Target keywords are the queries and related expressions a page is intended to answer. Choosing them requires audience demand, a useful answer, an appropriate destination and a realistic view of competing content. Search volume and difficulty help organize the research; a universal monthly-volume range or KD cutoff cannot decide every site's priorities.
Begin with a reader and a task
Start from what the organization can genuinely explain or provide. What does a buyer need to compare? Which information is missing from inquiries? What causes confusion during use? Preserve models, applications, locations and constraints when translating these questions into possible searches.
Do not restrict the list to immediate-purchase terms. Definitions, troubleshooting and preparation guides can serve relevant future customers and useful audience needs. Explain that connection. A trending topic without the knowledge or evidence to answer it is a weak editorial starting point.
Google's SEO Starter Guide recommends considering the different terms beginners and experienced users may search. It also explains that language-matching systems can connect a page with queries without every exact variation appearing. A keyword group should organize the answer, not prescribe repeated phrases.
Collect candidates with their provenance
- Internal information: genuine support questions, product records, site searches and existing content. Remove private details before publishing customer material and do not turn a few questions into an invented industry survey.
- Search Console: use an authorized property's queries and pages to find needs already appearing for the site. A small or new site's report cannot represent its whole market.
- Keyword tools: expand seeds into related phrases, questions and estimates. Record the tool, database, market, date and original export.
- Search results: identify pages serving the same task. Search competitors need not be the same companies as your sales competitors.
Search Console's query documentation describes anonymization and data truncation. An absent query does not establish zero market demand. A recorded impression also does not establish a reliable conversion opportunity.
When using a competitor-gap tool, choose the correct regional database, keyword type and comparison scope. Semrush's Keyword Gap documentation distinguishes Missing, Weak and Untapped terms. These describe relationships within the selected comparison and database; they are not automatic writing assignments. Another brand's login queries, unrelated recruitment and products outside your scope may be unsuitable.
Understand the metrics before prioritizing
| Metric | Useful interpretation | Do not equate it with |
|---|---|---|
| Average monthly searches | A demand estimate with a defined tool, place and period | Unique people, guaranteed traffic or independently additive synonym demand |
| Keyword Planner competition | Advertiser competition for the selected settings | Organic SEO difficulty |
| KD or PKD | A proprietary estimate of organic competition or domain-specific difficulty | A Google score, success probability or deadline |
| CPC and bid ranges | Signals about the advertising market | Your organic visitor value or acquisition cost |
| Google Trends | Relative interest within the chosen comparison | An absolute search count |
| GSC impressions and clicks | Observed performance for your site and report scope | Complete market demand |
Keyword Planner's documentation says average monthly searches include close variants and depend on the selected months, location and Search Network. Figures are rounded. Historical volume for the same keyword does not become three independent demand estimates when you change broad, phrase and exact match types; advertising forecasts consider match types and other settings. Do not sum overlapping estimates and call the result unique searchers.
Ahrefs KD estimates difficulty reaching the top ten. Semrush PKD additionally considers the entered domain's relevance and competitive situation. Scores from different systems are not interchangeable. A percentage sign does not turn a difficulty value into a probability of ranking.
Google Trends uses sampled, normalized interest. Low-volume terms may display zero, and apparent spikes for low- or no-interest queries can include statistical noise. Preserve the comparison settings; a value of 100 is not one hundred monthly searches.
Map related searches to a useful destination
Group queries by the searcher's task, then inspect your existing pages. The question is whether one page can naturally answer the group completely. Shared vocabulary alone is insufficient, and there is no required number of secondary keywords per page.
Hypothetical packaging-supplier plan: no live search volumes or difficulty scores are claimed for these candidates.
| Candidate question | Task and evidence needed | Page decision |
|---|---|---|
| What information is needed for a packaging quote? | Prepare dimensions, quantity, structure, printing and delivery requirements | Improve the existing quotation-preparation page if it serves this task |
| Internal vs external box dimensions | Understand definitions, measurement directions and recording conventions | A measurement-guide section or a separate explanation if the answer warrants it |
| Buy a particular box model | Check actual specifications, variants and supply conditions | The corresponding product page |
| Another brand's account login | Reach a specified website | Usually not a sales-content opportunity for this supplier |
Mark uncollected data as unknown. A small technical demand may still be important to a specialized audience. Whether it needs its own page depends on the completeness and distinctness of the answer, not a minimum search-volume rule.
Explain what your content can add
Read representative pages and record their type, answer, evidence, functionality and unresolved questions. Do not use domain metrics, word counts or dates as a substitute for evaluation. An older technical explanation may remain accurate; a newer article may simply repeat it.
Turn a genuine advantage into the brief: verifiable model differences, usable instructions, original product documents or a working tool. If key data or testing capability is missing, record the dependency instead of claiming a hands-on review. Where no meaningful contribution is available, defer the candidate or research other lower-competition opportunities.
Make the final list actionable
Each row should make the intended reader, main task, query group, market, data source and date, destination URL, questions, required evidence, owner and decision clear. The decision may be to update a page, create a distinct answer, gather information or defer. These are project choices rather than Google ratings.
Let the title describe the subject and benefit, then answer the important questions in the appropriate sections. Use model names and constraints where relevant, with descriptive links to the next resource. Do not change an established URL merely to add a phrase or pack all variants into the description. Google's title guidance also discourages repetition and keyword stuffing.
Check publication before judging the keyword choice
Verify content, metadata, crawlability, mobile reading and relevant links first. Then inspect actual queries, impressions and clicks under comparable market, device, page and time settings, alongside useful task completions or qualified inquiries. A free plan, a trial and a paid subscription may provide different data; consult current product terms rather than assuming an old tutorial's allowance still applies.
If a page gets no impressions, distinguish indexing, sparse demand, competition, a mismatched answer and insufficient observation. If it gets impressions without clicks, inspect the queries and positions before rewriting. The click-change diagnostic develops that process. The purpose of a keyword plan is an explainable, reviewable decision—not a traffic promise made from estimates.