What Is Keyword Research?

How Keyword Research Works

Keyword research runs in three moves: discovery, enrichment, and prioritization. Discovery starts from a handful of seed keywords — the obvious terms that describe what you do — and expands them into the long tail of real phrasings people use. Tools do this by mining query databases, autocomplete, and the terms a site already ranks for. The output of discovery is raw and long, often thousands of phrases, and most of them are noise.

Enrichment attaches data to each candidate so you can judge it. The two numbers that matter most are search volume, an estimate of how often a term is searched, and keyword difficulty, an estimate of how hard the current top results are to displace. A third dimension, search intent, asks what the searcher actually wants — a definition, a comparison, or a checkout page. Intent is the dimension most people skip and the one that most often decides whether ranking translates into results.

Prioritization is the real work, and it is subtraction. You score each candidate on volume, difficulty, and intent together, then cut everything that fails on any axis: too competitive to reach, too thin to bother, or wrong intent for what the page can offer. What survives gets grouped and mapped. Because Google understands synonyms and related concepts rather than matching literal strings, many phrasings of one idea belong on a single page — so research increasingly means clustering terms by shared intent, not treating each string as its own target.

The Metrics That Drive Prioritization

• Search volume — average monthly searches for a term and its close variants. In Google Keyword Planner this is averaged over a 12-month window by default and, without ad spend, shown as a range rather than an exact figure.
• Difficulty — how strong the incumbent results are. A term you have no realistic chance of ranking for is not an opportunity, however high its volume.
• Intent — informational, navigational, commercial, or transactional. It determines what kind of page can satisfy the query, and whether ranking will ever produce revenue.
• Relevance and business value — whether the searcher is a plausible customer. A term can score well on every other axis and still be worthless if the traffic it brings never converts.

Example of Keyword Research

The clearest documented starting point for keyword research is Google Keyword Planner, a free tool inside Google Ads. Its behavior is publicly specified, which makes it a reliable worked example of how the discovery-and-enrichment loop actually functions.

You begin in one of two documented modes: Discover new keywords, where you enter seed terms or a URL and Keyword Planner returns related phrases, or Get search volume and forecasts, where you paste an existing list. For each keyword, the tool reports average monthly searches — and Google states this figure is, by default, “averaged over a 12-month period” and covers the keyword and its close variants, not just the exact string you entered. That single detail changes how you read the data: two different phrasings of the same idea can report the identical bucketed number because Google is grouping them.

The volume itself arrives as a range, not a precise count, for accounts without sufficient ad spend — Google’s documentation describes buckets like 1K–10K rather than a figure such as 6,720. Alongside volume, Keyword Planner reports competition, labeled low, medium, or high, which Google defines as how many advertisers bid on the keyword relative to all keywords. Critically, that competition column measures ad-auction demand, not organic ranking difficulty — a distinction people routinely miss, and the reason serious research pulls organic difficulty from professional SEO databases rather than reading Keyword Planner’s competition column as if it were a ranking forecast.

Put those documented facts together and the prioritization logic falls out. Suppose your seeds surface two candidates: a broad head term reporting 10K–100K average monthly searches with high competition, and a specific long-tail phrase reporting 100–1K with low competition and unmistakable buying intent. Volume-first thinking picks the head term. Research picks the long-tail one, because the documented signals — reachable competition plus transactional intent — predict it will actually rank and actually convert. The head term goes on a someday list, not the roadmap.

Frequently Asked Questions

The Bottom Line

Keyword research turns a vague sense of what a market wants into an evidence-backed shortlist of the exact phrases worth pursuing. It is less about harvesting long lists and more about scoring each candidate on demand, competitiveness, and buyer intent, then assigning the survivors to the pages best placed to satisfy them. Done well, it decides what you write before you write a word.