Affiliate disclosure: BeingAffiliate.com may earn a commission when you purchase through certain links, at no additional cost to you. Our recommendations are based on relevance and usefulness, not commission rates.
Keyword research is not a hunt for phrases with attractive numbers. It is a method for discovering how people describe problems, what they need next, and which questions your website can answer better than the current results. Paid tools can accelerate this work, but a beginner can make sound publishing decisions with search results, forums, program documentation, and a spreadsheet.
Begin with audience problems, not keywords
Write five situations your reader encounters and five decisions each situation creates. For a hypothetical home-office site, “poor video-call sound” may lead to questions about room echo, microphone type, placement, compatibility, and budget. These are seed topics grounded in a real problem.
If you cannot identify a coherent reader, use the niche decision framework first. Research becomes noisy when every popular query appears relevant.
Understand search intent
| Intent | Reader need | Possible article |
|---|---|---|
| Learn | Understand a concept or task | How microphone pickup patterns work |
| Solve | Fix a specific problem | Reduce keyboard noise on video calls |
| Compare | Choose between approaches | USB vs XLR microphones for remote work |
| Evaluate | Judge a product or shortlist | Best microphones for untreated rooms |
| Navigate | Find a known brand or page | Usually belongs to the brand itself |
Commercial intent does not automatically make a query valuable. A specific troubleshooting question can attract the right reader earlier and create a natural path to a later comparison.
A free step-by-step process
1. Collect audience language
Read public communities, product reviews, support questions, retailer questions, and your own messages. Record exact problem language without copying anyone’s content. Reviews are especially useful for discovering criteria: setup difficulty, size, durability, compatibility, cancellation, support, and hidden costs.
2. Use search suggestions carefully
Type a seed question and observe autocomplete, related searches, and “People also ask” where available. These features change by location and personalisation, so treat them as clues rather than volume reports. Search several natural variations rather than forcing an exact phrase.
3. Inspect the result page
Ask what search engines believe the query means. Are results tutorials, product pages, videos, forums, category pages, or comparisons? A page that does not match the dominant need may struggle even if it repeats the phrase perfectly.
4. Find content gaps
Open several strong results and note what they cover, what they omit, how recently facts were checked, and whether the advice is supported. A gap might be an underserved audience, unclear explanation, missing limitations, outdated interface, weak comparison criteria, or inaccessible presentation.
5. Group related questions
Cluster questions that can be answered completely on one page. “How high should a microphone sit?” may belong inside a setup guide; “USB versus XLR” deserves a separate comparison. Avoid creating several nearly identical pages that compete with each other.
6. Prioritise realistically
Score each proposed article from one to five for reader value, relevance to the site, confidence in producing credible content, commercial connection, competition, and maintenance cost. Do not pretend the total predicts traffic; use it to make trade-offs visible.
Example: prioritising a small cluster
A hypothetical beginner coffee site finds four topics: “coffee grinder,” “why coffee tastes bitter,” “burr vs blade grinder,” and “best grinder.” The broad term is ambiguous and competitive. The bitterness guide is useful but not always product-led. The comparison has a clear decision and teaches criteria. The roundup requires testing several products. A new publisher might produce the troubleshooting guide and comparison first, then attempt the roundup only when evidence supports it.
Estimate demand without false precision
Free trend tools can reveal seasonality and relative change. Search Console later reveals the queries for which your own pages appear. Community frequency and repeated support questions show qualitative demand. None provides a guaranteed visitor number. Record uncertainty and update priorities after publishing.
Mistakes to avoid
- Selecting a query only because a tool labels it “easy.”
- Writing separate pages for trivial word variations.
- Ignoring the format and intent visible in results.
- Targeting product decisions you cannot evaluate credibly.
- Assuming a high-volume term will produce qualified buyers.
- Copying competitor headings instead of finding a useful angle.
Research checklist and action plan
- Define one audience situation.
- Collect twenty real questions from at least three source types.
- Label each by intent and decision stage.
- Inspect current results for ten promising questions.
- Cluster overlaps into distinct articles.
- Score value, fit, evidence, difficulty, and maintenance.
- Brief the strongest article and apply the on-page SEO process.
Frequently asked questions
Do I need search volume?
No precise threshold is required. Volume can help prioritise, but strong relevance, clear intent, and your ability to produce the best answer for a narrow audience often matter more.
Can I use AI for keyword ideas?
AI can help expand language, but it does not prove that people search a phrase or that results match the proposed intent. Validate ideas against real sources and current results.
How many keywords should one article target?
Target one coherent need and naturally answer its closely related questions. A useful page may appear for many phrases without treating each phrase as a separate target.
Research decisions, not just phrases
A free process can outperform an expensive dashboard when it keeps you close to reader language and visible search intent. Choose one cluster, publish the most defensible article, and use real impressions and feedback to improve the next research cycle.