What Is Keyword Research? Finding Terms Worth Ranking For

Jul 31, 2026 · SEO Glossary

The most common content strategy failure I see isn't bad writing — it's good writing aimed at nothing. A client came to me with forty blog posts, professionally written, genuinely useful… and collectively earning about 30 organic visits a month. Not one post targeted a query anyone actually typed. Their competitor's scrappier blog, half the quality, pulled thousands of visits — because every post answered a real search. That gap is what keyword research exists to close.

What is keyword research? It's the process of discovering what your potential customers type into search engines, measuring how many of them do it and how hard those terms are to win, and then deciding which queries each page of your site will target. It's market research wearing an SEO costume — done before content is written, because it determines what's worth writing at all.

What is keyword research trying to balance?

Every keyword decision juggles three variables, and the tension between them is the whole craft:

• Demand — monthly search volume. But raw volume misleads: "insurance" has enormous volume and sends you nothing you can use, while "freelancer liability insurance germany cost" might show 90 searches a month and convert at 8%. Treat volume as one input, never the score.
• Winnability — who currently holds page one. If a query's results are all major brands with thousands of referring domains, a new site targeting it is donating content to the void. Difficulty scores in tools approximate this; looking at the actual results yourself beats the score every time.
• Value — what a visitor from this query is worth to you. "how to fix a leaking tap" brings a plumber DIYers; "emergency plumber near me" brings customers. Same topic, wildly different business value. This is where search intent enters: informational, commercial, and transactional queries deserve different pages and different priorities.

The strategic sweet spot for most small sites is deliberately down-market: specific, lower-volume queries with clear intent and beatable competition. Ten long-tail keywords at 100 searches each, where you can actually rank, out-earn one 10,000-volume term where you'll sit on page four indefinitely.

Where the keywords come from

A workable research session, with or without paid tools:

1. Seed from the business, not a tool. List what you sell, the problems it solves, and the exact phrases customers use in emails and sales calls — that last source is gold most agencies never mine. Clients say "our website looks broken on phones"; they don't say "responsive design services."
2. Expand with free data. Google's autocomplete, the "People also ask" boxes, and related searches are Google telling you verbatim what people type. Search Console's Performance report shows queries you already half-rank for — the cheapest wins in all of SEO, since improving an existing position 12 is far faster than ranking from nothing.
3. Study the winners. For each candidate query, look at what ranks: what format (guides? product pages? videos?), how deep, how fresh. The results page tells you what Google believes the query means — and whether your planned page matches or fights that interpretation.
4. Layer in tool metrics if you have them. Volume estimates and difficulty scores from any mainstream tool are fine for relative comparison. Their absolute numbers are estimates with wide error bars; treat "500/month" as "meaningfully more than 50, meaningfully less than 5,000."

From list to map: the step everyone skips

Research that ends in a spreadsheet of 300 keywords has produced nothing yet. The output that matters is a keyword-to-page map: each priority query assigned to exactly one URL — existing or planned — with its intent noted. This mapping is what prevents three future problems at once: content written about nothing (my opening client), multiple pages drifting onto the same term (keyword cannibalization is almost always a mapping failure), and topics covered at the wrong depth — a 3,000-word guide answering a query that wanted a price list.

Group mapped keywords by theme and you've also discovered your site's structure: clusters of related queries become hubs and supporting pages, which is how keyword research feeds directly into information architecture rather than just a publishing calendar.

Judgment calls the tools won't make

A few opinionated rules from years of doing this for clients. Don't chase queries you can't serve — ranking for a question your business can't credibly answer earns bounces, not customers. Revisit the research quarterly, lightly: search language shifts, new questions emerge, and Search Console keeps revealing queries you never planned for. Weight your own data over any tool's — a query that already sends you converting traffic at position 8 outranks any tool's "opportunity score." And remember that targeting is only half the delivery: the page has to be crawlable, indexable, and fast for the targeting to matter, which is why keyword work and a technical site audit belong in the same project plan — I've watched beautifully targeted content underperform for a quarter because the section it lived in was noindexed. Research decides what to say; the site's health decides whether Google hears it.