Keyword research: SEO method and tools

What is keyword research?

Keyword research is the step that turns an offer into a list of queries to work on. It answers three questions. What do the people you want to reach type? How many of them are there? What content do they expect in return?

Google gives the principle in its starter guide: think about the words a user might search for to find your content. An expert and a beginner do not phrase them the same way (Google, SEO Starter Guide, December 10, 2025). The same guide states that Google does not use the meta keywords tag. Keywords are there to help you write for the right people, not to fill a tag.

The expected result fits in a short list of prioritised topics, each tied to a page. It guides writing and internal linking over the following months, where a 2,000-row spreadsheet guides nothing.

What criteria should you use to choose a keyword?

Volume catches the eye, but it is never enough. A good keyword combines three things: real demand, an intent your page can serve, and value for your business.

Criterion	Question to ask	Where to find it
Volume	How many searches per month, on average?	Keyword Planner, SEO tools, API
Difficulty	How much competition on page one?	SEO tool score, reading the SERP
Intent	Is a guide, tool, product or service expected?	The SERP itself
Business relevance	Can this audience become customers?	Your knowledge of the offer
SERP features	AI Overview, local map, videos: is there still room for a link?	The SERP itself

Read search volume with caution

The volume shown by tools usually comes from Google Ads. The DataForSEO documentation describes it as an approximate monthly average. It states that Google Ads provides combined volumes for groups of similar keywords (DataForSEO, Search Volume). Values are also rounded into brackets.

Elev8 Lab’s data from September 24, 2026 shows this, on French queries. “cocon sémantique” and “cocon semantique” (with and without the accent) each show 880 monthly searches. “pagerank” and “page rank” each show 1,000. These are not two demands that add up, but the same group counted twice. Adding up variants inflates the potential.

How do you do keyword research?

The method fits in seven steps. It works for a new site as well as for a redesign. On an existing site, start step 2 with your own data.

1. List your topics: offers, customer problems, questions asked to sales and support, the actual vocabulary of your customers.
2. Expand the list: Google suggestions, People also ask, related searches, keyword tools, Search Console queries.
3. Collect the data: average volume, difficulty, seasonal trend, for each query.
4. Qualify the intent by opening the SERP for the main queries.
5. Group the queries whose SERP shows the same pages: they are handled on a single page.
6. Assign a page to each group, existing or to be created. Check that no other page already targets the same group.
7. Prioritise: potential, difficulty and business value, in a dated plan.

Steps 4 and 5 make the difference between a list of words and a strategy. They rely on analysing search intent. The result naturally takes the shape of a semantic cocoon: a parent page, child pages, one intent each.

Example: Elev8 Lab’s keyword study

Elev8 Lab planned its first 50 pieces of content from a DataForSEO data pull on September 24, 2026. It includes the volumes and difficulty of each target query, for the French market. Excerpt:

Query (French)	Average monthly volume	Difficulty (0 to 100)	Decision
seo	22,200	44	Pillar page, reference guide
audit seo	1,600	21	Dedicated sub-pillar
cocon sémantique (semantic cocoon)	880	5	Article, rework of existing content
maillage interne (internal linking)	590	0	Article, rework
recherche de mots clés (keyword research)	320	29	Article, with “mots clés seo” (SEO keywords, 390) as secondary
google search console	135,000	54	Article focused on the long tail: the query is mostly a brand search

The last row shows the volume trap. The 135,000 searches for “google search console” mostly come from people who want to open the tool. The intent is navigational, and the tool’s official page serves it first. Our Google Search Console guide therefore targets usage questions, less searched but within reach.

Which tools should you use for keyword research?

No tool gives the truth: each has its source and its rounding. What matters is to cross-check a source of volumes, a source of real queries and the SERP.

Tool	What it provides	Limit
Google Keyword Planner	Keyword ideas and volumes from Google Ads	Designed for advertising; requires a Google Ads account
Google Search Console	Real queries, impressions, clicks and position for your site	Only the queries where you already appear
Google Trends	Seasonality and relative change of a topic	Relative index, no absolute volume
Suggestions and People also ask	Real phrasings and sub-questions	No volume
SEO tools (Semrush, Ahrefs, Haloscan, SE Ranking)	Volumes, difficulty, competitors’ keywords	Paid; estimates specific to each vendor
DataForSEO API	Raw data at scale	Billed per call; requires some technical skills

Google’s Keyword Planner opens from a Google Ads account. Search Console, for its part, shows the queries that already generate impressions, with their clicks (Google, Get started with Search Console, December 10, 2025). On an established site, it is the first source to use. It reveals queries that no third-party tool attributes to you.

Is keyword research changing with AI?

Queries are getting longer and more conversational, and a single question hides several others. Google says its AI Overviews and AI Mode can run several related searches on subtopics: this is “query fan-out”. It adds that no special optimization is needed to appear in them (Google Search Central, December 10, 2025).

In practice, keyword research keeps its place, but it broadens to questions. People also ask in the SERP, customer verbatims, subtopics of a query: all of these become sections to cover on the page. This is the raw material of good SEO writing.

What mistakes should you avoid?

• Choosing on volume alone, without opening the SERP or checking the intent.
• Creating one page per variant (“keyword”, “keywords”, “SEO keywords”) instead of one page per intent.
• Targeting queries unrelated to your offer just to “get traffic”.
• Repeating the keyword until the text is saturated.
• Forgetting to date the study and review it every year.

Google describes two of these mistakes. Keyword stuffing appears in its spam policies (Google, spam policies, August 28, 2026). Its documentation on helpful content also warns against entering a topic “without any real expertise”, simply because it seemed likely to bring traffic (Google Search Central, December 10, 2025).

Once chosen, keywords go where they count. The title tag, the H1, the subheadings and the first lines. On an existing site, a semantic audit checks that each page carries the group of queries assigned to it. All these topics are gathered in the content and semantics sub-pillar.