Skip to main content

Guides

A keyword research method should end with an owned page and a next check

A repeatable keyword research method that connects competitor terms, demand, difficulty, intent, SERP evidence and a page decision.

By Kai · Reviewed

The right keyword research method depends on the evidence you already have. A new product can start from customer language; a team entering an established category can learn from competitor pages; a site with search history should inspect its own queries. The work converges on the same useful outcome: a page with a clear reader, task and contribution.

On this page
  1. When you have customers but little search data
  2. When competitors already serve the category
  3. When your site already has search impressions
  4. One example, three possible research decisions
  5. Keep the first research round small enough to publish from
  6. When the methods disagree, choose the evidence closest to the task

When you have customers but little search data

Use the words people use while trying to finish a task. A support request saying “I need to compare last month’s traffic with this month, but the current month has not ended” contains the task, the constraint and the likely mistake. An internal label such as “advanced analytics workflow” loses those details.

Collect a small set of such questions from conversations you are permitted to use, remove personal details and group them by the desired output. Then inspect search results for representative wording. A recurring customer problem may need documentation, a product change or a public guide; the keyword check helps you decide how other people seek the same answer.

If the keyword tool has no volume value, keep the customer evidence and its origin. You can still improve an existing page that fails to answer the question. A standalone acquisition page needs a clearer case: a distinct task, enough substance and a plausible way for the intended reader to find it.

When competitors already serve the category

Choose a few relevant domains and inspect ranking terms beside their landing pages. The page is the important unit: it shows whether the query is served by a template, a product feature, an explanation or a comparison. Keep navigational demand separate from problems your own product can address.

For each candidate topic, try the receiving page. A template you can use immediately teaches a different lesson from a long article that eventually asks for a sales call. Write down the part that helps the reader and the part you could improve with material you actually have.

This method is strongest for discovering language and page formats. It becomes weak when the competitor’s full topic coverage is treated as your publishing plan. Your buyers, features and distribution may differ. A rejected candidate with a clear reason is a useful research result.

When your site already has search impressions

Open Search Console and inspect relevant query groups by page over comparable periods. Start with the question you can act on: which pages attract the intended queries, which relevant queries land on an unsuitable page, and which useful pages have little visibility? Keep country and device filters consistent while making the comparison.

A page with relevant impressions and few clicks deserves an inspection of the displayed result, position and competing formats. A page with clicks but little task completion deserves an inspection of the actual resource. These observations lead to different edits, so avoid treating every weak result as a request for another article.

Use complete windows where possible, and inspect seasonality when it could matter. A very small sample can change sharply after a handful of visits. Keep the counts with the rates and write a provisional interpretation until more evidence arrives, rather than repeatedly changing the page’s subject.

An aggregate CTR can change because the query mix changes. In a teaching example, one query group keeps 500 impressions and 50 clicks, while a new group adds 9,000 impressions and 90 clicks. The combined CTR is about 1.47%, compared with 10% for the original group alone, although that original group has not weakened. Inspect the groups before rewriting a title solely because the overall rate fell.

One example, three possible research decisions

Return to the incomplete-month question. If several customers ask it and the product already has a complete-month selector, the most useful work may be a help article with a screenshot and a worked comparison. If competitor search pages reveal demand for traffic-growth explanations, a public guide can show how month length and baselines affect the result.

If visitors already reach a guide on your site but cannot find that answer, edit the relevant section and link to it from the product. The same underlying question has produced documentation, acquisition content or a local improvement depending on the evidence. Keyword research is valuable when it selects among those outcomes.

Keep a record with the exact query, its evidence source, intended reader, proposed URL and the contribution to produce. Add market, metric date and numerical estimates when you use them. A teammate should be able to open the record and understand both the choice and what still needs to be created.

Keep the first research round small enough to publish from

Pick a small group of decisions you can complete with the resources available now. Use Vaneform’s Site keywords for observed competitor language and its keyword brief for the supported market’s demand and results. Keep customer questions and owner analytics in the same reasoning process, even when they live in different tools.

Review the published page after it has relevant observations. Ask whether the reader found the expected resource and whether the page’s evidence is still current. Expand the research when a new task emerges, not merely because the spreadsheet could hold more rows.

When the methods disagree, choose the evidence closest to the task

A customer may ask for a very specific feature while the keyword tool shows a much broader high-volume topic. A competitor may rank for that broad topic through a free resource that your product does not offer. These are different observations, and they do not have to produce the same publishing decision. Begin with the person and the job your page can realistically serve.

Use the customer question to identify the constraint, the competitor page to inspect a possible format and the keyword data to understand the measured demand. Then ask what material you can produce. If the answer is a precise help example for existing customers, publish that improvement even while the larger acquisition topic remains under research.

Conversely, a broad query can reveal a useful public resource when you have original data or a functioning tool that completes the task. Confirm what the result pages expect and what you can maintain. The decision should be explained in terms of the resource and audience, with numerical estimates serving as context.

Keep the disagreements in the research note. “Customer need is clear; search demand is unmeasured; a documentation update is ready” is a useful conclusion. “Competitor topic is visible; our required asset is missing; defer the acquisition page” is another. Each tells the team what to do without pretending all evidence has the same strength.

Revisit the decision when its missing evidence becomes available or the product changes. A useful keyword process is not a one-time export; it connects new information to a page that someone owns. Preserve the previous reasoning so later changes reflect learning rather than a fresh guess about which phrase might rank.

Research a website or a search opportunity · /guides/keyword-research-methods