Most keyword research goes wrong in the same way: a large list is exported, sorted by search volume, and handed to a writer. The result is content that targets words rather than people, and it rarely ranks.
Intent comes before volume
Every search sits somewhere on a spectrum from learning to buying. A page can serve one point on that spectrum well; it cannot serve all of them at once.
- Informational: the searcher wants to understand something. Guides, explainers and comparisons win here.
- Navigational: the searcher wants a specific brand or page. You rarely compete for these unless it is your own brand.
- Commercial: the searcher is evaluating options. Comparison pages, case studies and detailed service pages work.
- Transactional: the searcher is ready to act. Product, pricing and contact pages.
Let the results page tell you
The fastest way to confirm intent is to run the search and look at what already ranks. If the first page is full of tutorials, a product page will not break in, no matter how well optimised. Treat the current results as the clearest available statement of what search engines believe that query means.
Build a map, not a list
A keyword map assigns one primary query and a small cluster of closely related variants to a single page. This prevents two of your own pages competing for the same term, and it makes it obvious when you need a new page rather than another paragraph.
- One page, one primary intent.
- Group variants that would be satisfied by the same answer onto that page.
- When a variant needs a genuinely different answer, it earns its own page.
- Record the target page for every query so writers and developers work from the same document.
Where to find candidate terms
Keyword tools are a starting point, not the source of truth. Some of the most valuable queries never show meaningful volume in a tool.
- Your own site search logs and Search Console queries, which reflect real demand from real visitors.
- Sales and support conversations, where customers use their own words rather than industry vocabulary.
- Community forums and question sites, which surface the phrasing people use when no marketer is listening.
- Competitor pages that rank consistently, which reveal clusters you may have missed.
Judge difficulty honestly
Difficulty scores are estimates built largely from link metrics. They tell you little about whether you can realistically compete. A better question: do the pages ranking now demonstrate experience, depth and trust that your page can match or exceed? If not, target something more specific and build from there.
Review it regularly
Intent drifts. A query that returned guides last year may return product listings now, and a page written for the old result set will quietly decline. Re-check the results pages for your priority terms a few times a year and adjust rather than waiting for traffic to fall.