Most keyword research produces a spreadsheet of high-volume terms the business has no chance of ranking for. Useful research does the opposite: it finds the terms where you can win, and tells you honestly which ones to leave alone.
Good keyword research starts with what your customers actually type, not with volume. Group terms by intent, then filter by whether you can realistically compete against what already ranks. A term with 200 monthly searches you can win is worth more than one with 20,000 you cannot.
Start with your own data, not a tool
- Search Console. Queries you already appear for, including ones you never targeted. This is the most under-used source anyone has.
- Your enquiry inbox. The words customers use when describing what they need.
- Sales calls. How people phrase the problem before they know your terminology.
- Site search. What visitors look for once they arrive and cannot find.
- Support tickets. The confusion that becomes a search.
Tools are for expanding and validating. They are a poor starting point because they hand you the industry's vocabulary rather than your customers'.
Intent decides everything
A term is only worth targeting if the page you would build matches what the searcher wants. Volume is irrelevant if the intent is wrong.
| Intent | Looks like | What ranks |
|---|---|---|
| Informational | how does X work, what is X | Guides, explainers |
| Commercial investigation | best X, X vs Y, X reviews | Comparisons, listicles, directories |
| Transactional | X company, hire X, X near me | Service pages, local results |
| Navigational | brand name | The brand's own pages |
The most common mistake is writing a blog post for a transactional term. Someone searching for a web design company wants a company, not an article about companies. That search needs a service page.
Judging whether you can win
Difficulty scores from tools are a rough proxy. The reliable method is to look at what actually ranks.
- Search the term in a private window.
- Look at who holds the top five. Directories and national brands mean it is hard.
- Read the top result. Could you produce something genuinely better, or only equivalent?
- Check how many results are from sites of your size. If none are, that is your answer.
- Check whether the page type matches what you would build.
If the first page is Clutch, Forbes and three national agencies, that term is not available to you this year regardless of how good your page is. Move down the tail.
The long tail is where small sites win
Shorter terms have more volume and more competition. Longer, more specific terms have less of both, and considerably higher intent.
- web design is unwinnable and half the searchers are students.
- web design company is dominated by directories.
- dental website design is competitive but reachable.
- dental website design for a single-site practice has small volume and a buyer at the end of it.
Ten long-tail terms that convert beat one head term you will never reach. Head terms tend to arrive later as a by-product of ranking for the tail.
Turning the list into pages
- Group terms that share intent and would be answered by the same page.
- One page per group. Two pages for closely related terms compete with each other.
- Assign each group a page type: service page for transactional, article for informational.
- Prioritise by intent first, then by whether you can win, then by volume.
- Write down which page owns which term, so nobody later publishes something that competes.
That last step is what prevents cannibalisation. Without a documented map, six months of publishing produces several pages fighting over the same query.
Volume figures are estimates
Every tool's volume number is modelled, not measured, and different tools disagree substantially for the same term. Treat them as relative signals rather than facts.
The only volume data that is genuinely yours is in Search Console, and it only covers terms you already appear for. Use tool numbers to compare terms against each other, not to forecast traffic.
Frequently asked questions
What is keyword research?
The process of finding the terms your customers search, judging what they want when they search them, and deciding which you can realistically rank for. The output should be a map of which page targets which term, not a spreadsheet of volumes.
What are the best keyword research tools?
Google Search Console first, since it shows terms you already appear for. Beyond that, Ahrefs, Semrush and Google Keyword Planner are the common choices. Every tool's volume figure is an estimate, so use them to compare terms rather than to forecast traffic.
Should I target high-volume keywords?
Usually not at first. High volume means high competition, and a term you cannot reach produces nothing. Long-tail terms with lower volume and clearer intent convert better and are winnable, and head terms often follow once you rank for the tail.
How many keywords should one page target?
One primary term and a handful of close variants that mean the same thing. A page cannot rank well for several unrelated topics, which is why one page per service and one per question works better than combined pages.
What is keyword cannibalisation?
Two or more of your own pages competing for the same term, which splits signals so neither ranks well. It is prevented by documenting which page owns which term before publishing, and fixed by merging or consolidating the weaker pages.