SEO
Keyword Research Fundamentals
Keyword research is not a tool. It is a question: what are people actually searching for, which of those searches can I answer better than the pages already there, and which of them matter to my site?
What a keyword is — and what it is not
A keyword is a query: the words a person types when they want something. It is not a magic string to be stuffed into a page. The words matter because they carry intent — what the searcher wants — and the intent decides what kind of page should answer them. A query of four or more words, a long-tail query, is usually more specific and less contested than a one- or two-word head term, and for a new site the long tail is where the winnable ground is. The head terms in this niche — “affiliate marketing”, “SEO”, “make money online” — are owned by large publishers and platform help centres. The long tails — a specific question, a specific comparison, a specific problem — are held by small blogs, and that is where a new site can win.
Two warnings about keyword data. First, the volume and difficulty figures that paid tools show are estimates — guesses built from samples — not measurements of real searches. A beginner can do real keyword research without them. Second, no keyword figure on this site is presented as a measured fact, because none was observable free. The method below uses what is observable: the suggestions search engines show, the questions they surface, and the pages already ranking.
The free method, step by step
You need no paid tool. You need a search engine, a notebook (a spreadsheet works better once the list grows), and the patience to read results instead of trusting a score.
Step 1: seed from your topic
Start with your topic and write down ten seed phrases: the questions your readers ask, the problems they describe, the decisions they face. Seeds come from your own experience, from reader comments and questions, and from forums where your audience talks. A seed is not a keyword yet; it is the start of a search.
Step 2: harvest the suggestions
Type each seed into a search engine and read the suggestions that appear as you type. Those suggestions are real queries people have typed. Note them, including the longer variants. Then run the search and read the “people also ask” questions beside the results — those are the questions search engines have learned are related, and they are often the exact questions your readers are asking. Also read the bottom of the result page, where related searches appear. By the end of this step you should have dozens of real queries for every seed, all observed, none invented.
Step 3: read the top ten
For each query that matters, open the top ten results and read them as a reader would. Ask: are they thorough or thin? Are they current or stale? Are they written by someone who knows the topic, or by a generalist content team? A top ten full of thin, stale or generic pages is a winnable query. A top ten full of thorough, current, expert pages is a query to avoid for now. Note the shape of the results too: are they guides, tools, videos, or product pages? The shape tells you what kind of page the searcher wants — and it is the shape your page must match, a subject covered in our search intent guide.
Step 4: judge what you can answer better
A query is winnable for you when you can answer it better than the pages already there — more thoroughly, more currently, with real experience, or with a tool the incumbents lack. Be honest about this. If the top page is a tested comparison by an expert and you have never touched the product, you cannot answer it better today. Pick the queries where you can. The honest test is a sentence: “I can answer this better because…” If you cannot finish the sentence, the query is not yours yet.
Grouping keywords into pages
One page should answer one intent. Two queries that want the same answer belong on one page; two queries that want different answers belong on two pages. The grouping rule is simple: if the same page would satisfy both queries, they are one group; if it would not, they are two. A group usually contains one primary query — the clearest statement of the intent, which becomes the page’s title and H1 — and several secondary queries, the variants and related questions the page also answers in its headings and FAQs. Grouping prevents the two classic errors: writing two pages that compete with each other for the same query, and writing one page that tries to answer two different intents and satisfies neither.
Spreadsheets are the practical tool for grouping at this scale: one row per query, one column for the group, one for the intent, one for the verdict — winnable now, winnable later, or not winnable. This site’s own inventory was built that way: 123 keywords mapped to their pages, with each page carrying exactly one primary keyword. Our topic cluster guide explains how the groups then become a cluster that supports itself.
How to write posts that rank
This section answers the intent behind “how to write blog posts that rank” — it is the full, substantial treatment of that question on this site, not a separate thin page, because the answer is the method this guide teaches, applied to a single post.
Start with the answer, not the introduction
A post that ranks answers its question in the first paragraph. The reader arrived with a question; the first thing they see should be the answer. Search engines have learned to prefer it, and readers reward it by staying. Everything after the first paragraph is the evidence: the explanation, the worked example, the comparison, the exception. A long introduction that repeats the question before answering it is the most common reason a good post ranks badly — the reader leaves before the answer arrives, and the search engine notices.
Match the page to the intent, exactly
A post ranks when its kind matches the searcher’s intent. A “how to” query wants steps, in order, that a reader can follow. A comparison query wants a structured side-by-side and a verdict. A definition query wants a short, precise answer and then depth. A buying query wants a decision and the evidence for it. If the top results for a query are all step-by-step guides and you publish a definition, you have answered the wrong question — however well you wrote it. Read the results first, match their kind, and then beat them at it. Our search intent guide works through all four kinds with examples.
Be thorough where it matters, and cut where it does not
The advice to write long posts is repeated everywhere in this niche, and it is usually stated without its condition: length helps only where depth helps. A post should be as long as its answer needs. A question with a complicated answer deserves a long post with worked examples, tables and FAQs. A question with a simple answer deserves a short post that says it and stops. The guides that rank for blogging advice suggest a range of roughly fifteen hundred to twenty-five hundred words for a substantial post — but the number is a side-effect of thoroughness, not a target. A long post full of padding ranks worse than a short post that answers completely. Every page on this site has a minimum length because thoroughness is the standard, not because length is the goal.
Write for the reader who arrived, not the reader you wish for
A post that ranks is written for the person who typed the query, in the words they used, at the level they are at. That means using the query’s language in the title, the H1 and the first paragraph — naturally, not stuffed — and answering the follow-up questions that same reader will ask next. The FAQs at the bottom of a post are not decoration; they are the follow-up questions, answered, and they are often the part of the page that earns the extra visibility in search results. Write them from real questions, not from imagination: the “people also ask” box, reader comments, and the questions your own experience says come next.
Show evidence, not adjectives
Rankings follow pages that show their work. A claim with a source, a date, and a link is stronger than the same claim stated as an opinion. A worked example with round numbers, labelled as an example, is stronger than a general statement. A table that compares options side by side is stronger than a paragraph that describes them. A photograph you took, a measurement you made, a step you performed — all of these are evidence that a summary cannot reproduce, and in the age of AI-generated summaries that is the ranking advantage that matters most. Our blogging guide makes the same point: a new blog wins by publishing things a summary cannot.
Publish, then improve — do not wait for perfect
A post does not need to be perfect to rank; it needs to be published, and then improved. New pages take weeks to be found and longer to rank, and a page that sits in draft earns nothing. Publish the thorough version, watch which queries bring readers to it in your search analytics, and then improve it: answer the queries it is appearing for but not yet answering, fix the section readers leave at, update the facts that changed. The best-ranking pages on any site are usually the ones that have been improved three or four times, not the ones that were perfect on day one.
What not to do
Do not stuff keywords. A page that repeats its target query in every sentence reads as spam to a reader and to a search engine, and the search engine has been penalising it for years. Do not buy keyword data you cannot interpret, and do not chase a high-volume head term because its number looks impressive — a high-volume query you cannot win is worth less than a low-volume query you can. Do not write two pages for one intent, and do not write one page for two intents. And do not treat a keyword list as a content plan: a list of queries is the raw material, and the plan is the set of pages that answer them, grouped, prioritised, and written one at a time.
The last warning is about expectations. Keyword research does not promise a ranking. It identifies a winnable query and tells you what the winning page looks like; the ranking still depends on writing that page, and writing it better, and keeping it current. Gate A in the research language means winnable on observable evidence — not a promise. The pages that rank are the ones where the research, the writing and the maintenance all happened.
Last checked: 14 November 2026.