Keyword Research Is Intent Research

Most keyword research starts in a tool: type a seed term, export a long list, sort by volume. The output looks rigorous, but it answers the wrong question first. Volume tells you how often a phrase is typed. It doesn't tell you what the person typing it wants, whether your page could satisfy them, or whether they're anyone you need to reach. Better keyword research starts from intent and uses tools to check and size what you find.

Start from problems, not phrases

Before opening any tool, list the situations in which someone would need what you offer, and how they'd describe the problem in their own words. Sales conversations, support emails, reviews, and forum threads are better sources for this than a brainstorm, because they capture vocabulary your customers actually use. Professionals often describe their work in terms buyers never search for.

Read the results page to learn the intent

The classic intent categories (informational, navigational, commercial, transactional) are useful, but the most reliable evidence of intent is the results page itself. Search the term and look at what ranks:

• If the top results are how-to guides, the intent is to learn, and a sales page won't compete.
• If they're product or category pages, people want to compare or buy.
• If they're local listings and a map, location is part of the intent.
• If they're mixed, the query is ambiguous, and the page that ranks will be the one that resolves the ambiguity best.

Google has already run the experiment on what satisfies that query. Matching the format and depth of what ranks is usually the price of entry; doing it better is how you win.

Use the tools for what they're good at

Google Search Console: your own data

For an existing site, the Performance report in Search Console is the most valuable keyword source available, and it's free. It shows the actual queries your pages appear for, with impressions, clicks, and average position. The most productive opportunities are usually queries where you already get impressions but rank on page two, or rank well but get few clicks because the title doesn't fit what people wanted.

Google Keyword Planner: relative size, not precise counts

Keyword Planner was built for advertisers. Accounts without meaningful ad spend typically see volume as broad ranges (for example, 1K to 10K monthly searches) rather than specific figures. Even the specific figures are rounded estimates. Use it to compare the relative size of options and to find related phrasings, not to forecast traffic.

Third-party tools: modeled estimates

Tools such as Ahrefs and Semrush estimate volume and difficulty from their own data and models. They're useful for competitor research and for discovering terms you hadn't considered. Their numbers aren't measurements, and different tools disagree, so compare options within one tool rather than across them.

The results page again

"People also ask" boxes, related searches, and autocomplete suggestions show the subtopics and follow-up questions people associate with a query. They're often a better guide to what a page should cover than any tool's keyword list.

Drop the "LSI keywords" idea

A persistent piece of SEO advice says to sprinkle "LSI keywords" (supposedly from latent semantic indexing) through a page. Google representatives have stated plainly that Google doesn't use LSI keywords, and the technique dates from information retrieval research long before modern search engines. The real principle underneath is sound: a page that thoroughly covers a topic naturally uses the related vocabulary. Write for completeness and clarity, and the related terms appear on their own.

Long-tail queries and "zero-volume" terms

Long, specific queries each have low volume, but together they make up a large share of searches, and their intent is usually clearer. A page that fully answers one specific question often ranks for dozens of variations of it. Don't dismiss terms a tool reports as zero volume, either. In specialized B2B and professional fields, a handful of the right searchers a month can matter more than thousands of the wrong ones, and the tools' estimates are least reliable at the low end.

Organize by intent, one page per need

Group keywords by the need they express, not by shared words. Each group should map to one page. Two pages chasing the same intent compete with each other (often called cannibalization), and neither ranks as well as one strong page would. When you find overlap on an existing site, consolidate the pages and redirect the weaker URL to the stronger one.

Measure what the keywords are for

Rankings fluctuate by location, device, and personalization, and they're a means, not an end. Track clicks and impressions in Search Console, and in your analytics track whether visitors from those queries do what the page is for: read, sign up, call, buy. A keyword that brings traffic that never converts is a content decision to revisit, not a win.

Once you know what a page should target, the next step is making that obvious on the page itself; see what still matters in on-page SEO.