Keyword research is the compass for everything else in search, and doing solid keyword research follows one repeatable pipeline: seeds, expansion, qualification, mapping and maintenance. Rush the front half and you spend months writing for queries nobody types or nobody buys from. This guide walks the entire process end to end — from a blank spreadsheet to a finished keyword map your content plan can hang on.
Expensive software isn't required to start; free sources comfortably cover a first pass. What matters is discipline about sequence: gather wide, filter hard, assign each survivor precisely, then keep the map breathing as real data arrives.
Quick Answer: Build seed terms from your products and customers' problems, expand them into hundreds of candidates using tools plus live SERPs, qualify each candidate on demand, difficulty and intent, map one primary keyword per page, and refresh the map quarterly using Search Console data.
Step 1: Build Seed Lists From What You Already Know
Seeds are broad themes, not targets. Sources hiding in plain sight:
- Your own product and service names, including industry jargon and plain-English versions.
- The problems your offer solves, phrased the way customers describe them in emails, calls and reviews.
- Competitor homepage language — their headings encode research they already did.
Aim for ten to twenty seeds spread across several buckets. Too few seeds narrow the funnel before it opens; too many create cleanup work later.
Step 2: Expand Every Seed Into Raw Candidates
Feed each seed through multiple expansion sources and collect without judging:
- Google Keyword Planner for volume-scaled ideas straight from Google's own ad data.
- Autocomplete and Related Searches for authentic user phrasing.
- People Also Ask chains for question-format entries.
- Competitor gap reports in tools like Semrush or Ahrefs to lift terms rivals already rank for.
Several hundred raw candidates is a healthy haul for a typical small-to-mid site. Volume now, verdicts later.
Step 3: Qualify Each Candidate on Three Filters
- Demand: is anyone searching? Exact volume numbers matter less than directionality; true zero-demand terms drop unless they're clearly converting brand variants.
- Difficulty: skim the live SERP. Forums and shallow pages signal openings; wall-to-wall major brands signal retreat.
- Intent: what does the searcher want, and can your site credibly provide exactly that?
Tier the survivors — A for act-now targets, B for later, C for watchlist. The tiers become your publishing queue.
Step 4: Map Survivors Onto Pages
Mapping prevents the classic self-inflicted wound of three pages competing for one query. Build a sheet with columns for keyword, intent, tier, target URL and status. Group near-synonyms sharing intent onto a single page; split distinct intents onto distinct pages. The exercise also exposes gaps needing new content versus existing pages that merely need strengthening.
Step 5: Treat the Map as a Living Document
Research decays. Refresh quarterly, and monthly pull the queries Search Console reveals that you never targeted — real searcher language outranks any tool's imagination. Strike dead terms, promote surprise performers, log seasonal patterns, and version the file so decisions stay traceable.
| Stage | Output | Helpful Tools |
|---|---|---|
| Seeds | 10–20 starter themes | Customer language, competitor homepages |
| Expansion | 300+ raw candidates | Keyword Planner, autocomplete, gap reports |
| Qualification | Tiered A/B/C shortlist | Live SERPs, difficulty metrics |
| Mapping | Keyword-to-URL assignment sheet | Plain spreadsheet |
| Maintenance | Updated map each quarter | Search Console query data |
Key Takeaways
- Seeds come from your products, your customers' words and competitors' headings.
- Expand widely first; qualify demand, difficulty and intent second.
- Map exactly one primary keyword per page to avoid cannibalizing yourself.
- Real Search Console queries beat tool estimates for revealing actual language.
- Refresh the whole map quarterly — research has a shelf life.