How to Do Keyword Research for SEO: A Beginner Workflow
Build a useful keyword list, group related searches and map each topic to the right page without inventing search-volume figures.
By New Delhi Website Technologies · Practical SEO learning series
Start with a website you own or are authorised to edit. Examples are teaching exercises; search rankings and traffic results are not guaranteed.
What you will learn
Keyword research is the process of learning how people describe a problem and deciding which pages can answer it well. A spreadsheet full of phrases is only the starting point. The useful output is a prioritised page plan with a clear audience, search intent and evidence behind each choice. In this lesson you will turn customer questions into topic groups, inspect search results and choose a realistic first page. Any numerical example below is illustrative, not a claim about current market demand.
1. Begin with the language customers already use
Collect questions from sales calls, enquiry forms, support messages and conversations you are allowed to use. Remove personal details before sharing notes. A website agency might hear questions about development cost, delivery time, maintenance and whether a shop needs ecommerce. Write those questions in ordinary language. Do not start by multiplying every service phrase by every neighbourhood. That produces a large list without demonstrating that the pages would answer different needs. Customer language gives you useful topics to investigate, even before a keyword tool is involved.
2. Build seed topics before collecting variations
Create a small set of seed themes such as website cost, online-store setup, mobile design and ongoing support. Under each theme, list related questions and the kind of answer required. Cost questions may need a scope comparison; troubleshooting questions need a diagnostic guide. Keep spelling variants in the same working group unless they reflect a genuinely different meaning. Your goal is to understand the topic, not to create one page for every wording. Add columns for audience, location relevance and the business service the topic can reasonably connect to.
3. Use search results to test your interpretation
Search a candidate phrase and look at the types of pages that appear: tutorials, service pages, product listings, videos or local results. Record the pattern and date as an observation, not a fixed law. If the results are mostly beginner explanations, a hard-selling landing page may be a poor match. Look for questions that competitors leave unanswered and examples you can provide from your own work. Do not copy their structure mechanically. The purpose is to understand expectations and find a useful angle you can support with accurate detail.
4. Treat tool metrics as estimates with context
If you use a keyword tool, record the provider, location, language and date beside its volume estimate. Different tools may report different figures, and small local queries may have incomplete coverage. Never turn an unavailable figure into a fabricated number. A low-volume query can still be valuable when it closely matches your service and produces qualified enquiries. Keep tool difficulty scores separate from your own assessment of the work needed. Consider whether you have the expertise, examples and website structure to publish a credible answer.
5. Group related searches into a page map
Assign one primary intent to each planned page. Closely related variations can live on the same strong guide when they require the same answer. Separate topics when the visitor task changes, such as learning to build an online shop versus hiring a developer. In your spreadsheet, include the proposed URL, primary question, supporting questions and relevant existing links. Check your current pages before creating anything new. Improving an existing page is often clearer than publishing another article that competes with it for the same reader and repeats most of its contents.
6. Prioritise and validate after publication
Choose topics using a combination of relevance, usefulness, available evidence and realistic effort. Start with the page that fills a clear gap in the customer journey. After publishing, inspect the queries that actually generate impressions and clicks when reporting data becomes available. You may discover that readers use different words from your original list. Update the article when those queries reveal a question it should answer. Keep the original hypothesis in your research log so you can learn from the difference between an assumed keyword opportunity and observed visitor behaviour.
Worked example
Illustrative page map: group “website maintenance cost” and “what does website maintenance include” into one guide with a scope table. Keep “hire website maintenance company Delhi” as a service-page question if the business genuinely offers that service. The guide can link to the service page after explaining options. No volume figure is needed to demonstrate that the two pages serve different tasks.
Your practical exercise
- Collect fifteen real or clearly labelled hypothetical customer questions.
- Group them into five themes with a shared visitor task.
- Check three search queries and record the types of results shown.
- Map one theme to an existing or proposed URL with an original angle.
Save the page URL, the observed problem, your proposed change, the publication date and the check you performed. Record real report data separately from examples or expectations.
Frequently asked questions
Should I choose only high-volume keywords?
No. Consider business relevance, the visitor task and whether you can create a useful answer. Volume estimates alone do not establish value.
How many keywords should one page target?
There is no fixed quota. Organise the page around one clear task and cover the related questions needed to complete it naturally.
Continue learning on this website
Explore our original practical guides below. Each page includes explanations, examples and checks you can apply to your own project.