Collecting your own semantic core

A keyword is a single word or phrase capable of correctly describing a certain page or ad. A word’s value is determined by its popularity, and its main characteristics are frequency, word form, and exact and phrase.

First and foremost, keyword selection is required in order to define a semantic core – a list of keywords with their forms and combinations that describe the structure of a business and the search terms on which users can find pages on a site. The semantic core takes into account the interests and desires of users, matching the goals of the project.

Creating a semantic core is a fairly complex process, in which you first need to form a cloud describing the basic services of the tags, and then, using the relevant frequency, select high-frequency queries that form the logical backbone of the entire breakdown of queries. When compiling your own semantics, you need to follow a certain workflow.

Collecting basic queries

Basic keywords are the main phrases and words that match the subject resource. You need to make a list of key phrases that fully describe the product or site. You can use different services such as Yandex. Wordstat. The final list should be saved in a separate file, adding to its self-compiled phrases.

Collect semantics should be primarily for the main sections of the site. To identify them, you must consider the following indicators:

1. Seasonality. The categories should be selected for elaboration at least three months before the beginning of the season.
2. assortment. Work with a category with more offers.
3. frequency. Choose sections that have a sufficiently high number of monthly impressions by phrase frequency in the promotion region.

Extension of the semantic core

All of the selected phrases from the compiled list should be made the basis of query groups. Such groups should be assembled and expanded through paraphrasing and the use of synonyms.

Removing unnecessary words

Once the keyword database is collected, it should be cleaned as not all of the words exactly match the subject of the resource, which may result in attracting a non-target audience. To get rid of redundant keywords, a list of stopwords should be compiled, which include:

• mentions of countries and cities in which it is not planned to operate;
• mentions of competitors;
• non-target words for commercial resources, such as “picture”, “download”, “what”, “how”, “which”, “list”, “review”, “instruction” and so on;
• repetitions;
• misspelled words;
• implicit duplicates, i.e. phrases that consist of the same set of words arranged in a different order.

Stop words are removed manually. Software filters can be used to speed things up.

Query grouping

In order to identify patterns in the distribution of keys on individual pages, queries need to be grouped together to form semantic clusters. These are groups of queries that are similar in meaning and have a multi-level structure. For example, the first level cluster “bed linen” includes clusters of the second level “blankets”, “plaids”, “pillows”, “sheets”, “mattress covers”, and so on.

Usually, the keys from the first and second level clusters are allocated already at the first stage of work, because it is enough to have a good understanding of your own product. For ease of compilation, you can be guided by the structure of competitive sites. The semantics of the remaining sublevels is determined at the stage of a careful compilation of the core and its clustering.

It should be taken into account that each group of queries belonging to the last level should meet a single need of the user, i.e. a specific product. For this reason, when sorting the phrases into categories, use words that can later become the name of the category or landing page.

Groups of queries can be combined. For example, creating separate categories for “beautiful product” and “stylish product” is not rational, so it is better to combine them. To determine the importance of the phrases for the categories, copy them into the relevant section of Google’s Keyword Planner, which allows you to check the frequency of the query. All queries can be divided into:

Low-frequency (LF) queries. They contain five or more words and are rarely used by users. They are as specific as possible and have a lot of specifications. It is advantageous to use low-frequency queries when setting up contextual advertising, as high clickability and conversion is possible at a minimal cost per click.

2. Medium-frequency (MF) queries. They consist of two to four words. They contain less specificity than low-frequency, but it is still present, expressed in the names of sections and categories of directories and pages. With competent use of phrases in the text of the promotion of BF-queries is quite easy.
3. High-frequency (HF-queries). These are the most popular queries consisting of one or two words. SEO on them the most difficult, expensive, and long, because it requires a considerable amount of work.

There are no exact numbers to attribute a query to a particular group, and everything is determined by the subject of the site. For some sites, 1000 hits a month may be considered low-frequency (music, movies), while 200 hits may be considered high-frequency (finance).

The highest-frequency queries should be included in meta tags, while low-frequency ones should be used to optimize site pages. Since the latter has little competition, it is enough to do a good job with the texts to bring the relevant pages to the top.

After all the above manipulations, a detailed structure of the resource can be obtained, including the key phrases for

• landing pages;
• categories and subcategories;
• information pages.

The XMind service can be used to visualize the structure of the site.

Using alternative methods of semantics expansion

Along with the growth of the site should increase and semantic core, for which you can use alternative sources of semantics, which include related queries, search terms, and similar phrases. Collecting keywords is easier and quicker within a single group.

Tools for keyword research

A number of tools can be used to collect semantics:

1. Yandex.Wordstat. Facilitates the selection of basic queries. After entering the desired phrase in the right column will display similar queries that users usually specify along with a given query.

To make working with the service easier, you can install Wordstat Helper plugin (for Google Chrome and Mozilla Firefox), which allows you to select appropriate queries and save them as a ready list, avoiding manual copying.

2. Keyword Planner from Google. Can work in two modes – searching for new keywords and viewing the number of queries and predictions. To run the search, you need to enter several keyword phrases suitable for a group of queries. The resulting word choices can be downloaded.

The frequency check is carried out under “Retrieving query and trend statistics”.

3. Key Collector. This program is included in the basic package of self-promotion of the site. It can be used not only to compose a semantic core but also to test promoted queries for a number of parameters (frequency, competition, number of main pages in the output, link budget for aggregators, etc.), to divide queries into groups and check the position of the resource in search engine results. The program is flexibly configurable. All information is unloaded into Excel.

You can also use the Key Collector to make a list of stop words.

Clustering is available in the program.

4. Rush-Analytics or Just-Magic. These are paid services allowing automatic clustering. They are optimal for broad semantics.
5. Serpstat. Provides data on keyword phrases, allowing a better understanding of the audience. With the help of this service, it is possible to highlight a number of synonyms to the keys, which were not taken into account in the initial search, and choose the best topics for the blog, picking up search cues.

Using an advertising analytics service

An important additional source of semantics is competitors’ websites, which should be carefully analyzed. A number of free services offering such a service, among others, can be used for this purpose. However, as they are aimed at highlighting a wide range of keywords, some important words present in the competitors are often overlooked, which reduces the effectiveness of the promotion. Using the ad analytics service from SpyCon allows you to collect all the key phrases missed during your own work in the context of multiple competitors.

When performing the analysis, the service takes into account the peculiarities of competitor resources such as:

1. type of site. Most sites belong to one of the following types:

• Landing pages consisting of a basic sales page and may include several auxiliary non-promotional pages;
• business cards sites with a small number of service pages and service pages that contain information about the activities of the company
• service sites and corporate web resources containing a small number of pages with services and hidden functionality available to registered users, and they are promoted through a forum or blog, branded queries, and traffic links;
• online shops that focus on selling products and their delivery, promoted by means of sales queries and have an extremely voluminous structure that may take up to several weeks to be evaluated manually;
• boards/portals/catalogs, which are similar to online shops but are aimed at attracting not only buyers but also sellers.

2. The structure of the site. Its analysis allows us to reveal a number of important points:

• The way in which the categories of services and products are separated on the page;
• the features of the service pages, such as the presence of guides for buyers, promotions, and delivery;
• features of the blog page categories;
• The principle of making the URL, which for sites that are in the top, usually in the form of CNC, ie the name of the links to identify the content of the page.

3. Content. When analyzing the content, the commercial part is determined, because it has the greatest importance for the conversion. From her highlights keywords that can later be used to create new content.
4. UTP. When analyzing competitors, it is important to highlight the advantages that make their pages more attractive than others. These may include price, lead time or delivery times, working conditions, guarantees provided, available forms of payment, and various promotional offers.

The selection of a complete list of competitive keywords allows for a more efficient grouping of queries. To do this, after identifying search queries from the top sites, common pages are searched and then, if a certain number of pages are available, the keywords are grouped together. The collected queries are further divided into commercial and informational queries, which significantly increases conversion rates, as it allows the user’s intent – the intent behind the keyword – to be highlighted more accurately.