Keyword research is the job of finding out which phrases your customers actually type, rather than the ones you assume they type. Get it right and everything downstream becomes easier. Get it wrong and you can write brilliant pages that nobody was ever going to search for.
This article gives you a method you can follow without buying anything.
Start keyword research with the words customers use
Most keyword research goes wrong at the first step, because it starts with a tool. A tool can only tell you about phrases you thought to ask it about. So begin somewhere better: with the language your customers use when nobody is watching.
Read your own enquiry emails. Listen to how people describe the problem on the phone. Look at the questions your sales team answers twice a week. Those are real phrases from real buyers, and they are almost always plainer than the words a business uses about itself.
Write them down exactly as they were said. Resist tidying them up. "Why is my roof leaking in one spot" is a better starting point than "roof leak diagnostics", because somebody genuinely typed the first one.
Turn one phrase into a list
Now you can widen it. Type one of your real phrases into a search engine and look at what it offers you. The suggestions that appear as you type are drawn from what other people search for. The related searches at the bottom of the results page are the same thing again.
Meanwhile, the "people also ask" boxes are a gift for keyword research, because each one is a question phrased the way somebody asked it. Collect them. You are not judging yet, only gathering.
Ten minutes of this usually produces thirty or forty phrases. That is enough to work with.
Read the intent before anything else
Intent is what somebody wants when they search, and it matters more than volume. Four kinds cover nearly everything.
- Someone wants to understand something. They are learning, not buying.
- Someone wants to compare options. They are close, but not ready.
- Someone wants to buy or book. They have decided.
- Someone wants a specific site or place. They already know where they are going.
The test is simple and it costs nothing. Search the phrase and look at what already ranks. If the first page is full of guides, the search engine has decided that phrase is a learning question, and a product page will not win it however good it is. If the first page is full of product pages, a guide will not win it either.
This single habit prevents most wasted effort. Keyword research is largely the skill of noticing what kind of page a phrase wants, and then writing that kind of page.
Judge whether a keyword research target is winnable
Volume is the number everybody looks at first and it is the least useful on its own. A phrase searched very often is usually contested by businesses far larger than yours, so appearing on the third page brings you nothing.
Instead, look at who currently ranks. If the first page is national brands, big publishers and comparison sites, that phrase is a long project rather than a quick one. If it contains forums, thin pages, or nobody who really answers the question, there is room. That judgement assumes you know what puts a page on the first page at all. Start there if the wider subject is new to you.
Narrower phrases are usually the better bet for a smaller business. Fewer people search them, but the people who do are closer to buying and there is far less standing in your way. Consequently a page about one specific problem often earns more than a page about the whole category.
Worked example, with made up numbers purely to show the shape of the trade. A broad phrase gets 10,000 searches a month and you might reach position 20, which realistically brings almost nobody. A narrow phrase gets 200 searches and you could reach position 2, which might bring 40 visitors. The narrow phrase wins, even though its number looks fifty times worse.
Group the phrases before you write
Many of the phrases you collected are the same question wearing different clothes. "How much does a new boiler cost", "new boiler price" and "cost to replace boiler" all want one page, not three.
So group them. Each group becomes one page, with the clearest phrase as the main target and the rest as the ways people ask for it. This is the step that stops you writing four thin articles that compete with each other, which is a problem a search engine solves by ignoring most of them.
One idea per page. If a group contains two genuinely different questions, split it.
Decide the order
You cannot write everything at once, so rank the groups. A simple order works well for most businesses. First, phrases where somebody is ready to act and you have a real chance. Second, questions your sales team answers constantly, because those pages save time as well as earning traffic. Third, the broad educational subjects, which take longest and pay off last.
Keyword research is not a one off exercise. Revisit the list every few months, because the way people search shifts, and because Search Console will start showing you phrases you never thought of.
Use your own data once you have some
Google Search Console reports the searches that already bring people to your site, including plenty you never targeted. It is the most honest keyword research you will ever do, because it is a record rather than an estimate.
Look for two patterns. Phrases where you rank around positions five to fifteen are often one improved page away from the first page. Phrases that bring visitors to a page that does not really answer them are usually asking you to write a new page.
What good keyword research looks like when it is done
You should end up with a short list of groups, each with one main phrase, a note on what kind of page it needs, and an honest view on whether you can win it. Not a spreadsheet of two thousand rows. Twenty well chosen groups will keep most businesses busy for a year, and they will be twenty pages that had a reason to exist.
Frequently asked questions
What is keyword research?
It is the process of finding out which phrases people type when they are looking for what you offer, and deciding which of those phrases are worth building a page around. It combines listening to real customer language, widening that with search suggestions, and judging whether a phrase is realistically winnable.
Do I need a paid tool to do keyword research?
No. Search suggestions, related searches and the questions shown in results pages are free and come from real behaviour. Google Search Console is free and reports the searches already bringing people to you. Paid tools speed the work up and estimate volume, but they are not required to start.
What is search intent and why does it matter more than volume?
Intent is what somebody actually wants when they search: to learn, to compare, to buy, or to reach a specific place. It matters more than volume because a page that answers the wrong intent cannot rank however well it is written. Searching the phrase and reading what ranks tells you the intent in seconds.
How many phrases should one page come out of research with?
One main phrase, plus the closely related ways people ask for the same thing. If two phrases would need genuinely different pages to answer them properly, they are two pages. Pointing one page at two intentions usually means it satisfies neither and ranks for neither.
Are low volume keywords worth targeting?
Often yes, especially for a smaller business. A narrow phrase attracts fewer people, but those people are usually closer to buying and far fewer competitors are chasing them. Ranking near the top of a small phrase beats ranking on page three of a large one.
How often should keyword research be repeated?
Every few months is enough for most businesses. The way people search shifts slowly, new products change what you need to rank for, and Search Console keeps revealing phrases you never thought to target. A full restart is rarely necessary once you have a solid list.
Should I target the same keyword on more than one page?
No. Two pages aimed at one phrase compete with each other, and the search engine picks one, usually not the one you wanted. If you already have two pages doing this, the usual fix is to merge them into the stronger address rather than to keep both.
How do I know if a keyword is too competitive?
Search it and look at who ranks. National brands, large publishers and comparison sites on the first page mean a long project. Forums, thin pages or results that do not really answer the question mean there is room. That reading is more useful than any single difficulty score.