Finding keywords in an existing article is more useful than counting repeated words. A practical review identifies the article’s main topic, related phrases, entities, questions, headings, and search-intent clues, then maps those findings to clear optimization decisions. This workflow shows how to use manual review, search data, and keyword-extraction tools together without forcing keywords into the copy.
Overview
A keyword review starts with the text you already have. The goal is not to make every phrase appear more often. It is to understand what the article currently communicates, which topic it appears to target, and whether that target matches the reader’s likely query.
For each article, separate findings into four groups:
- Primary topic: the central subject the article explains or solves.
- Secondary terms: closely related phrases that clarify subtopics, variations, or steps.
- Entities: specific people, products, concepts, places, formats, or tools discussed in the article.
- Questions and intent clues: problems, comparisons, definitions, and actions implied by the wording.
This distinction prevents a common mistake: treating frequency as importance. A phrase may be repeated because it is part of the brand name, navigation, or a generic sentence pattern. Another important concept may appear only once but deserve a heading or a fuller explanation.
Before changing the draft, record its current title, URL, headings, publication or update date, and intended audience. If you are preparing a new article, a SEO content brief template can help you define the target before drafting.
What to track
1. Extract visible terms and phrases
Copy the article into a plain-text workspace or a keyword extractor for articles. Remove menus, author bios, related-post modules, comments, and other elements that are not part of the main copy. If you analyze the full page, those elements can distort the results.
Review the extracted list for:
- Repeated words and two- to five-word phrases
- Words used in the title, introduction, headings, image text, and conclusion
- Synonyms and alternate wording
- Specific nouns that define the subject
- Verbs that reveal the reader’s desired action, such as compare, calculate, fix, plan, or learn
Use frequency as a prompt for review, not as an instruction to repeat a phrase. A simple character counter, word counter, or text-analysis tool can support this first pass, but editorial judgment is still needed to remove irrelevant repetitions.
2. Review structure and intent
List every heading in order. Headings often reveal the article’s intended subtopics more clearly than body paragraphs do. Then summarize each section in one sentence. If two sections answer the same question, the article may have overlap. If a major question is missing, the extraction process has exposed a content gap.
Look for intent signals such as:
- Informational: what, why, definition, meaning, or explanation
- Procedural: how to, steps, checklist, workflow, or tutorial
- Comparative: versus, alternatives, best, pros and cons, or differences
- Transactional or navigational: pricing, download, login, tool, template, or brand-specific language
An article can contain several intents, but one should normally lead. If the title promises a tutorial while most of the text is a broad definition, the keyword map should flag that mismatch.
3. Validate with available search data
If the article has been published, review the queries associated with its impressions and clicks in your search-performance platform. Compare those queries with the terms found in the copy. The data may reveal that readers use a different phrase, ask a narrower question, or discover the page for a subtopic that is barely covered.
Search data is evidence about how the page is being discovered, not a complete list of everything to add. Treat low-volume or limited-impression observations cautiously, especially when the data set is small.
4. Compare methods
| Method | Best for | Limitation |
|---|---|---|
| Manual review | Meaning, headings, entities, tone, and intent | Slower and less consistent across large inventories |
| Search-performance data | Queries that already lead people to the page | Only available for eligible published pages and recorded activity |
| Keyword-extraction tool | Repeated phrases and a fast first-pass inventory | May miss context, synonyms, and terms that appear only once |
5. Build a keyword-mapping worksheet
Use one row for each meaningful topic rather than each word:
| Section | Observed phrase or entity | Role | Action |
|---|---|---|---|
| Title and introduction | Central subject | Primary topic | Keep clear and specific |
| H2 or H3 | Distinct subtopic | Secondary term | Expand, combine, or retitle |
| Body copy | Related concept | Supporting term | Use where naturally relevant |
| Search data | Reader wording not covered | Potential gap | Add a section only if it fits the promise |
For each row, add the reader question it answers and the URL that should own the topic. That final field helps prevent several pages from competing for the same subject.
Cadence and checkpoints
Use the workflow at different points in the editorial cycle.
- Before drafting: extract terms from a brief, trusted reference set, or competing page types to define the article’s scope. Do not copy another page’s wording.
- After drafting: check whether the title, introduction, headings, and conclusion agree about the primary topic.
- During a content audit: compare the current map with performance data, internal links, and nearby articles on the same subject.
- During a refresh: mark terms that are outdated, ambiguous, unsupported, or no longer central to the reader’s task.
A monthly review is reasonable for actively changing publishing programs or important pages with regular data. A quarterly review may be enough for stable evergreen content. Use the same worksheet each time so changes are comparable. Record the review date, editor, data period, target topic, changes made, and next checkpoint.
After editing, use a text comparison tool to check that intended additions remain in the final version and that useful details were not lost during revision. A readability pass should follow the keyword review; the blog readability checklist can help with paragraphs, headings, and sentence clarity.
How to interpret changes
If one phrase becomes more frequent, do not assume the article is better optimized. Ask whether the repetition improves clarity, supports a heading, or simply replaces natural wording. A strong update usually makes the subject easier to identify while adding useful information for the reader.
Use these interpretations as working hypotheses:
- The primary topic appears clearly, but related questions are missing: add useful subheadings or brief answers rather than repeating the main phrase.
- Several phrases describe the same idea: choose a natural preferred term and use variations where they improve readability.
- Search queries do not match the article’s promise: review the title, introduction, audience, and internal linking before changing the entire page.
- A related term attracts attention but has no supporting section: decide whether it belongs in this article or deserves a separate page.
- The extraction tool returns generic words: refine the input, remove boilerplate, and rely more heavily on manual entity and intent review.
Do not add a keyword solely because a tool lists it. Relevance, accuracy, completeness, and clear writing should decide whether a term stays.
When to revisit
Revisit the keyword map on a monthly or quarterly cadence, after a substantial rewrite, or when search-performance data shows a meaningful change in the queries associated with the page. Also review it when the subject, product, terminology, audience, or related article group changes.
Refresh the workflow itself at least every six months, or after a major change to the search platform or the tools used by your team. Interface labels, available reports, and extraction features can change, so verify the current steps before documenting them for others. Keep the underlying editorial principles stable: isolate the main content, inspect context, map topics to sections, and validate changes after publication.
For a practical next pass, choose one existing article and complete these five actions:
- Copy only the main article text into a clean workspace.
- Extract repeated phrases, headings, entities, and questions.
- Check available search queries against the extracted list.
- Complete the keyword-mapping worksheet and identify one genuine gap or mismatch.
- Make only the changes that improve the article’s usefulness, then record the date and schedule the next review.
This process turns “find keywords in text” from a counting exercise into a repeatable editorial workflow. For older pages that need broader treatment, pair it with a content refresh workflow and track the changes in your publishing calendar.