Keyword research when you have no budget and no idea where to start

Start with what you already rank for

Before hunting for anything new, look at the keywords you fell into by accident. Almost every site ranks for something nobody planned. A service page written in a hurry, an old post answering a question a customer asked twice, a page you forgot existed.

Google Search Console shows you this for free. Connect the site, give it a few days to gather data, then open the Performance report and sort by impressions. The rows with plenty of impressions and almost no clicks are the ones to care about. Google already thinks you are relevant enough to show. People are seeing you and picking someone else, which is usually a title problem or a position problem, and both are fixable in an afternoon.

Write down the ten terms with the most impressions. That list beats anything a keyword tool hands you cold, because it is built on what Google has already decided about your site rather than what you hope is true.

Search volume is the least useful number

Beginners fixate on volume. A tool says 40,000 searches a month and it feels like a jackpot. Then you look at who ranks and it is Wikipedia, three national publishers and a competitor with fifteen years of links.

Volume tells you how many people search. It says nothing about whether those people buy, or whether you have any chance of reaching them. A term with 90 searches a month where every result is a small business page is worth more to a plumber than one with 40,000 searches dominated by Home Depot.

A rough filter that works: if the first page is entirely made of sites much larger than yours, skip it for now. Come back in a year. There is always a longer, more specific version of the same query where the competition is people your own size.

Write down what customers actually say

The best keyword source is not a tool. It is your inbox, your phone, and the notes you take after a job.

People search the way they speak. A customer does not call you asking about "HVAC preventative maintenance solutions". They ask why the upstairs is always cold. That question, more or less word for word, is what they typed into Google an hour earlier.

Keep a running note for a fortnight. Every question a customer asks, write it down verbatim. Do not tidy the grammar. At the end you will have twenty or thirty phrases that are already proven to come out of a real buyer's mouth, and most of them will have almost no competition because your competitors are busy chasing the 40,000-a-month term.

Free tools that are genuinely enough

You can get a long way without paying anyone.

• Google's own autocomplete. Start typing your service and let it finish. Those suggestions come from real search behaviour.
• The "People also ask" box on a results page. Every question in there is one someone genuinely asks, and each one can be a section of a page or a page of its own.
• The "Related searches" list at the bottom of a results page. Same idea, broader.
• Google Search Console, again. Free, and specific to you rather than to the market in general.
• Your competitors' navigation menus. If four rivals all have a page on the same service, that service gets searched for.

Sort by what the searcher wants, not by topic

Two people searching similar words can want completely different things. Someone typing "how much does a new boiler cost" is researching. Someone typing "boiler installation near me" is close to booking. One needs an article, the other needs a service page with a phone number near the top.

Group your list into three buckets before you write anything. People learning, people comparing, people ready to buy. Then check what Google is already showing for each term. If the whole first page is blog posts, Google has decided the term wants information, and a sales page will not rank there no matter how good it is.

This one habit saves more wasted writing than any other. Match what is already ranking, or accept that you are arguing with Google about what its users want.

Pick five and stop

A list of 300 keywords is a way of avoiding the work. Pick five. Ideally five where you already have impressions, where the competition looks like you, and where the searcher is somewhere near a decision.

Give each one a page, or improve the page you already have that is closest to it. Then leave them alone for two or three months. Search results move slowly and checking daily will only make you miserable.

Once those five are settled, look at the next five. That is the whole loop. If you want the list kept for you, along with a plan that says which page to touch first, that is what our action plan does, and you can join the beta to try it. Otherwise a spreadsheet works fine, and our guide to the numbers worth tracking covers what to check when you come back to it.