Keyword Research: The Complete Step-by-Step Guide

Keyword research is the decision layer of SEO: choosing which searches you'll compete for before spending a word of content or a single link on them. Done well, everything downstream gets easier — pages match real demand, clusters assemble themselves, and effort lands where winning is possible. Done badly, whole sites are built for queries nobody types or nobody can win. This pillar is the complete step-by-step process, from seed to prioritised plan.

Step 1: Start from seeds, not tools

Tools expand; they don't originate. The seed list comes from the business: what you sell, the problems it solves, the words customers actually use (support tickets, sales calls and community threads are gold — customers rarely speak in your internal vocabulary). Ten to twenty seed terms per topic area is plenty; the expansion machinery below turns each seed into hundreds of candidates.

Step 2: Expand the universe

Feed every seed through the expanders and collect everything:

• Keyword tools — a paid suite (Ahrefs/Semrush keyword explorers) or the free stack plus Keyword Planner. Pull the matching-terms and questions reports per seed.
• The SERP itself — autocomplete, People-Also-Ask, related searches: Google telling you what its users type.
• Competitors — export what ranking rivals already rank for; the full method is competitor keyword analysis, and it routinely doubles the universe with proven-demand terms.
• Your own data — Search Console queries where you already get impressions are pre-validated demand with a head start.

Expect thousands of raw candidates per topic. That's correct — the next three steps are filters.

Step 3: Understand each keyword's three properties

Every candidate gets judged on the same triad:

Demand (volume, honestly read)

Monthly search volume is an estimate with wide error bars — treat it as order-of-magnitude (10s vs 100s vs 1,000s), not arithmetic. And don't dismiss the small numbers: the long tail — thousands of low-volume specific queries — collectively outweighs the head terms and converts better, because specificity is intent.

Intent (what the searcher is actually doing)

The decisive property, and the most skipped: is the query informational, navigational, commercial-investigation or transactional — and what page shape does its SERP reward? The intent guide covers the full taxonomy; the operating rule is that intent is read from the ranking pages, never assumed from the words.

Difficulty (can you actually win it)

Tool KD scores are a first-pass sort; the real judgement is the SERP audit — who ranks, with what authority, what content, and what link entry fee. The difficulty guide turns this into a repeatable check.

Step 4: Group into topics, map to pages

Individual keywords aren't pages — intents are. Group candidates that share an intent (their SERPs largely overlap) into one target; separate intents get separate pages. Then assign every group a home: one keyword-group, one URL, recorded — the keyword mapping discipline that prevents two pages half-covering one query and cannibalising each other. At topic scale, the groups arrange naturally into pillar-and-cluster structure: the head term is the pillar, the subtopic groups are the clusters — keyword research and site architecture are one exercise viewed from two ends.

Step 5: Prioritise like an investor

Score each mapped group on value (what a ranking is worth to the business), winnability (difficulty vs your authority), and effort (what the content costs to do properly). The portfolio that works for most sites: quick wins first (low difficulty × decent value — often long-tails and evergreen topics the big players ignore), strategic bets second (the head terms your link programme will spend months earning), and defensive coverage third (queries where absence hurts the brand). Publish in cluster runs, not scattershot — coverage compounds, as this site's own calendar demonstrates in public.

Step 6: Revisit on a schedule

Keyword research is a cycle: quarterly, feed Search Console's actual queries back in (reality outperforms every tool), harvest new PAA/autocomplete shifts, and re-audit difficulty on your strategic bets as your authority grows — keywords unwinnable at DR 20 open up at DR 40. The map is a living document; sites that research once in 2019 are competing for a web that no longer exists.

Frequently asked questions

How many keywords should one page target?

One intent-group per page — which typically means one primary keyword and its dozens of natural variants, ranked via the semantic coverage a complete answer produces. Counting "keywords per page" is the wrong unit; counting intents is right.

Do I need paid tools for keyword research?

The free stack covers seeds, expansion and intent-reading honestly; paid suites add volume/difficulty data at scale and competitor exports. Start free, pay when list size makes manual work the bottleneck.

What comes after the keyword plan?

The on-page execution per page, cluster-linked architecture, and the authority budget for whatever the SERPs priced in — which is where our website database and team plug into the plan.

The keyword research series

1. Long-Tail Keywords: How to Find and Use Them
2. Search Intent: The 4 Types and How to Match Them
3. Keyword Difficulty: How to Judge If You Can Rank
4. Keyword Mapping: One Keyword, One Page
5. Keyword Cannibalization: Find and Fix It
6. Best Free Keyword Research Tools
7. Google Keyword Planner for SEO
8. Competitor Keyword Analysis
9. Evergreen Keywords: Topics That Never Die