Keyword research is the foundation of search engine optimization. Without sound keyword research, it is impossible to build a website structure, create content, or evaluate the results of promotion. At the same time, working with keywords is not a one-off task. It is a process that starts before a website launches and continues throughout its life. This guide covers keyword research, from basic concepts to mapping keywords to pages and common mistakes that prevent sites from reaching the top.
What keywords are and why they matter
Keywords are words and phrases that users enter into a search bar to find information, a product, or a service. From an SEO perspective, keywords are the bridge between what users are looking for and what your website offers.
An SEO specialist's job is to determine exactly which words and phrases the target audience uses and optimize website pages for them. If a user searches for “washing machine repair in Kazan,” but your website describes the service as “household appliance maintenance,” the search engine simply will not connect the query with your page.
Why research keywords
- Understand demand. Keywords show exactly what potential customers are looking for, how they phrase their searches, and how often they search.
- Build the structure. The semantic core determines which pages you need to create and how they will be connected.
- Optimize content. Knowing the keywords lets you write texts that answer real queries rather than cover made-up topics.
- Evaluate competition. Analyzing keyword volume and competitiveness helps you choose realistic targets.
- Measure results. Rankings for keywords are the main indicator of SEO effectiveness.
A semantic core is the complete set of keywords and phrases for which a website should be visible in search engines. Collecting and refining it is the first step in any SEO strategy.
Types of keywords
Keywords can be classified according to several criteria. The most common classification is by search volume.
By search volume
| Type | Search volume (Yandex) | Example | Characteristics |
|---|---|---|---|
| High-frequency (HF) | from 10 000 impressions/month | “apartment renovation” | Broad, competitive, vague intent |
| Mid-frequency (MF) | 1 000 – 10 000 | “turnkey apartment renovation price” | More specific, moderate competition |
| Low-frequency (LF) | 100 – 1 000 | “40 sq m apartment renovation in a new building” | Specific intent, low competition |
| Micro-frequency (MF) | fewer than 100 | “renovation of a 30 sq m Khrushchev-era one-room apartment” | Very specific, almost no competition |
The boundaries between categories are approximate and depend on the niche. For a highly specialized B2B website, 500 impressions per month is a high-frequency query, while for an electronics online store it is a micro-frequency query.
Long tail
The long-tail concept holds that the combined traffic from low-frequency queries exceeds the traffic from a few high-frequency queries. Long-tail queries usually contain 4–7 words, have low competition, and convert well.
Example:
- HF: “laptop” — enormous competition, and it is unclear what the user wants
- Long tail: “buy a laptop for programming for under 80000 rubles” — clear intent, high likelihood of purchase
A long-tail strategy is particularly effective for new websites that cannot compete with large portals for high-frequency queries.
By user intent
Another important classification is based on the user's search intent:
- Informational — the user is looking for information: “how to configure a router,” “what is a VPN.” Suitable for blogs, FAQs, and reference materials.
- Navigational — the user is looking for a specific website or brand: “Wildberries account,” “GitHub login.” It is difficult and rarely effective to rank for another brand's navigational queries.
- Transactional — the user is ready to take action: “buy iPhone 16,” “order pizza.” These are the most valuable from a commercial perspective and are suited to product pages and service pages.
- Commercial investigation — the user is comparing options: “best CRM 2026,” “Bitrix24 vs AmoCRM.” Suitable for reviews and comparison pages.
Understanding intent makes it possible to create content that precisely matches user expectations. A page with informational content targeting a transactional query is a waste of resources.
Keyword research tools
There are many tools for building a semantic core. Here are the main ones, from free options to professional platforms.
Yandex.Wordstat
A free tool from Yandex that provides search query statistics. It is the main tool for working with Russian-speaking audiences.
What it provides:
- Query volume (overall, by region, and by device)
- Related queries (the “Also searched for” column)
- Query history (seasonality)
- Regional filters
Wordstat operators:
"ремонт квартир"— phrase matching (words in any order, any forms)"!ремонт !квартир"— exact word form+в— force the inclusion of prepositions and conjunctions-евро— exclude a word
Limitations: Wordstat provides data only for Yandex. Additional sources are needed to build a complete semantic core.
Google Keyword Planner
A free tool within Google Ads. It provides data on Google search queries.
What it provides:
- Estimated search volume (ranges: 100–1000, 1K–10K, etc.)
- Click and impression forecasts
- Suggestions for new keywords based on a URL or seed query
- Cost per click (CPC) — an indirect indicator of commercial value
Limitations: Exact search volume figures are available only when active advertising campaigns are running. Without an advertising budget, the tool shows approximate ranges.
Ahrefs
A professional SEO analysis platform. One of the most powerful tools for working with keywords.
Features:
- Keywords Explorer — keyword research with exact search volume, a difficulty score (Keyword Difficulty), and clickability estimates
- Content Gap — find keywords that competitors rank for but your website does not
- Site Explorer — comprehensive analysis of the keywords any domain ranks for
- Clustering by parent topics (Parent Topic)
Cost: from $99/month. A significant investment that pays off when working on several projects or a large website.
Serpstat
A multifunctional SEO platform with a strong presence in CIS markets. A good balance between functionality and price.
Features:
- Keyword research with search volume and competition estimates
- Competitor analysis — domains, pages, and keywords
- Keyword clustering
- Website audit
- Rank tracking
Cost: from $59/month. Limited free access is available.
Keys.so
A Russian service specializing in search visibility analysis. It works particularly well with Yandex and Google data for the Russian-language internet.
Features:
- Complete keyword data for any domain — all the keywords for which a website is visible in search
- Compare the keyword data of several domains
- Find competitors based on keyword overlap
- Visibility trends
Cost: from 4 900 ₽/month. It is focused on the Russian market, so its data on the Russian-language internet is often more complete than that of international services.
Additional sources
In addition to the main tools, you can collect keywords from:
- Search suggestions — autocomplete in Yandex and Google as you type a query
- “Related searches” block — at the bottom of the search results page
- Google Search Console / Yandex.Webmaster — actual queries for which your website already appears
- Internal site search — what users already on your website are searching for
- Forums and Q&A sites — the actual wording used in questions by your target audience
- Reviews and comments — the language customers use to describe the product
Step-by-step process for collecting keyword data
Building a semantic core is systematic work that is best done in a specific order.
Step 1. Identify seed queries
Start with a brainstorming session. Write down all the words and phrases that come to mind when you think about your product, service, or website topic.
Sources for seed queries:
- Product and service names
- Catalog categories and subcategories
- Problems your product solves
- Questions customers ask
- Professional terminology and jargon
For an electronics online store, seed queries might include: “laptop,” “smartphone,” “headphones,” “tablet,” “smartwatch,” and “charger.”
Step 2. Expand the list using tools
Run each seed query through Yandex.Wordstat and Google Keyword Planner. Collect all relevant suggestions, related queries, and variations.
Practical tip: use Wordstat with the * operator to find unexpected combinations. For example, ноутбук * программирования will show variations such as “laptop for programming,” “laptop for programming work,” and so on.
Step 3. Analyze competitors
Choose 5–10 competitor websites and export their keywords using Ahrefs, Serpstat, or Keys.so. This will provide a substantial set of keywords that you may have missed during manual research.
Step 4. Combine everything in one spreadsheet
Combine data from all sources in one spreadsheet (Excel, Google Sheets). Record the following for each keyword:
- The keyword itself
- Search volume (basic Wordstat)
- Search volume (exact, using the
"!слово"operator) - Source (Wordstat, competitor, or suggestions)
- Intent (informational, transactional, or navigational)
Step 5. Clean up the list
Remove irrelevant queries:
- Queries that mention competing brands (if you do not plan to target them)
- Queries for other regions (if you run a local business)
- Queries with zero search volume
- Duplicates and obvious variations of the same query
- Irrelevant homonyms (for example, “key” as a tool vs. a search query)
Clean up the search queries in the spreadsheet and check demand in the selected source using the same region and operators. The page word and phrase analysis tool on reChecker is needed later: it counts occurrences in the text of a specific URL. These figures are not Wordstat search volumes, do not show search demand, and do not determine a phrase's priority in the semantic core.
Step 6. Determine search volume and competition
For the remaining queries, collect exact search volume (Wordstat with operators) and assess competition (Keyword Difficulty in Ahrefs, competition level in Serpstat). This will help you set priorities in the next stage.
Keyword clustering
Clustering is the process of grouping keywords by semantic similarity. The goal is to determine which queries can be targeted on one page and which require separate pages.
Why clustering is necessary
Without clustering, it is easy to make two mistakes:
- Cannibalization — several pages on a website compete with one another for the same query. The search engine cannot determine which page to show and ends up showing none of them.
- Insufficient relevance — trying to target too many unrelated queries on one page. The page becomes unfocused and does not fully answer any of the queries.
Clustering methods
Manual clustering
A specialist groups queries manually based on logic and an understanding of the subject matter. Suitable for small semantic cores (up to 500–1000 queries).
Advantages: accuracy and consideration of business-specific nuances. Disadvantages: slow, subjective, and not scalable.
SERP-based clustering
The most popular automated method. The principle: if two queries have the same URLs in the top 10 search results, they can be promoted on the same page.
Types of SERP-based clustering:
- Soft — queries are grouped if they have at least N URLs in common with any query in the group. Creates large clusters.
- Hard — queries are grouped if they have N URLs in common with every query in the group. Creates precise but small clusters.
- Moderate — a compromise between soft and hard.
The overlap threshold (N) is usually set at 3–4 URLs in common out of the top 10.
Semantic similarity clustering
This method is based on analyzing word meanings, rather than search results. NLP algorithms and embeddings are used to determine the semantic similarity of queries. It complements the SERP-based method well, but does not replace it, because search engines may interpret semantically similar queries differently.
Grouping and prioritization
After clustering, you will have a list of keyword groups. Assign a priority to each group.
Prioritization criteria:
| Criterion | High priority | Low priority |
|---|---|---|
| Commercial value | Transactional queries | Informational queries |
| Search volume | High total cluster volume | Individual low-volume queries |
| Competitiveness | Realistic (a chance to reach the top) | Excessive (the top is dominated by giants) |
| Page readiness | The page already exists and needs optimization | A page needs to be created from scratch |
| Profit margin | High profit per conversion | Low or zero margin |
Start with commercial clusters that can deliver results sooner. Work on informational clusters in parallel through a blog and reference sections.
Competitor analysis
Competitor analysis is one of the most effective ways to expand your keyword set and find growth opportunities. Instead of guessing which keywords to target, you use existing data on what works for other websites in your niche.
How to find competitors
SEO competitors are not necessarily your business competitors. They are websites that rank for the same keywords as you. An online store may compete in search with a marketplace, a blog, or even a forum.
Ways to find SEO competitors:
- Enter your main queries in a search engine and see which sites are in the top 10
- Use the “Competitors” report in Serpstat or Ahrefs
- In Keys.so, use the feature for finding competitors based on overlapping keywords
What to analyze about competitors
1. Keywords the competitor ranks for
Export the competitor’s full list of keywords using Ahrefs (Site Explorer → Organic Keywords) or Serpstat. Filter by:
- Rankings (top 10 — keywords that bring the competitor actual traffic)
- Search volume (medium and higher)
- Your own coverage (keywords you do not rank for)
2. Pages that bring in the most traffic
The Top Pages report in Ahrefs shows which competitor pages attract most of their organic traffic. This can indicate which types of content work in your niche.
3. Content Gap
Content Gap (Ahrefs) or Missing Keywords (Serpstat) shows queries that your competitors rank for but you do not. This gives you a direct action plan: create or optimize pages for these queries.
4. Competitor site structure
See how the competitor has organized its catalog and which sections and subcategories it has created. This can help when mapping keywords to pages.
Practical example
Suppose you are promoting a law firm’s website. An analysis of three competitors using Keys.so showed that they all attract substantial traffic from informational queries: “how to contest a Russian traffic police fine,” “statute of limitations on a loan,” and “how to file for alimony.” You do not have pages on these topics. This is an opportunity: create a section with legal articles to cover these queries.
Mapping keywords to pages
Mapping is the process of assigning keyword clusters to specific pages on a site. Essentially, it is an optimization plan: which page will be promoted for which queries.
Mapping principles
One page — one cluster. Do not try to promote one page for dozens of unrelated queries. Each page should have a clear focus.
One target page per cluster. If two sections of your site are optimized for the same cluster, this is keyword cannibalization. Choose one page and focus your efforts on it.
The page type should match the intent. Transactional queries should lead to commercial pages (product pages, service pages); informational queries should lead to blog articles.
How to create a mapping plan
Create a table with the following columns:
| Cluster | Primary query | Additional queries | Page URL | Page type | Status |
|---|---|---|---|---|---|
| Bathroom renovation | bathroom renovation | full bathroom renovation, bathroom renovation cost | /uslugi/remont-vannoj | Service | Optimization |
| Bathroom design | bathroom design | bathroom design photos, bathroom design ideas | /blog/dizajn-vannoj | Article | Creation |
Statuses:
- Existing — the page exists and is already optimized
- Optimization — the page exists but needs improvement
- Creation — the page needs to be created from scratch
- Consolidation — several pages need to be combined into one
Typical mapping structure for a commercial site
- Homepage — brand + a high-volume query for the main area of business
- Category pages — medium-volume queries describing product groups
- Product pages / service pages — low-volume transactional queries
- Blog / reference section — informational queries
- Location pages — queries tied to a city or region
How to use keywords on a page
Collecting and clustering keywords is half the work. The other half is placing them appropriately on your site’s pages. Search engines assess how relevant a page’s content is to a query, and where keywords appear on the page plays a significant role.
Title (page title)
The title is the most important on-page element for search engines. It appears in search results as a clickable heading.
Guidelines:
- Put the primary keyword at the beginning of the title
- Length: 50–60 characters (so it is not truncated in search results)
- Each page should have a unique title
- Readability matters more than keywords: write titles for people, not robots
Example:
- Bad:
Ремонт квартир ремонт квартиры под ключ ремонт квартир цена - Good:
Ремонт квартир под ключ в Москве — цены от 5 000 ₽/м²
You can check how your title will look in search results using the SERP snippet preview. This is useful for making sure the title is not truncated and looks appealing.
Meta description
The description is not a direct ranking factor, but it affects CTR (click-through rate) in search results.
Guidelines:
- Length: 150–160 characters
- Include the primary keyword and one or two additional keywords
- Add a call to action or specific details (prices, time frames, benefits)
- Do not duplicate the title
Prepare unique titles and descriptions for the page that reflect its content and search intent. Check the result with the Meta tag checker.
H1 (level-one heading)
H1 is the main heading on the page itself. There should be one per page, and it should contain the primary keyword.
Guidelines:
- One H1 per page
- Include the cluster’s primary query
- It does not have to match the title (it can be longer and more detailed)
- Use natural phrasing
H2–H6 (subheadings)
Subheadings structure the text and define the topics of its sections. Use them for additional keywords from the cluster.
Example structure for a “bathroom renovation” service page:
- H1: Turnkey bathroom renovation in Moscow
- H2: Types of bathroom renovation
- H2: Bathroom renovation cost
- H2: Stages of the work
- H2: Portfolio — bathroom renovation photos
- H2: Frequently asked questions about bathroom renovation
Main text
Keywords should appear naturally in the body text. There is no need to count exact occurrences — write useful content that answers the user’s question. If the text covers the topic fully, keywords will appear in it organically.
Recommendations:
- Include the primary keyword in the first 100 words of the text
- Use morphological variants and synonyms
- Avoid unnatural phrasing just to include an exact-match keyword
- Structure the text with lists, tables, and callout boxes
You can check how densely keywords appear in the text using keyword analysis on reChecker — the tool shows how often words and phrases are used on the page. Compare the results with the reader’s needs and the keyword set you prepared: the number of occurrences alone does not measure how fully a topic is covered or prescribe a required density.
Image alt tags
The alt attribute of the <img> tag describes image content to search engines and screen readers.
Guidelines:
- Describe what is shown in the image
- Include a keyword if it is appropriate
- Do not spam — the alt tag should be informative
<!-- Плохо -->
<img src="photo.jpg" alt="ремонт ванной комнаты ремонт ванной ремонт ванных комнат">
<!-- Хорошо -->
<img src="photo.jpg" alt="Ванная комната после ремонта — укладка плитки и установка душевой кабины">Page URL
A URL should be readable and contain the keyword in transliteration.
Guidelines:
- Use Latin letters, lowercase, and hyphens instead of spaces
- Keep it short and clear:
/uslugi/remont-vannojinstead of/page?id=12345 - Avoid parameters and technical artifacts
- Make sure it reflects the page content
Internal links
Internal link anchors are another place to use keywords. When linking from one page to another, use meaningful anchors that contain keywords for the target page.
Example:
- Bad:
подробнее читайте [тут](/uslugi/remont-vannoj)— the link text does not explain what is on the target page. - Good:
мы также выполняем [ремонт ванной комнаты под ключ](/uslugi/remont-vannoj)— the link describes the service. This is an example of markup for a renovation company’s website.
You can check the quality of meta tags across all pages of your site using the meta tag checker, and use the word counter to analyze the amount of text content.
Common mistakes when working with keywords
Even experienced specialists make mistakes when working with keywords. Here are the most common ones.
1. Over-optimization (keyword stuffing)
Repeating a keyword in every sentence, using unnatural phrasing, or adding hidden text can all lead to search engine penalties. Yandex applies the “Baden-Baden” filter, while Google lowers rankings through the Panda and Helpful Content algorithms.
How to avoid it: write for people, not robots. If a phrase sounds strange when read aloud, rewrite it.
2. Ignoring intent
Creating a commercial page for an informational query, or vice versa, is a common mistake. If a user searches for “how to choose a washing machine,” they need an advice article, not a product page.
How to avoid it: check the search results for the target query before optimizing a page. What types of content are in the top 10? If the results contain only articles, create an article. If they contain product pages, create a product page.
3. Working only with high-volume queries
Focusing only on high-volume queries is a beginner’s mistake. High-volume queries are difficult to compete for, conversion is low, and results take a long time.
How to avoid it: build your strategy from low-volume to high-volume queries. Start with specific low-volume queries, attract traffic and build authority, then gradually target more competitive keywords.
4. Keyword cannibalization
When several pages are optimized for the same queries, they compete with each other. The search engine gets “confused,” alternates between the pages in search results, and neither reaches its highest possible rankings.
How to avoid it: maintain a mapping plan and make sure each cluster has exactly one target page.
5. One-time keyword research
The semantic core is not a static document. Search demand changes, new queries appear, and seasonality shifts. A core assembled a year ago may no longer be relevant.
How to avoid this: Review your keyword set every 3–6 months. Track new queries in Search Console and monitor changes in your niche.
6. No link to business goals
Collecting thousands of keywords without understanding which ones will lead to conversions is pointless work. Traffic for traffic’s sake is not the same as results.
How to avoid this: Evaluate keywords not only by search volume, but also by commercial value. The query “online mortgage calculator” may bring in thousands of visitors, but if you sell furniture, that traffic is useless.
7. Copying competitors’ keyword sets without adapting them
Exporting a competitor’s keywords and blindly copying their strategy is a bad idea. You have a different product range, different strengths, and different resources.
How to avoid this: Use competitors’ data as a source of inspiration, but adapt your strategy to your business. Look not only for overlap, but also for untapped niches.
8. Ignoring regional targeting
For a local business, it is critical to account for the geographic targeting of queries. “Pizza delivery” and “pizza delivery Yekaterinburg” are different queries, with different competition and audiences.
How to avoid this: Build your keyword set using regional filters. In Yandex Wordstat, select the desired region; in Ahrefs, select the target country and language.
Conclusion
Keyword research is not a mechanical process, but analytical work that requires an understanding of the business, its audience, and search algorithms. Even the best tools cannot replace a specialist’s expertise—they only provide data; people make the decisions.
Key principles to keep in mind:
-
Start with intent. Before chasing search volume, understand what lies behind a query. A page that directly meets the user’s need will outperform a page stuffed with keywords.
-
Work systematically. Collection → cleanup → clustering → mapping → optimization → monitoring. Skipping any stage reduces overall effectiveness.
-
Prioritize. Don’t try to cover everything at once. Start with the clusters that will deliver the greatest results with the least effort, then move on.
-
Use competitors’ data. Competitor analysis saves time and shows what actually works in your niche.
-
Update your keyword set. Search demand changes. Regularly check whether your core is still relevant and add new queries.
-
Place keywords thoughtfully. Title, H1, first paragraph, subheadings, alt tags, URL—each area matters, but keyword stuffing is more harmful than using too few keywords.
Sound keyword research is the foundation for a website’s visibility in search. Combine manual analysis with automated tools, focus on users’ needs, and results will follow.