to ruin behavioral metrics.
Duplicate keys. For example, of the three keys “custom birthday cakes,” “custom birthday cakes,” and “custom birthday cakes,” leave out the first one.
Keys mentioning inappropriate regions or addresses. For example, if you serve residents of the Northern district of Cherepovets, you do not fit the key “cakes to order industrial district”.
Phrases entered with errors or typos. Search engines understand that the user is looking for croissants, even if he enters the key “croissants” into the search box.

You’ve got a ready-made query list for the basic keyword “cakes to order” after removing the irrelevant phrases. Now you need to make the same lists for the other basic keys.

Search phrases with which users find or will find your site are grouped into semantic clusters. These are groups of queries that are similar in meaning. For example, the semantic cluster “Cake” includes all key phrases related to this word: cake recipes, order cake, cake photos, wedding cake, etc.

A semantic cluster is a group of queries united by meaning. It is a multi-level structure. Within the cluster of the first order “Cake” there are clusters of the second order “Cake Recipes”, “Cake Ordering”, “Cake Photos”. Inside the second-order cluster “Cake Recipes” theoretically there is a third order of clustering: “Cake recipes with mastic”, “Cake recipes for sponge cakes”, “Cake recipes for sand cakes”. The number of levels in the cluster depends on the extensiveness of the subject. In practice, in most topics it is enough to allocate business-specific clusters of the second order inside the first-order clusters.

When, during the brainstorming session, you wrote down the basic key phrases, you identified most of the clusters.

It’s very important to do the clustering correctly at the second level. Search phrases are modified with specifiers that denote user intent.

For example, the clusters “cake recipes” and “cakes to order”. The search phrases of the first cluster are used by people searching for information. The keys of the second cluster are used by customers who want to buy a cake.

Using Wordstat and manual screening, you have identified the search phrases for the “cakes to order” cluster. You need to distribute them among the pages of the “Cakes” section.

For example, the cluster has the search terms “custom soccer cakes” and “custom football-themed cakes”.

You need to create a corresponding page in the “Mastic Cakes” section, if there is a corresponding product in the assortment of the company. Enter it into the structure of the site: specify the name, URL and search phrases with frequency.

See what other search phrases potential customers are using to find football-themed cakes. Do this by using the Keyword Selection service or similar tools.

In the list of search phrases cluster in a way that is convenient for you, mark the allocated keys. Allocate the remaining search phrases.

Change the structure of the site, if necessary: create new sections and categories. For example, the page “cakes to order puppy patrol” should be included in the section “Children’s cakes”. However, at the same time it can also be included in the section “Cakes from mastic”.

The cluster may not have the right phrases for the page you plan to create. For example, this can happen because of imperfect search phrase collection tools or incorrect use of them, or because of the low popularity of the product.

If there isn’t a suitable product in the cluster, that doesn’t mean you should give up creating the page and selling the product. For example, a confectionery company sells children’s cakes depicting the characters of the cartoon “Peppa Pig”. If the list does not include the relevant keys, you need to clarify the needs of the audience with Wordstat or other service. In most cases, suitable queries will be found.

After removing unnecessary keys in the cluster, there may still be search phrases that do not fit the created and planned pages. You can ignore them or use them in another cluster. For example, if the confectionery for some reason in principle does not sell cake “Napoleon”, the relevant key phrases can be used in the section “Recipes”.

At the end you have a list of existing and planned site pages with URLs, search phrases and frequency. Next, let’s figure out what to do with it.

A table made with a semantic core should be the main source of ideas when making a content plan. You have a list with preliminary page titles and search phrases. They define the needs of the audience. When creating a content plan, all you need to do is refine the title of the page or publication. Include the main search term. It’s not always the most popular key. In addition to popularity, the query in the title should best reflect the need of the page’s audience.

Use the rest of the search phrases as the answer to “what to write about.” Content should reveal the topic and answer users’ questions.It is important to focus on information needs, not on search phrases and how they fit into the text.

The list of detrimental actions covers the stages of building a semantic core and planning and implementing a content strategy:

Refusing to use overly competitive keywords. You don’t have to be at the top for the competitive query “tours to Egypt” to attract traffic and sell. Keep the phrase in your semantic core and use it as a content idea.
Refuse to use low-frequency phrases. As mentioned above, representatives of the old school consider queries junk if they do not pay back the money invested in the purchase or rental links. And you need to collect as many low- and ultra-low-frequency keywords as possible. They, too, will become your content ideas.
Using formulas and ratios to estimate keywords. For example, keyword effectiveness index (KEI) is the ratio of popularity to competitiveness. And semantics is a branch of linguistics. Leave the formulas to the mathematicians. Otherwise, you’ll lose a lot of valuable content ideas, which may get discarded along with keys with supposedly low effectiveness or KEI.
Creating pages for the sake of queries. By “buy” and “order” queries you mean one intention. That’s why you need one page.
Trying to automate the construction of the semantic core as much as possible. Without careful manual analysis, the value of the key list is low. A machine can collect phrases and their characteristics, but only a human can analyze the industry, really assess the level of competition and plan information campaigns. The latter, by the way, is related to the construction of the site structure and the distribution of keys.
Excessive focus on keyphrase collection. To get content ideas, at first, small new projects only need to use one source, such as Yandex or Google statistics. There’s no need to spy on competitors, study search cues, and collect keys from different search engines. The need for alternative sources of keys will appear when you have a shortage of ideas and want to expand your core.

Understanding exactly why you’re building a core is enough to keep you from turning into an old-school optimizer.

The mindset of a businessman or marketer and some of the skills of an optimizer will help you build search phrases competently. Pay attention to the structure of the site, understanding what you want to communicate to customers. Figure out how audiences search for the information you plan to publish. Keyword phrase selection services and programs, statistics sources, and search engine spying tools can help you do this. Analyze the collected keywords and remove phrases that do not fit the meaning. Group the keys and distribute them among the pages of the site. Next, proceed to the creation of a content plan.

Source link