Keyword research is the process of discovering, analyzing and prioritizing the search queries that a target audience types into search engines, so that a website can create or optimize pages that match them. In SEO, it is the planning step that connects what people actually search for with what a business can credibly offer.
Good keyword research does more than collect popular phrases. It evaluates each query by search intent, search volume, keyword difficulty and business value, groups related queries into topics, and assigns them to specific pages. The result is a prioritized content and optimization plan rather than a list of words.
How the keyword research process works
Keyword research follows a repeatable sequence of stages, from broad discovery to page-level decisions. Practitioners start with seed keywords that describe the offer, expand them through tools and observation, then filter and cluster the list until each cluster maps to one URL.
The table lists the typical stages and what each one produces.
| Stage | Main activity | Output |
|---|---|---|
| Seed collection | List core products, problems and audience language | 10-50 starting terms |
| Expansion | Use tools, autocomplete, related searches, competitors | Hundreds to thousands of candidates |
| Qualification | Check intent, volume, difficulty, relevance | Filtered shortlist |
| Clustering | Group queries that one page can satisfy | Topic clusters |
| Mapping | Assign clusters to new or existing URLs | Keyword-to-page map |
| Prioritization | Rank by value, effort and chance to rank | Content roadmap |
Sources for expansion include keyword tools, Google Search Console query data, autocomplete suggestions, “People also ask” boxes, forums, customer support logs and sales conversations. Internal data is often the most valuable input because it reflects the vocabulary of real customers rather than industry jargon.
Metrics used to evaluate keywords
Each candidate keyword is judged on a small set of metrics, none of which is decisive alone. Volume shows demand, difficulty estimates competition, and intent determines whether ranking would bring the right visitors at all.
| Metric | What it estimates | Main caveat |
|---|---|---|
| Search volume | Average monthly searches in a market | Tool estimates are rounded and often seasonal |
| Keyword difficulty | Effort needed to reach page one | Proprietary score, not comparable across tools |
| Search intent | Goal behind the query | Mixed intent is common for short queries |
| Cost per click | Commercial value from paid search bids | Reflects advertiser competition, not organic value |
| Click potential | Share of searches that produce organic clicks | SERP features and AI answers reduce it |
Volume figures deserve particular skepticism. Tools model them from clickstream and panel data, so two tools can differ by a factor of two or more for the same query. They are useful for comparing queries relative to each other, not as precise traffic forecasts.
Search intent and keyword types
Search intent is the single most important filter in keyword research, because a page that does not match intent rarely ranks, however well it is optimized. The fastest way to determine intent is to inspect the current SERP: the types of pages Google already ranks reveal what it believes searchers want.
| Intent type | Example query | Content that fits |
|---|---|---|
| Informational | how does solar inverter work | Guide, explainer, tutorial |
| Commercial investigation | best solar inverter for home | Comparison, review, buying guide |
| Transactional | buy 5 kW hybrid inverter | Product or category page |
| Navigational | brand name inverter login | Brand or account page |
| Local | solar installer near me | Location page, business profile |
Keywords are also classified by length and specificity. Head terms of one or two words carry large volume but vague intent and fierce competition, while long-tail keywords of three or more words have low individual volume but clearer intent and higher conversion rates. Collectively, long-tail queries account for a large share of all searches, which makes them the realistic entry point for newer sites.
Keyword research compared with related SEO activities
Keyword research sits upstream of most SEO work and feeds several neighboring practices. It is easily confused with them, so the boundaries are worth stating.
| Activity | Main question | Relation to keyword research |
|---|---|---|
| Keyword research | Which queries should the site target | Defines targets and priorities |
| On-page SEO | How should a page be optimized | Uses the mapped keywords on each URL |
| Content strategy | What should the site publish and why | Uses clusters as a demand signal |
| Competitor analysis | Where do rivals win traffic | Supplies gap keywords and difficulty context |
| Rank tracking | How do pages perform over time | Measures results for chosen keywords |
Modern search engines interpret meaning rather than exact strings, so research now targets topics and entities, not single phrases repeated on a page. One well-structured page can rank for hundreds of semantically related queries, which is why clustering has replaced the old one-keyword-per-page model.
Common mistakes in keyword research
Most weak keyword strategies fail for the same predictable reasons: they chase volume, ignore intent or overlook the competitive reality of the SERP.
- Chasing volume over relevance: high-volume terms that do not match the offer bring traffic that never converts.
- Ignoring the SERP: skipping a manual check misses intent signals, SERP features and dominant domain types.
- Keyword cannibalization: assigning one cluster to several pages makes them compete against each other.
- Treating difficulty scores as facts: they are estimates and should be checked against the authority of ranking pages.
- Researching once: demand, language and AI-generated answers change, so research needs periodic revision.
The idea in brief
Keyword research means finding out which words people type into search engines when they look for something a business offers. It is like listening to customers before writing a shop sign.
Some phrases are searched often but are hard to win. Others are searched less but by people who know exactly what they want. Keyword research helps a site choose the phrases worth pursuing and decide which page should answer each one.
A practical scenario
A regional company that installs heat pumps wants more inquiries from search. Its marketing lead begins with seeds such as “heat pump installation” and “heat pump cost”, then expands them with a keyword tool and twelve months of Search Console data. The list grows to about 1,400 queries.
After filtering for relevance and inspecting the SERPs, the team groups the queries into 35 clusters. The broad head term is too competitive, so it prioritizes long-tail clusters with clear intent, such as heat pump running costs for older houses and installation grants in its region. It maps each cluster to one page, merges two pages that were cannibalizing each other and writes six new guides.
Within about eight months, organic visits rise steadily and, more importantly, quote requests from organic search roughly double, because the new pages match what local homeowners actually ask.