Keyword research is where most content strategies quietly go wrong. Not because people skip it — but because they do it in a way that produces a list of keywords they have no realistic chance of ranking for, then spend a year writing against that list and conclude that SEO does not work.
Good keyword research is less about finding keywords and more about honestly assessing which battles you can win.
What a keyword actually represents
A keyword is not a magic phrase to sprinkle into text. It is a record of a real person’s problem, in their own words, at a moment when they wanted help.
Treating it that way changes everything. You stop asking “how many times should I repeat this phrase?” and start asking “what did this person actually want, and what would make them feel that they found it?”
The four numbers that matter
Every keyword has a handful of attributes worth checking before you commit hours to writing.
Search volume
Roughly how many people search it monthly. Useful, but routinely over-weighted by beginners. High volume attracts high competition, and a keyword with 200 searches that you rank first for beats a keyword with 50,000 that you rank fortieth for. Fortieth position gets nothing.
Keyword difficulty
An estimate of how hard it will be to rank, usually based on the strength of sites currently ranking. Treat the score as a rough signal, not gospel — and always sanity-check it by actually looking at the results page.
Search intent
What kind of answer the searcher wants. This determines the format of what you must write. Get it wrong and nothing else saves you.
Business value
The forgotten one. Some high-volume keywords bring readers who will never do anything useful for you. Some low-volume keywords bring exactly the right person. Volume is not value.
Step 1: Start with your topic, not a tool
Before opening any software, write down what you actually know and what your readers repeatedly struggle with. Tools are excellent at expanding a list and terrible at inventing one from nothing.
List 5–10 broad areas within your subject. These are seeds, not keywords.
Step 2: Expand the seeds
Now expand each seed into real phrases people search. Free sources are surprisingly good:
- Search autocomplete. Start typing your seed and note the suggestions — these are real, common queries.
- “People also ask” boxes. A direct feed of related questions, and each one is a potential H2 or a whole article.
- Related searches at the bottom of the results page.
- Your own site search and analytics. What are people already looking for once they arrive?
- Communities and forums. Real phrasing, real frustration, and no marketing gloss.
Step 3: Read the results page before you decide
This is the step that separates useful keyword research from wishful thinking. Search the keyword and study the first page carefully.
Ask yourself:
- What format dominates? If it is all listicles and you planned a tutorial, adjust or move on.
- Who is ranking? If the entire page is enormous, established brands, a new site will not break in. If there are forums, small blogs, or genuinely mediocre pages, that is an opening.
- Could I plausibly make the best page here? Be honest. If your answer is “I’d write something equally good”, that is not enough — equal does not displace incumbent.
If you cannot articulate why your page deserves to outrank what is already there, you are not ready to write it.
Step 4: Prioritise ruthlessly
Score each candidate keyword on three things: how winnable it is, how valuable the reader is to you, and how much you can genuinely say about it.
Then start with the winnable ones. Early wins matter enormously — not only for traffic, but because ranking builds site authority that makes the next, slightly harder keyword achievable. Sites climb; they do not teleport.
Step 5: Group keywords into pages, not one keyword per page
A very common error is creating a separate thin article for every close variation. “How to build an email list”, “building an email list” and “email list building tips” are the same search intent. They should be one strong page, not three weak ones competing with each other.
Group by intent. If two keywords would be satisfied by the same article, they belong on the same article.
Step 6: Build topic clusters
Rather than a scattered list, organise your keywords into clusters: one broad pillar page covering a subject comprehensively, surrounded by focused articles on specific sub-questions, all interlinked.
This does two things. It demonstrates genuine depth on a topic, and it channels authority around the cluster instead of spreading it thin across unrelated pages.
Common mistakes worth avoiding
- Chasing volume. The most-searched keyword in your niche is almost certainly the one you have the least chance of winning.
- Ignoring intent. The single most common reason a well-written article never ranks.
- Trusting difficulty scores blindly. They are estimates. The results page is the evidence.
- Keyword stuffing. It reads badly and it has not worked for many years.
- Writing for keywords instead of people. If the article only exists to catch a phrase, readers can tell, and increasingly so can search engines.
A simple starting workflow
- Pick one narrow topic you genuinely know.
- Collect 30–50 real questions people ask about it.
- Search each one and note the format that ranks and how strong the competition looks.
- Discard anything dominated by major brands with excellent content.
- Group what remains by intent — you will likely end up with 12–20 real article ideas.
- Order them by winnability and start at the top.
Do that properly once and you will have a content plan worth six months of work — and, more importantly, one where the work has somewhere to land.
Frequently asked questions about keyword research
Do I need a paid keyword research tool?
Not at the start. Autocomplete, “People also ask” boxes, related searches and the communities where your audience already talks will give you more real queries than you can write about in a year. Paid tools mainly save time and add difficulty estimates — useful once you are publishing at volume, unnecessary when you are writing your first twenty articles.
What search volume is too low to bother with?
Very little, if the reader is right. A keyword with fifty monthly searches from people about to hire someone like you is worth more than one with fifty thousand searches from people who will never buy anything. Volume is a proxy for value, and often a poor one. Judge the searcher, not the number.
How do I know if a keyword is too competitive?
Search it and look honestly at page one. If every result is a large, established brand with comprehensive, current content, you will not break in as a new site — regardless of how good your article is. If you can see forums, outdated pages, thin content or small sites, that is an opening. The results page is better evidence than any difficulty score.
Should I target keywords my competitors already rank for?
Only if you can genuinely do better, and can articulate how. Matching an established page is not enough to displace it — search engines have no reason to demote an incumbent for an equal newcomer. You need a clear reason: more current, more complete, more specific, or drawn from experience they visibly lack.