What Is Keyword Research?
Keyword research is the process of finding, evaluating, and prioritizing the words and phrases people use when they search, so that content and site structure follow real demand instead of guesses. It answers a question intuition can get wrong: how does your audience describe what you offer when they search for it? It is not a one-time project. Terms, products, and search results change, so the research feeds content planning on a schedule.
The Process
- Start with seed terms. Write down the core topics, products, and services in plain language before opening a tool.
- Expand the list with keyword tools, search suggestions, “People also ask” boxes, competitors’ pages, and your customers’ own words from support tickets, reviews, and forums.
- Group by intent. Put terms together by what the searcher is trying to do, not by shared words, and let the results page settle close calls. “Running shoes review” asks for a different page from “buy running shoes online,” while “running shoes” and “buy running shoes online” may both be served by one category page if their results overlap.
- Filter by relevance, competition, and value. Drop terms that do not fit the business, then weigh the rest by how realistic ranking is and what ranking would be worth.
- Map groups to pages. Assign each group to an existing or planned page, and check for overlap so two pages do not chase the same intent.
Intent Before Volume
Search volume tells you how often a term is searched, not whether the searchers want what you offer or are ready to act. A lower-volume term with clear buying or comparison intent can be worth more to a business page than a higher-volume informational term. As an illustration, not a measurement: a thousand visits from people ready to buy can produce more revenue than ten thousand visits from people learning about a topic. Your own conversion data shows the real difference.
Look at the results page before committing to a term. If shopping results, a video block, or an AI-generated summary dominate the page, the opportunity for a standard result can be smaller than the volume suggests. Note what appears above the standard results for each priority term and weigh it with intent and difficulty.
Why Tools Disagree
The same keyword can show three different volumes in three tools. Each builds its figures from its own data and method, so the figures can differ. Compare terms within one tool rather than across tools, and treat volumes as relative.
For terms your pages already earn, Search Console reports Google’s own data: the actual queries that produced impressions and clicks. It is not complete. Google’s Search Central blog explains that some queries, which it calls anonymized queries, are left out for the privacy of the people who searched them.
Reading Difficulty Scores
The major third-party keyword databases publish a keyword difficulty score, and each calculates it differently, for example from the links of the pages that currently rank. Google’s own tools in the table below do not: Keyword Planner is built for advertisers, and Trends shows relative interest. Use the scores to sort a long list. For the terms that survive, look at the actual results: who ranks, what format their pages take, and whether your site can offer something better.
Tools
| Tool | What it contributes |
|---|---|
| Google Keyword Planner | Search frequency for terms and how it has changed over time, from Google Ads |
| Google Search Console | The queries already bringing your pages impressions and clicks |
| Google Trends | Relative interest over time and by region |
| Ahrefs, Semrush, Moz | Large keyword databases, difficulty scores, and grouping |
Google describes Keyword Planner as giving “insight into how often people search for certain terms” and how those searches have changed over time. Google Trends works differently. Its data FAQ explains that it uses “a largely unfiltered sample of actual search requests,” that “Google Trends normalizes search data to make comparisons between terms easier,” and that each data point is compared with all searches for its time and place: “The resulting numbers are then scaled on a range of 0 to 100 based on a topic’s proportion to all searches on all topics.” Trends shows relative popularity, not search counts.
Mistakes to Avoid
| Mistake | Why it hurts |
|---|---|
| Sorting by volume alone | Ignores whether searchers want what you offer |
| Building for the wrong intent | A page that answers a different question does not serve the searcher who asked this one |
| Treating difficulty scores as exact | They are each tool's estimate, not your site's real difficulty |
| Ignoring the results page | A term can look easy in a tool and be hard on the actual page |
| Researching once | Demand and wording change; lists go stale |
| Chasing phrases instead of topics | A page that covers a topic well can earn many related terms |
Where the Research Goes
- Content planning: which topics get a page, and in what order.
- Site structure: how groups of related terms become hub pages and supporting pages linked together.
- On-page wording: which terms belong in titles, headings, and text, used naturally; Google’s spam policies treat keyword stuffing as spam.
- Measurement: which queries bring impressions and clicks over time, read in Search Console, and which organic landing pages bring conversions, read in analytics or a CRM. Search Console does not report conversions, so a conversion figure for a single query is an inference, not a measurement.
Questions About Keyword Research
How does conversational and AI search change the research?
Google’s documentation on AI features explains that AI Overviews and AI Mode may answer one question by running several related searches across subtopics; Google’s name for this is query fan-out. That makes the questions and topics behind a search worth researching, not only short phrases. Finding what people ask and grouping it by intent still applies.
What is the difference between keyword research and keyword strategy?
Research finds and evaluates terms. Strategy decides which terms map to which pages, in what order, and how to keep pages from competing with each other.