Two searches can use quite different words and still deserve the same page. Two others may look similar but require completely different answers. Keyword research is how you identify those distinctions, investigate demand and competition, and decide which pages your website should create or improve.
The result should be more useful than a spreadsheet full of phrases. An editor needs to know the intended reader, the question worth answering, the appropriate page type, the evidence for choosing it, and whether an existing URL already does the job. Without those decisions, even a convincing keyword report leaves the difficult work untouched.
A Small Example: Which Searches Actually Need a Page?
Imagine a website that publishes practical advice on caring for houseplants but does not sell them. Its editor starts with the broad topic monstera care. That is a seed topic: a useful starting point for discovering searches, not a command to publish a page with those exact words in its title.
Several possible queries emerge. The following choices are illustrative editorial hypotheses, not measured search volumes or verified live search-result findings.
| Candidate search | Reader task | Initial content decision |
|---|---|---|
| why are monstera leaves yellow | Diagnose a plant problem | Check whether an existing troubleshooting guide can answer it; update that page or create one if needed. |
| monstera yellow leaves | Potentially the same diagnosis | Review the results before treating it as a separate topic. Different wording alone does not justify another URL. |
| how to prune a monstera | Follow a specific procedure | Consider a distinct practical guide if the site can provide accurate instructions and useful visuals. |
| buy monstera plant online | Find a retailer and purchase | Defer or ignore. This publishing site has no product page that can fulfil the task. |
Nothing in this example requires a traffic forecast. It requires knowing what the site is qualified to offer and checking whether two queries lead to the same useful outcome. Only then is it worth deciding how much search demand each opportunity may have.
For any real project, define the audience, target country and language, business or editorial purpose, and the content the team can maintain. Customer questions, product categories, service descriptions and gaps in existing articles provide starting points. The right list begins with the site’s actual remit, not with a tool’s largest suggested keyword.
Collect Evidence Without Mixing Up What the Numbers Mean
Start with a manageable set of candidate phrases and related questions. Existing Search Console queries, audience language, Google search suggestions, competitor pages and keyword platforms can all contribute. They are not interchangeable sources, and none is a complete view of what people want.
| Research source | What it helps establish | What it cannot establish alone |
|---|---|---|
| Google Search Console | Queries, impressions, clicks, CTR and average position associated with your site’s Google Search visibility. | Total market demand or a complete list of every query. Some query data is withheld or limited in reports. |
| Google Keyword Planner | Keyword ideas and historical search estimates, using the selected market, time period and Google Ads settings. | Organic ranking difficulty or the visits an SEO page will receive. Its competition figure concerns advertisers. |
| Google Trends | Relative interest and how it changes by time and geography, using normalised values. | Absolute monthly searches or a precise measure of public opinion. |
| Google Autocomplete | Possible query wording and predictions influenced by real searches, language, location and other context. | A reliable ranking of the most-searched terms or proof that a suggestion deserves a page. |
| Customer questions and site search | The language people use when asking for help, comparing options or looking for information. | Independent evidence of Google search volume. |
| SEO and competitor research tools | Candidate topics, competing pages and proprietary demand or difficulty estimates. | A single authoritative score of how Google will rank your page. |
The settings matter as much as the number. Do not compare a UK estimate from one period with a global estimate from another and call the larger one the better opportunity. Record the source, geography, language where relevant, date range and date checked. This is especially important when a project covers more than one market.
Keyword Planner is built for advertising research. Google’s current help documentation says access to basic keyword-idea features requires completion of account setup, including billing information. It should not be described as a tool everyone can use without an account requirement. For keyword discovery, a smaller combination of sources you understand is preferable to an elaborate collection of incompatible figures.
Let the Search Results Challenge Your Initial Assumption
A candidate phrase may look attractive in a dashboard but point towards the wrong kind of page. Search the shortlisted term in the intended market and inspect the results that appear. What are the prominent page types: informational guides, shopping pages, comparisons, videos, local listings or something else? Notice what problem the titles and visible results appear to solve.
Suppose the houseplant site hopes to publish an illustrated pruning guide. If the results predominantly show practical demonstrations, images or video may be necessary to make that guide genuinely useful. If a phrase instead returns sellers and product listings, a purely informational article may be a poor answer. These are conditional examples, not claims about the present results for any of the sample queries.
Search results can also be mixed. That may mean a broad term covers several tasks rather than one neat intent category. Don’t force everything into a single label. Choose a task your page can fulfil, then examine whether closely related queries share that task. The separate MOCOBIN guide to understanding search intent covers the detailed diagnosis; here, the purpose is to make a publishing decision.
Record what you observed and when you observed it. SERP composition can differ by place, device and time, so a screenshot or note attached to an editorial brief is more useful than treating today’s interpretation as permanent.
Prioritise Relevance, Then Test the Opportunity
Search volume is useful only after the proposed page makes sense. It estimates the scale of searches in a defined dataset and setting, not the number of readers your page will gain. Google Trends can show whether interest is seasonal or changing, but its 0-to-100 values are not raw monthly search totals.
Competition needs a similar distinction. Keyword difficulty is a metric devised by SEO platforms, usually drawing on information about existing ranking pages and links. It is not a Google score. A low number does not compensate for the wrong page type; a high one does not prove that a relevant topic should be abandoned.
For a workable priority, ask whether the audience needs the answer, whether your site can supply it credibly, whether the live results indicate a suitable page format, and whether the expected editorial or commercial value warrants the work. Then compare demand, competing pages, required evidence, production effort and likely maintenance. A topic may be strategically important despite modest reported volume. Another may be popular and entirely beside the point.
Be careful with apparent shortcuts. A long-tail phrase may express a narrower need, but length alone does not make it easy to rank for. Branded and non-branded queries also lead to different decisions: an established brand’s navigational demand should not automatically be counted as a new customer-acquisition opportunity. These distinctions matter to the choice, while their detailed analysis belongs in the specialist guides.
Decide Which URL Should Own the Search Need
Grouping happens after you have examined the reader’s task, not simply after sorting phrases that contain the same words. If two queries require substantially the same answer for the same audience, one well-planned page may serve both. If they lead to different decisions, a distinct page can be justified.
Return to the houseplant example. The two yellow-leaf queries might belong on one diagnosis page, depending on the search results. Pruning is a different task, while buying a plant falls outside the site’s offering. A separate page for every minor variation would increase editorial work without necessarily helping anyone.
Check the existing site before assigning a URL. A relevant page may need clearer coverage or a better title, not a companion article. If two current pages already provide the same answer, review them for consolidation. A missing topic can justify new content only when the site can add a distinctive and maintainable answer.
Keyword mapping turns these findings into URL ownership and helps prevent competing briefs. The broader research workflow should establish the decision, while a detailed mapping guide can explain the spreadsheet, clusters and implementation rules. Once pages have different purposes, an internal linking strategy can connect them naturally. Internal links cannot make two essentially identical pages serve different readers.
Hand the Editor a Decision, Not a Keyword Export
Before research becomes a writing assignment, give each approved topic a brief containing five essentials:
- Reader and market: who is searching, where, and what they need to achieve.
- Target task and page type: the question to answer and the format supported by the observed results.
- Query group and evidence: useful primary and supporting phrasing, the data source, market settings and review date.
- Existing URL and action: create, update, merge, or defer or ignore, with a reason rather than a tool score alone.
- Work and verification: what information or visual evidence is needed, who will maintain it, and what would later justify another review.
For the illustrative yellow-leaf query group, the brief would first ask whether an existing diagnosis page already answers the question. If it does, the assignment is to check that answer and improve genuine omissions. If it does not, a new troubleshooting page may be justified after the live results and editorial expertise have been reviewed. No traffic figure needs inventing to make the instruction usable.
After publication, use Search Console to see which queries and pages gain impressions or clicks, with the usual reporting limitations. A commercial website may also need enquiries, purchases or other business measures; an educational page may be assessed by the usefulness of its answer and the behaviour of the audience it attracts. A keyword estimate was evidence for a decision, not a promise of rankings.
If a page is already doing the intended job, leave it in place. If new evidence shows a different intent, revisit the brief. The value of keyword research is that it lets a team make and revisit those choices deliberately, rather than producing more pages because another phrase has appeared in a tool.
