How to Do Keyword Research: A Practical Process
By The seo.bike team, SEO & GEO practitioners · Last reviewed July 8, 2026
Start with a handful of seed terms describing what you offer, expand them with autosuggest and related searches, then classify each keyword by search intent. Score candidates on volume, difficulty, and business value together, group them into topic clusters, and map exactly one intent to one page to avoid competing with yourself.
Start with seed terms
Seed terms are the five to fifteen plain-language phrases a customer would use to describe the problem you solve, before any tool gets involved. Write them the way a person talks, not the way your industry does: a customer searches "leaky faucet fix," not "potable water fixture remediation." Pull them from your product names, the questions your support inbox repeats, and the words prospects use on sales calls.
A good seed list mixes broad category terms ("running shoes") with problem statements ("shoes for flat feet") and outcomes ("stop knee pain running"). Each seed is a doorway; the expansion step walks through it. If your seeds are all one narrow angle, your whole research inherits that blind spot, so deliberately include a competitor's angle and a beginner's angle.
Don't over-invest here. Ten honest seeds you can expand beat a polished list of fifty. The tools do the multiplication in the next step.
Expand with autosuggest and related searches
Type each seed into the search box and read what autocomplete offers. Those suggestions are real, high-frequency queries ordered roughly by popularity, and they cost nothing. Append a letter ("running shoes a," "running shoes b") or a question word ("how," "why," "best," "vs") to fan out dozens more.
Run the search and mine two more spots: the "People also ask" boxes, which reveal the sub-questions behind a topic, and the "related searches" block at the bottom of the page, which shows adjacent phrasings and the direction searchers wander next. Both are free windows into demand.
Free keyword tools and the autosuggest scrapers built into many SEO suites will batch this for you and attach rough metrics. Use them to widen the net, but the raw search box remains the fastest way to sanity-check that a phrase is one real humans actually type.
Read the intent behind each query
Every keyword carries an intent, and matching it matters more than matching the words. The four buckets: informational ("how to clean suede," the searcher wants to learn), navigational ("nike store," they want a specific site), commercial investigation ("best trail runners 2026," they are comparing before buying), and transactional ("buy hoka speedgoat," they are ready to act).
The fastest way to read intent is to search the term and look at what already ranks. If page one is all listicles, the query wants a comparison, and a product page will not rank no matter how optimized. If page one is product pages, an article will struggle. Google has already told you what it thinks the intent is; argue with it and you lose.
A single keyword can even carry mixed intent, and the results page will show a blend. When it does, note it, because it changes which page type you build and how you structure it.
Score volume, difficulty, and value together
Three numbers decide whether a keyword is worth chasing, and using any one alone leads you astray. Search volume estimates monthly demand but is directional at best; treat the tool's number as an order of magnitude, not a fact. Difficulty estimates how strong the incumbent results are, usually from their link profiles, and tells you the effort to compete. Value is the one no tool prints: how likely this searcher is to become a customer.
The classic mistake is chasing high volume alone. "Marketing" gets enormous volume, brutal difficulty, and near-zero conversion for most businesses. A low-volume term like "invoice software for freelance plumbers" may convert ten times better and rank in a month. Long-tail, lower-volume, higher-intent terms are where a smaller site should start.
A simple filter: prioritize keywords where value is high and difficulty is realistic for your site's current authority, even if volume is modest. Bank early wins on attainable terms, use the traffic and links they earn to raise your authority, then revisit the harder, higher-volume heads later.
Group keywords into topic clusters
Once you have a few hundred candidates, they naturally collapse into clusters of near-synonyms and sub-questions orbiting one topic. "How to clean suede shoes," "remove stains from suede," and "suede shoe care" are not three articles; they are one article's worth of intent phrased three ways. Grouping prevents you from writing five thin pages where one thorough page belongs.
Build the clusters by intent, not by string similarity. Two phrases can share words but split on intent ("apple pie recipe" vs "apple pie near me"), and those genuinely are two pages. Sort your list, tag each row with its intent, and the clusters draw themselves.
Each cluster becomes a candidate page or a pillar-plus-supporting-pages structure: one broad hub page for the head term, focused pages for the distinct sub-intents beneath it, all interlinked. This is also how you plan internal links before you write a word.
Map one intent to one page
The final and most-skipped step: assign each cluster to exactly one URL, and never let two pages target the same intent. When two of your pages chase the same query, they compete with each other, split their link equity, and confuse the search engine about which to rank. That self-inflicted problem is keyword cannibalization, and it quietly caps the ceiling of both pages.
Keep a simple map: a spreadsheet with one row per target intent and the single URL that owns it. Before publishing anything new, check the map. If an existing page already owns that intent, strengthen it instead of building a rival; if the new angle is genuinely distinct, give it its own row.
Revisit the map quarterly. Search demand shifts, you will spot two old pages that drifted onto the same intent, and merging them (with a redirect from the weaker URL to the stronger) usually lifts the survivor. The discipline of one-intent-one-page is what turns a keyword list into an organized, non-competing site.
Key terms
Related guides
- How to Write Title Tags and Meta Descriptions
- The Types of SEO: On-Page, Off-Page and Technical
- On-Page SEO: How to Optimize a Page That Ranks
- What Is SEO? A Plain-English Guide for Beginners
Frequently asked questions
No. The search box's autocomplete, the "People also ask" boxes, and the "related searches" block are free and reveal real demand. Free keyword tools add rough volume and difficulty estimates. Paid suites mainly save time at scale and sharpen the difficulty scores; a small site can build a solid keyword map with free tools and manual searching alone.
Search the term and study page one. The page types Google already ranks, whether listicles, product pages, how-to articles, or local packs, reveal the intent it has assigned. If comparison articles dominate, the query is commercial-investigation and a product page will not rank. Matching your page type to what already ranks is more reliable than guessing from the words alone.
A newer or lower-authority site should start with long-tail, lower-volume, higher-intent terms. They are easier to rank for, convert better because they are specific, and the traffic and links they earn raise your authority. Use those early wins as a foundation, then pursue the harder, higher-volume head terms once your site can realistically compete for them.
Cannibalization happens when two or more of your pages target the same search intent, so they compete with each other, split link equity, and confuse the engine about which to rank. Avoid it by keeping a map of one intent per URL, checking it before publishing, and merging or redirecting duplicate pages into a single stronger one when you find them.
Put this into practice
Try the free SEO tools, or let the managed service do the work for you — every change checked by a safety linter before it ships.