How Is Keyword Analysis Done? (With Free Tools)

What should keyword analysis produce?

The output isn't just an Excel column. Every group should show the core topic, the search intent, the fitting page type, the existing or planned URL, the priority, the evidence source and the measurement criterion. Without that information, a content team can open several pages for the same intent and turn its own results into rivals.

The phrases in the table are not real volume claims; they're hypothetical examples for an Azerbaijani B2B service. The point is showing how the same word "CRM" can turn into four different decisions. A separate URL isn't needed for every variation; close phrases serving the same intent can merge into one strong page.

Which free tools can be used?

Choose the tool by the question it answers, not the number it produces. For a business with an existing site, the most valuable start is usually Search Console, because it shows impressions and clicks already earned on Google, not hypotheses. A new site will lean more on Keyword Planner, Trends and the results page.

Per the official documentation, Google Keyword Planner gives new phrases related to the business and site, estimated searches and targeting costs. Those figures are primarily for ad-campaign planning. Don't read them as an organic-traffic forecast or a revenue promise.

In Google Trends, the number 100 doesn't mean "100 searches." Per the official Trends data explainer, values are normalised against all searches in the chosen time and region, then scaled to 0–100. So Trends is strong for direction and comparison, not exact volume.

Search Console doesn't show all queries either. Google's current data documentation says some queries are anonymised for privacy and only the main rows are kept in the table. When the table totals and the query-row sums differ, don't treat it as a measurement error.

How is keyword analysis done in 7 steps?

1. Write the offer and the audience in one sentence

"SEO services for everyone" is not a useful analysis input. Write whose problem you solve, in which geography and with which service. If that answer is weak, first define the target audience; a tool can't make a vague business precise.

2. Extract five to ten seed topics

Start from product names, the problems customers voice, usage scenarios, alternatives, prices and risks. A seed topic isn't a direct SEO title. It's the starting folder where new queries will collect.

3. Expand the phrases from three sources

For an existing site, export the Search Console queries, expand the seeds in Keyword Planner and compare season and direction in Trends. Then add the natural language from customer emails, sales meetings and support questions. The intersection of tool data and real customer language gives more useful topics.

4. Group the queries by intent

Separate informational, comparison, commercial and transactional intent. "What is SEO" and "SEO service pricing" belong to the same topic family, but not to the same decision. The first is searched for learning, the second for choosing and budgeting.

5. Check the Google results page by hand

Review the first screen and the first ten results in the target region. Note the page types, headlines, dates, features like maps and video, and the sub-questions left unanswered. Look not to copy the competitor, but to understand how Google interprets the intent and what you can add.

6. Map each intent group to one URL

If a fitting existing page exists, refresh it instead of opening a new URL. If a new page is needed, choose one of the pillar, cluster, service or comparison formats. That decision joins the content system and the SEO architecture in one map.

7. Assign priority and a measurement criterion

Score each group by business fit, intent clarity, capacity for proof and existing opportunity. Then write the URL, the owner, the deadline and the primary measure. After publishing, track impressions, clicks and the fitting business event; don't close the work with a position count alone.

How are keywords prioritised?

Putting the difficulty and volume columns side by side and choosing the biggest number isn't enough. A small-volume query very close to the offer can bring more fitting inquiries with less traffic. Conversely, a broad informational query can create awareness but may not deliver direct sales.

Build a simple editorial score: give each criterion 0, 1 or 2. This isn't Google's ranking formula; it's an internal decision tool clarifying why the team picks one topic before another.

A topic scoring 8 isn't published automatically. Legal risk, author competence, resources and conflicts with existing pages are checked separately. The priority score shortens the discussion; it doesn't replace the editor's decision.

The most common keyword analysis mistakes

• Repeating the same phrase in every paragraph and counting that as relevance.
• Opening a separate weak URL for every long variation.
• Putting volume above business value and intent.
• Missing Azerbaijani-language variations, spelling differences and local customer language.
• Copying competitor headlines without adding experience, facts or tools.
• Ignoring Search Console's anonymised and off-table query limitations.
• Opening a new article without checking the existing page.

In Google's people-first content approach, a specific word count and keyword density are not success conditions. The official self-assessment questions foreground original information, complete answers, clear authorship and usefulness to people. Placing more phrases replaces none of those.

The same foundation holds in AI-driven search. Google's 2026 AI optimization guide explains AEO/GEO not as a separate secret but as SEO's continuation; it demands no special schema, mandatory "chunking" or artificial page length. Instead of mass-converting query variations into separate pages, build clear, sourced answers for AEO and GEO that stand up even out of context.

If your pages have already multiplied, run the SEO mistakes audit first. Canonical, index and internal-link problems can technically weaken a keyword map's seemingly correct decision. At that stage the technical SEO checklist shows the next work order.

Practical extra: the SEO and content operations system

Instead of building the article's process from a blank sheet, you can use the workbook I prepared in the Anar Rustamli design system. The file contains a 100-topic keyword map, a 25-week sample publishing calendar, a 25-point quality gate, the internal-link architecture, technical SEO checks and a sources sheet. The dates and statuses are examples, editable for your own site.

The file doesn't end with finding topic ideas; it keeps the URL, status, owner, quality evidence and technical checks in the same system. When planning execution time and cost, run a separate resource calculation for SEO timelines and budget.