When launching a new site or tackling a theme on an existing blog, the classic reflex is to open a keyword tool, type in a generic query, and retrieve the raw list. The problem is that this list says nothing about what Google actually understands behind each query. Finding relevant keywords for SEO today requires cross-referencing several signals even before writing the first line.
Filter keywords by intent signal, not by volume
Most guides start with search volume. In practice, we find that queries with modest volume convert much better than generic high-traffic terms. The reason lies in intent: a user typing “women’s waterproof trail running shoes” knows what they want, while someone searching for “shoes” is still browsing.
Read also : Tips and advice for enhancing a pear-shaped bust
Google itself is pushing in this direction. Match types now work based on meaning, not characters. Even with exact targeting in Google Ads, the platform broadens the triggered queries through its AI. On the organic SEO side, the logic is the same: the engine groups variants around a common intent.
Specifically, before validating a keyword, we check the results page. If Google displays product listings, the intent is transactional. If it shows guides, the intent is informational. A keyword whose SERP does not match your content format is an unnecessary keyword, regardless of its volume. Knowing how to find effective keywords starts with this manual check of the results page.
Related reading : Effective Tips and Practical Advice for Adjusting an Oversized Soccer Jersey
Long-tail queries in SEO: leveraging natural formulations
Queries composed of four words or more represent a growing share of search traffic. This phenomenon is accelerating with voice search and conversational engines, where users ask complete questions instead of typing two words.
To capture them, we don’t just rely on a suggestion tool. Here are the sources that provide real formulations:
- Google’s autocomplete suggestions (search bar) and the “People also ask” section, which reflect frequent queries in your theme.
- Forums, comments, and customer reviews in your industry, where users employ raw vocabulary that keyword tools do not always capture.
- Google Search Console on an existing site: the “Performance” report lists the actual queries that trigger impressions, including long-tail variants that you might not have thought of.
The best long-tail opportunities often come from queries that are already visible but poorly exploited. If a page appears in position 15 for a specific query, it’s a signal: the content exists, it just lacks targeted optimization.
Keyword prioritization: the field sorting grid
Having a list of 200 keywords is useless without a sorting method. We use a simple grid that crosses three criteria for each term.
Business relevance and response capability
Does the keyword correspond to a topic on which you can provide a complete and concrete answer? If you sell windows in Lyon, “wood or PVC window Lyon” is a term where your expertise is direct. “Thermal insulation regulations 2025” is much less so, unless you are also a certified installer.
Actual difficulty on the results page
We look at who occupies the top positions. If they are only high-authority sites (Wikipedia, national media, marketplaces), the likelihood of ranking with a new site is low. Favor queries where at least two results on page 1 are comparable sites to yours in size and reputation.
Conversion potential
A purely informational keyword (“what is SEO”) attracts traffic but rarely generates a direct action. A keyword like “SEO agency quote Lyon” has a lower volume, but each visitor is close to making a decision. We balance the editorial calendar between the two, without over-investing in one or the other.
Free tools for keyword research: which provide reliable data
Free tools are sufficient to get started, provided you know their limitations. Google Keyword Planner remains the reference for volume ranges, but it heavily rounds the data if you don’t have an active campaign. The displayed figures are orders of magnitude, not precise measurements.
Ubersuggest offers suggestions and a difficulty estimate. Feedback varies on the reliability of its difficulty score, but the suggestion feature remains useful for expanding an initial list. Chrome extensions like Keywords Everywhere or Keyword Surfer display volumes directly in the SERP, speeding up sorting without leaving Google.
Cross-reference at least two data sources before validating a keyword. A term that appears in Keyword Planner with a decent volume, which you also find in Google suggestions and in Search Console, deserves to be worked on. A term present in only one tool may be an artifact.
Structuring web content around keyword groups
A common mistake is to create a page for each keyword. Google groups close variants under the same intent. Publishing three pages on “SEO keywords,” “natural referencing keywords,” and “keywords SEO” creates cannibalization: your pages compete against each other in the results.
We group terms by thematic cluster. A pillar page covers the main topic, while satellite pages delve into sub-themes. The internal linking between these pages enhances the visibility of the whole. For an e-commerce site, the cluster can be a category (pillar page) linked to advice sheets (satellite pages). For a blog, it’s a comprehensive guide linked to specialized articles.
This grouping also facilitates updates: when a query evolves or a new formulation emerges, we know exactly which page to adapt without starting from scratch.
Keyword research is not a one-time exercise done at the launch of a site. Search behaviors change, long-tail queries multiply, and Google’s algorithms interpret terms in increasingly semantic ways. Reviewing your keyword list every quarter, cross-referencing Search Console and a suggestion tool, remains the most reliable way to keep organic SEO aligned with what users are actually searching for.