Choose SEO keywords by the task you can answer, their relevance to your audience and the evidence you can provide; use search volume as one demand estimate. A high-volume query is not automatically informational, and a low-volume query is not automatically transactional or highly converting.
Start with the market and language you serve. Then examine representative results, check existing pages and choose a useful next action for the reader. This avoids spending a content budget on terms that look large in a tool but describe the wrong product, country or user need.
Define intent before applying volume filters
Informational, navigational, commercial-research and transactional are useful editorial categories, not monthly-volume bands. A query can have mixed intent, and the same wording may mean different things in different contexts. Moz’s search-intent guide explains these categories and the value of studying actual results; a tool’s automatic label still needs judgment.
| Intent | Illustrative query | Useful destination |
|---|---|---|
| Learn or solve a problem | how to measure a flange bolt circle | A practical measurement guide with units, method and limitations |
| Reach a known destination | Example Tools support | The relevant support page for that named business |
| Compare options | pneumatic vs electric actuator for a valve | A comparison with operating conditions and trade-offs |
| Take a specific action | request a quote for actuator model X | The correct model or enquiry page with required inputs |
These are invented query examples, not measured keywords. “Example Tools” and “model X” are placeholders. Confirm what your own buyers mean through sales questions and current search results. See the related search-intent and page-type guide for separating a definition, comparison and purchasing task.
Read the metrics on their own terms
Google’s Keyword Planner documentation defines average monthly searches using the selected period, location and Search Network, including close variants. Its default average covers 12 months and counts searches regardless of language. Record the actual tool settings and avoid interpreting a displayed number as unique people or guaranteed visits.
| Metric | What it helps you consider | What it does not establish |
|---|---|---|
| Average monthly searches | Estimated demand under the tool’s stated settings | Unique buyers, your attainable clicks or a fixed intent |
| Advertising competition and bids | Advertising activity and historical bidding information | Organic keyword difficulty or your profit per sale |
| Keyword Difficulty (KD) | The provider’s estimate of ranking difficulty | A percentage probability of success or a universal cutoff |
| Google Trends interest | Relative, normalized search interest | Absolute monthly searches or a representative customer survey |
| GSC impressions and clicks | What your site recorded under the report’s scope | All market demand or confirmed AI citations |
Ahrefs defines its KD score on a 0–100 scale and advises against making every SEO decision from it alone. Different providers use different methods; keep provider, date and market attached to a score. A score of 35 is not “a 35% chance of failure,” and a lower score does not make an unsuitable query worth targeting.
Google Trends documentation explains normalization to 0–100. Low-volume terms may appear as zero, and autocomplete predictions are filtered suggestions, not a demand table. Rounded volume estimates and overlapping close variants should not be added together as though every row represented different people.
Inspect the search results and your existing content
- Record query, date, language, location and device where known. Separate advertisements, regular results and special search features.
- Read representative pages. Identify the task they answer, source quality, practical detail, relevant product conditions and gaps you can genuinely fill.
- Check your site for a page already serving that intent. Improve it when appropriate; do not publish one near-duplicate article for each synonymous phrase.
- Ask what you can add: a verified procedure, a well-supported comparison, original observations or a clearer way to use reliable documentation. Do not invent testing to make a keyword seem feasible.
Search results describe an observed situation, not a rule that all successful articles must copy the same word count or list format. A specialist page can be valuable with modest estimated demand when it solves an important customer problem.
A prioritization example without fake traffic forecasts
Fictional planning exercise: suppose an industrial supplier sees the following values in a hypothetical tool. They are teaching inputs, not live keyword research. All rows use the same assumed market and period, and the estimates are not summed.
| Query | Hypothetical monthly estimate | Business and evidence check | Editorial decision |
|---|---|---|---|
| industrial automation | 8000 | Broad topic; the supplier cannot answer the full category well | Do not prioritize a generic article merely for volume |
| how to measure flange bolt circle | 300 | Relevant task; a documented method and clear diagrams are available | Consider a practical guide after checking existing coverage |
| actuator X replacement quote | 40 | Specific model; compatibility and quotation process must be confirmed | Improve the relevant model/enquiry page if the product is actually supported |
The choice depends on relevance, available evidence, cost of keeping the page current and the role it plays in the site. It does not imply the 40-search term converts better, that 300 searches will become 300 visits, or that a 6:3:1 publishing ratio is appropriate.
Keep uncertain inputs visible. If you lack actual demand data, label prioritization as an editorial judgment. If a query has value only in a market you cannot serve, volume alone should not override that constraint.
Map the chosen query to a complete page
Use one principal intent, related questions and the relevant entities or terms to plan the title and sections. Answer each important question with substance. Use natural wording, supported examples and useful internal links; do not target a keyword-density percentage or repeat an exact phrase in every image alt attribute.
Titles should accurately promise the answer, as shown in the guide to using keywords naturally in titles. An explanatory article can lead to a related model or service page without forcing a sales pitch into every paragraph. Other relevant pages should link back where the guide answers a genuine next question.
Evaluate what happened after publishing
In the Search Console performance report, compare the intended pages and queries with consistent country, device, search type and time filters. Check query mix as well as impressions, clicks and CTR. No impressions for a new page do not by themselves prove there is no demand.
Evaluate qualified enquiries separately with an agreed definition and duplicate handling. Rankings, clicks, AI citations and sales are different observations; a short engagement time is not a universal content-failure threshold. Retain a dated baseline and do not invent historical results that were never collected.
Guangsuan’s GHA English SEO article service offers a free trial article and develops content around readers’ questions and supporting material. The practical starting point is a useful topic and available evidence, rather than a fixed volume or word-count rule.