Keyword Research for Beginners: A 2026 Step-by-Step Workflow
Keyword research is not a list of words. It is a map of what people are trying to do, and how your site can be the clearest answer.
Keyword research has a reputation for being a tool problem. Beginners are told to buy a database, export ten thousand rows and sort by volume. The result is usually a spreadsheet nobody uses, full of terms that bring the wrong visitors or that a new site has no chance of ranking for. The work that matters happens before the export: understanding what people are trying to do.
This workflow is the one we use for small sites in 2026. It starts with conversations and a blank page, uses free data where it is good enough, and ends with a short list of pages you can actually write and win.
Start with seed topics, not tools
A seed topic is a broad subject your site can credibly cover: home security, invoicing software, beginner running shoes. Write five to ten of them from the perspective of a customer, not from your product catalogue. If you sell software, the seed is not your feature name; it is the job the customer is trying to finish.
Then talk to five people who match your audience. Ask what they searched for last time they had this problem, what words they used and what they were afraid of getting wrong. Their language is more valuable than a volume estimate, because it tells you which words carry intent. Write the phrases down exactly as they say them.
Build a list from four sources
Four sources are enough for a first list. Search engine autocomplete shows what people type after a seed. Related searches at the bottom of a results page show adjacent questions. Your own site search and support inbox show the words customers already use. Competitor pages show which topics the market has decided are worth covering.
Add every phrase to one sheet with three columns: phrase, source and the job behind it. Do not filter yet. The goal at this stage is coverage, not precision. A list of 80 to 150 phrases is plenty for a small site; anything larger becomes a research project instead of a content plan.
- Autocomplete and related searches for the exact wording people use.
- Site search and support messages for problems you already solve.
- Competitor headings for topics the market validates.
- Community threads for the questions people are embarrassed to ask.
Check search intent before volume
Every phrase implies a job. Someone searching "what is a dmarc record" wants an explanation. Someone searching "dmarc record generator" wants a tool. Someone searching "dmarc setup service" wants to hire someone. The same topic, three different pages, three different outcomes. If you answer the wrong intent, the ranking will not convert.
Read the first page of results for each phrase. If it is all definitions, write a guide. If it is all product pages, write a comparison or a buying page. If it is a mix, the intent is unclear and the phrase is risky. Our guide to search intent and keyword mapping goes deeper on classification and grouping.
Judge difficulty realistically
Volume is a demand signal, not an opportunity. A phrase with 20,000 monthly searches and a first page full of national brands is not a beginner target. A phrase with 300 searches and a first page of thin forum threads may be winnable in a month. For a new site, the second phrase pays the bills first.
Use three quick checks. How many results are dedicated pages rather than incidental mentions? Do the top pages answer the query completely? Does the site have a clear reason to be trusted on this topic? If two of the three answers favour you, add the phrase to a shortlist. If not, keep it for later.
A keyword is only an opportunity if you can be the best answer to it, not just another page about it.
Omar Haddad
Key facts
- Seeds: 5 to 10 topics, written in customer language.
- First list: 80 to 150 phrases from four sources.
- Intent: check the live results page before judging volume.
- Difficulty: prefer phrases where two of three checks favour you.
- Output: one primary keyword per planned page.
Map keywords to pages
Now turn the list into pages. Group phrases that share the same job and the same expected format. "How to set up dmarc", "dmarc setup steps" and "dmarc configuration guide" belong on one page. "Dmarc record generator" belongs on a tool page, and "dmarc consultant" belongs on a service page. One page per job, one primary phrase per page.
Write the page title from the primary phrase, then check that the supporting phrases appear naturally in the outline. If a phrase needs a different format, give it its own page. If it needs the same format, merge it. This is the step that prevents two pages competing for the same search.
Prioritise what you can win
Score each planned page on demand, difficulty, business value and effort. A simple 1 to 5 scale is enough. Multiply demand by value, then subtract difficulty and effort. The highest scores become the next four articles. The rest stay in the backlog, which is a normal and healthy place for good ideas.
Start with pages that support a decision: comparisons, costs, setup steps and troubleshooting. They attract people close to acting, and they are easier to write well because the questions are specific. Save broad definition pages for later, when the site has enough internal links to support them.
Review and prune every quarter
Search demand moves. A phrase that mattered in January can fade by summer, and a new question can appear after a platform change. Once a quarter, open the sheet and check the pages you published: which queries bring impressions, which pages get clicks, and which phrases have drifted to a different intent.
Merge pages that compete, rewrite titles that promise the wrong thing and delete pages that answer nothing. Then add the next four. This rhythm keeps the site focused and stops the content library from becoming an archive of guesses. If you are launching a new product at the same time, pair this workflow with our startup launch checklist so the first pages ship with the product.
Final word
Keyword research is a map of intent, not a hunt for the biggest number. Start with the words customers use, check what the results page expects, and choose phrases you can answer better than anyone. Then write the page, measure it and prune the list. That loop is the whole skill, and it works without a paid database.
For official guidance on how search works and what site owners should prioritise, read Google's SEO starter guide. Treat it as the rules of the road, then apply the judgement above.