Keyword Research Rules

A keyword list is not a content strategy until every term is tagged with the intent behind it and assigned to exactly one page — a pile of high-volume keywords with no intent or ownership attached produces content that ranks for the wrong reason or not at all.

Search intent classification

• Classify every keyword into one of four intents before doing anything else with it: informational ("how does X work"), navigational ("brand name login"), commercial investigation ("best X for Y," "X vs Y"), or transactional ("buy X," "X pricing"). The intent determines what kind of page can rank, not the keyword's volume.
• Check the current top-ranking pages for a keyword as the real signal of intent, not the keyword's surface wording — a phrase that looks informational can be dominated by product pages (and vice versa), and Google's actual results are the ground truth, not a guess from the query text.
• Never write a single page trying to satisfy two conflicting intents (e.g. a listicle trying to also convert like a product page) — pick the intent the keyword actually calls for and build the page format that intent expects.

Long-tail vs. head terms

• Head terms (1–2 words, high volume, high competition) rarely convert directly and are usually unrealistic ranking targets for a new or mid-authority site — treat them as category indicators, not primary targets.
• Long-tail terms (3+ words, lower individual volume, much lower competition, closer to a specific need) convert at a higher rate per visitor and are realistic targets much earlier — a strategy built on long-tail coverage compounds into head-term relevance over time, not the other way around.
• Don't dismiss a long-tail keyword for "low volume" without checking search intent and buyer proximity — ten highly specific, high-intent long-tail terms often outperform one broad head term for actual business outcomes.

Difficulty vs. relevance tradeoff

• Weigh keyword difficulty against topical relevance to the site's existing authority, not against difficulty alone — a moderately hard keyword directly on-topic for an established site is a better target than an easy keyword with no relevance to anything else on the site.
• Treat a difficulty score as a rough proxy for competition, not a hard gate — a keyword with weak or outdated top-ranking content can be winnable despite a high score, and that requires actually looking at the current results, not just the number.
• Sequence targets by realistic winnability given current site authority: build topical relevance and internal links from easier wins first, then use that accumulated authority to go after harder, higher-value terms later rather than attacking them cold.

Clustering into topics

• Group keywords that share the same search intent and could reasonably be satisfied by the same single page into one cluster — don't create a separate page per keyword when several keywords are really the same underlying question phrased differently.
• Name each cluster by its core intent/topic, then designate one primary keyword (the clearest, highest-value phrasing) as the page's main target and the rest as supporting terms the page should also naturally cover.
• Build clusters around a hub-and-spoke structure when a topic is broad enough: one comprehensive hub page plus several narrower spoke pages on subtopics, internally linked — this is how a site earns topical authority instead of a scatter of disconnected pages.

Mapping keywords to pages

• Assign every keyword cluster to exactly one URL before writing anything — an explicit keyword-to-page map, maintained as a living document, is what prevents both duplicate content and content gaps as the site grows.
• Match the page type to the intent already classified above: informational clusters get articles/guides, commercial-investigation clusters get comparison pages, transactional clusters get product/pricing pages.
• Revisit the map when adding new content ideas — check the map first for whether a cluster already owns the topic before creating a new page, rather than starting from a blank slate every time.

Avoiding cannibalization

• Before publishing a new page, search the existing map (and the live site) for any other page already targeting the same primary keyword or a keyword in the same cluster — two pages competing for the same query split ranking signals and usually cause both to underperform.
• When two existing pages are found to compete for the same keyword, don't just add more content to both — consolidate into the stronger page and 301-redirect the weaker one, or clearly differentiate the two by intent/subtopic so they stop overlapping.
• Watch for cannibalization created silently over time — different writers targeting near-duplicate long-tail variants of the same head term is the most common cause, and the keyword-to-page map is the tool that catches it before publish, not after.