Keyword research is not about collecting the most keywords. It is about deciding which searches deserve a page, which existing pages should target them, and which opportunities are worth your time.
A keyword with thousands of monthly searches can still be a poor choice. The search intent may not match your content, the SERP may answer the query directly, or a page you already have may cover the topic. This guide covers the full process, from finding keywords in Google Search Console to checking whether the page that targets them can actually rank.
What is keyword research?
Keyword research is the process of finding and evaluating the search queries people use to find information, products, and services, then deciding how your website should respond to them.
The output is not a spreadsheet. It is a keyword map: each priority keyword cluster connected to the one page responsible for it, with a reason that page deserves to exist.
Keyword vs. search query
A search query is the exact phrase a person types. A keyword is a term you choose to target and track. One keyword usually represents dozens of related queries, including long-tail variations no tool reports. That is why a page rarely ranks only for the keyword it was written for.
Step 1: Start with your existing search data
If your site already earns impressions, your first keyword tool is Google Search Console. The Performance report connects queries with impressions, clicks, click-through rate, and average position, and it reflects real searches rather than estimates.
Look for four patterns:
- Striking distance queries. Positions 8 to 20 with meaningful impressions. These pages are already relevant; they often need better coverage, stronger internal links, or a clearer title rather than a new URL.
- High impressions, low CTR. Check why before rewriting. The page may rank too low, the query may be broad, or a featured snippet or AI Overview may answer it directly.
- Unexpected queries. Queries you never targeted reveal subtopics. If they fit the page, expand it. If they represent a different intent, they may justify a separate page.
- Declining queries. Compare equal date ranges. Check whether demand fell, the SERP changed, another of your URLs took over, or a technical change affected the page.
Useful Search Console filters
In the Performance report, use Query, Custom (regex) to isolate query types:
| Purpose | Regex |
|---|---|
| Question queries | ^(who|what|why|how|when|where|which|can|does|is)\b |
| Comparison queries | \b(vs|versus|best|alternative|alternatives|compare)\b |
| Long-tail queries (5+ words) | ^(\S+\s+){4,}\S+$ |
Question queries are candidates for FAQ sections or supporting articles. Comparison queries signal commercial intent that an educational page may not satisfy.
Step 2: Build a keyword list
Expand from seed keywords: broad terms that describe your products, services, or expertise. For an SEO tool, seeds might be keyword research, technical SEO, site audit, and SEO monitoring.
- Google Search Console: queries that already produce impressions
- Google Keyword Planner and Google Trends: demand estimates and seasonality
- Autocomplete, related searches, and People Also Ask: how people phrase the problem
- Ahrefs, Semrush, or similar tools: search volume, keyword difficulty, parent topics, competitor data
- Support tickets, sales calls, and communities: the language customers actually use
Record each keyword with its source, estimated volume, likely intent, and any URL that already ranks for it. Collect broadly here; filtering comes next. To turn keyword ideas into article ideas, see how to find SEO content topics.
Step 3: Cluster keywords by SERP overlap
Most guides say to group keywords by intent but never say how. The most reliable method is to let Google show you: if two queries return many of the same URLs, Google treats them as the same need, and they usually belong on one page.
| Keyword | Likely intent | Probable cluster |
|---|---|---|
| what is keyword research | Learn the concept | Main guide |
| how to do keyword research | Follow a process | Main guide |
| keyword research tools | Find tools | Tools page |
| best keyword research tools | Compare products | Tools page |
| keyword research with search console | Learn a method | Tutorial or guide section |
Step 4: Read the SERP before choosing a keyword
The four intent labels, informational, navigational, commercial, and transactional, are a starting point. The SERP is the evidence. For each cluster, check the page type that ranks (guides, product pages, comparisons, tools, definitions), the depth and audience, the SERP features present, and whether the results answer one need or several.
If comparison pages dominate, a definition will struggle. If detailed tutorials dominate, a short landing page will struggle. A well written page that offers the wrong type of answer rarely wins.
Step 5: Decide whether the keyword needs a new page
Finding a relevant keyword does not mean publishing a new article. Existing pages come first, because a page that already has impressions usually ranks faster than a new URL starting from zero.
Step 6: Evaluate volume, difficulty, and competition
Keyword tool metrics are useful inputs, but each one leaves something out.
| Metric | What it tells you | What it cannot tell you |
|---|---|---|
| Search volume | Estimated monthly demand | Whether the traffic is valuable or clickable |
| Keyword difficulty | A tool's estimate of ranking effort, mostly from backlinks | Whether your page and site can rank |
| Competition (Keyword Planner) | Advertiser competition for paid ads | Organic ranking difficulty |
| Traffic potential | Estimated clicks to the top page across its cluster | How much traffic your page will get |
Volume is an estimate, so a zero-volume long-tail keyword can still bring qualified visitors. Difficulty is not a prediction; open the ranking pages and judge them yourself. And topical authority changes the math: a site that already ranks across a cluster usually wins a new related query faster than a stronger domain with no coverage of the topic.
Step 7: Account for AI Overviews and zero-click searches
Search volume says nothing about how many searchers will click. When the SERP shows an AI Overview, a featured snippet, or a direct answer, many informational queries end without a visit.
Before committing to a keyword, ask whether the SERP answers the query completely, whether the query needs depth, examples, or a tool, and whether the pages cited in the AI Overview match the organic results. Prioritize how-to and decision topics over pure definitions, and include first-hand evidence that answer engines are more likely to cite. For prompts and conversational queries, see how to research keywords for AI search and traditional SEO keywords vs. AI search queries.
Step 8: Find content gaps, not just keyword gaps
Competitor keyword analysis shows queries where other sites rank and yours does not. That is a keyword gap. A content gap is narrower: a real need your site does not serve well. A competitor may rank for a query your existing page already covers, and then the opportunity is improving that page, not publishing a duplicate.
Start with your search competitors, which may differ from business competitors. A keyword gap is worth pursuing only when the topic is relevant to your audience, your site can give a credible answer, and the expected value justifies the effort. If a competitor ranks for 1,000 keywords you do not, the right response is not 1,000 new pages.
Step 9: Check for keyword cannibalization
Keyword cannibalization means several of your pages compete for the same query. Two pages ranking for one keyword is not automatically a problem, since they may serve different intents. In Search Console, filter by the query and open the Pages tab to see which URLs receive impressions. Overlapping pages often show up first as duplicate content or duplicate title tags in an audit.
| Situation | Action |
|---|---|
| Different intents | Keep both, sharpen each page's focus, and link between them |
| Same intent, one stronger page | Merge the weaker content in and 301 redirect the old URL |
| Same intent, neither stronger | Combine them into one better page |
Step 10: Prioritize by opportunity, not volume
Score each keyword cluster from 1 to 3 on relevance, opportunity, feasibility, and business value. A total of 10 to 12 is high priority, 7 to 9 is worth planning, and 6 or below can wait. This is an internal planning aid, not a ranking formula.
Step 11: Map keywords to pages, then check the page can rank
A keyword map assigns each priority cluster to one URL. It is the document that stops your team from writing the same article twice.
| Primary topic | Supporting queries | Intent | Page | Action |
|---|---|---|---|---|
| Keyword research | what is keyword research, keyword research for seo | Learn and follow a process | Main guide | Create |
| Keyword research tools | best keyword research tools | Compare products | Tools comparison | Create |
| Search Console keyword research | find keywords in search console | Learn a method | Dedicated tutorial | Create later |
| Keyword cannibalization | keyword cannibalization fix | Solve a problem | Technical SEO guide | Expand section |
A complete map also records current URL, current performance, priority, and a review date.
The step most guides skip: technical eligibility
A perfect keyword choice fails if the target page cannot rank. Before assigning a keyword to a URL, confirm it passes these checks:
- IndexableNo noindex tag, not blocked by robots.txt
- Self-canonicalThe canonical tag points to the page itself
- 200 statusNo redirect chain, soft 404, or error status
- Title and H1 matchBoth reflect the target topic (titles, H1s)
- Internally linkedRelated pages link to it; it is not one of your orphan pages
- UniqueNot a near-duplicate of another mapped URL (duplicate content checker)
A technical SEO audit in SiteAuditLint flags these issues across every URL in your map at once. For a manual pass, use the indexability checklist.
Step 12: Measure, compare, and refine
After a page has had time to appear in search results, often several weeks, compare its target keywords with its actual queries. The difference is query drift, and it is one of the most useful signals in SEO, because it shows what Google thinks the page is about.
Track queries producing impressions, clicks, CTR, average position, which URLs receive visibility for each cluster, and the dates of content changes and deployments, as part of your SEO change management log. Account for seasonality and SERP changes: a drop in impressions does not always mean the page got worse.
Comparing crawl snapshots from before and after a content update or release connects ranking changes to their likely cause. When a page loses visibility the same week its canonical changed, the fix is technical, not editorial. Audit comparison in SiteAuditLint shows those changes side by side. For the ongoing process, see our guide to SEO monitoring.
Common keyword research mistakes
Every common mistake has a better question behind it.
- Picking keywords by volumeWill this traffic click and convert?
- Trusting difficulty scoresCan our site beat these specific pages?
- One page per variationDo these queries share ranking URLs?
- Skipping the SERPWhat page type does Google reward here?
- Copying competitor listsIs this a content gap or just a keyword gap?
- Ignoring existing pagesDoes a page already rank for this?
- Targeting a broken pageIs the page indexable and canonical?
- Never revisiting the mapWhat queries does the page attract now?
Keyword research checklist
- Search Console reviewed for striking distance, low CTR, unexpected, and declining queries
- Keyword list expanded from seeds using several sources
- Keywords clustered by SERP overlap
- SERP checked for page type, depth, and features
- Existing pages checked before planning new ones
- Volume, difficulty, and traffic potential evaluated
- AI Overview and zero-click risk assessed
- Content gaps separated from keyword gaps
- Cannibalization checked between your own URLs
- Clusters scored and prioritized
- Each cluster mapped to one URL
- Target pages confirmed indexable, canonical, and internally linked
- Target keywords compared with actual queries after publishing
Frequently asked questions
- What is the purpose of keyword research in SEO?
- It identifies the searches your audience makes and helps you decide which topics to cover, which pages should target them, and which opportunities to pursue first.
- How do you do keyword research for SEO?
- Start with Search Console and seed keywords, expand the list with tools, cluster by SERP overlap, check intent in the results, evaluate demand and competition, prioritize, map each cluster to one page, and measure after publishing.
- What is the difference between keyword research and keyword mapping?
- Keyword research finds and evaluates opportunities. Keyword mapping assigns each one to the page responsible for it, which prevents duplication and cannibalization.
- How do I know if two keywords need separate pages?
- Compare their top 10 results. If they share 3 or more URLs, one page usually serves both. If they share few or none, Google treats them as different needs.
- How many keywords should a page target?
- One primary topic. A page can rank for hundreds of related queries that share its intent, so cover the topic thoroughly instead of counting keywords.
- Should I target low-volume keywords?
- Yes, when they are relevant. Long-tail keywords often represent specific problems and qualified visitors, and many have more demand than tools report.
Key takeaway
Start with the search data you already have. Let the SERP tell you how to group and target keywords. Check existing pages before creating new ones, and confirm the target page can technically rank.
The goal is not to target the most keywords. It is to build the right pages for the searches that matter, and to keep them working.