The short version: you can do proper keyword research using nothing but Google Search, Google Search Console, YouTube, and Reddit, and honestly it often produces better keyword lists than a paid tool because you're looking at real questions instead of a database's guess at search volume. The catch is time, not quality: free research takes longer and forces you to think, which is exactly why most people skip it and pay £99 a month for a shortcut they don't need.
Why I stopped paying for keyword tools (mostly)
I used to have three separate SEO subscriptions running at once. Somewhere around 2021, when I was rebuilding my business after a rough stretch and watching every subscription like a hawk, I cancelled two of them in the same week and just... kept doing keyword research. Nothing broke. A blog post I wrote using free tools that quarter ranked page one for "small business marketing consultant UK" within four months, which is a term I'd previously been chasing with software that told me it had "medium difficulty" and a search volume of 90 a month. The software was wrong on both counts, as it turns out. It's now sending me two to four enquiries a month.
That's not an anti-tool rant, by the way. Paid tools are faster once your list of topics gets long. But if you're a small business, a freelancer, or you're just starting out and every pound matters, you do not need to spend money to find out what people are typing into Google. You need to spend attention instead.
The tools you already have open right now
Google itself is still the best free keyword research tool that exists, and almost nobody uses it because it's not shiny.
- Autocomplete. Type your topic into Google and stop. Don't press enter. The four to eight suggestions that drop down are real queries, pulled from real searches, updated constantly. Type "keyword research" and you'll currently see things like "keyword research for beginners" and "keyword research template" hanging off it.
- People Also Ask. Search your main term, look at the expandable question boxes, click two or three open. New questions appear underneath them. Keep clicking and you'll have fifteen to twenty related questions in about four minutes. Every one of them is a potential subheading or a potential blog post.
- Related searches. Scroll to the bottom of the results page. Google gives you eight more terms it associates with your search. This is where you often find the phrase real people use, which is frequently not the phrase you assumed they'd use.
- Google Search Console. Free, and it's the only place that shows you the exact queries that already bring traffic to your existing site, including ones you're ranking for on page two or three without knowing it. This one detail alone has been worth more to my clients than any keyword database, because it shows demand you're already half-capturing.
- YouTube search bar. Autocomplete works there too, and video searchers phrase things differently to Google searchers. "How to" prefixes show up far more.
- Reddit and forums. Search your topic plus "reddit" directly in Google. You'll land on threads where people ask the question in their own words, complain about the thing they can't find an answer to, and argue about it in the comments. That argument is often your best content angle.
A step-by-step process that takes about 40 minutes
Here's the exact process I still use for client blog posts, including the ones I write for this site.
- Write down the one thing your reader is trying to achieve. Not a keyword, a goal. "Find customers without a marketing budget" for example, not "cheap marketing."
- Type three versions of that goal into Google and note every autocomplete suggestion. That's usually 12 to 20 phrases in ten minutes.
- Open People Also Ask on each search and click through four levels deep. Copy every question into a spreadsheet.
- Check Google Search Console under Performance, filter by the page closest to this topic, sort by impressions. Anything with high impressions and low clicks is a query Google already thinks you're relevant for, but your title or content isn't winning the click.
- Search the topic plus "reddit" and skim two threads for phrasing and complaints.
- Group everything into three buckets: what it is, how to do it, what it costs or how long it takes. Almost every topic breaks down this way.
- Pick the phrase that sounds like something a person would say out loud, not the stiff, formal version. That's usually your title.
Forty minutes, no login, no card details. The output is a spreadsheet of 30 to 50 real phrases and questions, grouped by theme, which is more than enough to plan a month of content.
The bit nobody tells you about search volume numbers
Here's the uncomfortable part. Even the paid tools are estimating. Ahrefs, Semrush, and the rest pull from clickstream data samples and Google Ads bid data, then model the rest. Two paid tools will often give you two different volume numbers for the same keyword, sometimes off by 300 percent. I've seen a term show "40 searches a month" in one tool and "480" in another, for the exact same phrase, checked the same week.
So when you're doing free keyword research and you can't see a volume number at all, you're not missing that much certainty, you're just missing the illusion of certainty. What matters far more than the number is whether the phrase sounds like something a real customer would type, whether there's a cluster of related questions around it (which signals real, sustained interest rather than a one-off search), and whether you could write something better than what's currently ranking. None of that needs a subscription. It needs you reading the search results for ten minutes instead of trusting a graph.
Using competitors to fill in the gaps
Once you've mined Google itself, the next free source is your competitors' own content, because they've usually already done keyword research, and you can reverse it out of what they've published. Look at the headings on their top three blog posts, the FAQ sections at the bottom of their service pages, and the questions their own site's search bar suggests if they have one. I've written about this in more depth, including how to spot which of a competitor's pages are pulling traffic versus just sitting there, in how you do keyword research on your competitors, and there's a fuller walk-through of building this into a repeatable process rather than a one-off exercise in how to do competitor keyword research for your niche.
A quick example. A client of mine in the accountancy space had a competitor ranking well for "self assessment deadline extension." My client had never considered that phrase because they assumed everyone already knew the deadline. Turns out thousands of people search it every January because they've missed it or think they might. That one phrase, lifted straight from a competitor's heading and rewritten better, became one of their best performing pages that tax season.
What to do once you have the list
A spreadsheet of keywords is worthless if it sits there. Once you've grouped your phrases into themes, pick one theme, write the piece, and then track whether it moves. You don't need paid rank tracking software for this either. I cover the manual version, including how to check your position without triggering Google's spam detection by hammering search results with automated checks, in how do I track my keyword rankings without expensive software, and if you later decide tracking software is worth the money, I've laid out what it tells you (and what it quietly doesn't) in how an SEO rank tracker helps you monitor keyword positions.
If your site runs on WordPress, a lot of your keyword work gets undone by basic technical mess, slow load times, missing internal links, title tags that don't include the phrase you researched. Worth sorting before you write another word, and I've laid out the fixes that matter most in how to improve SEO on a WordPress website.
Where free research runs out
I'll be blunt about the limits, because pretending free tools do everything isn't useful either. Free methods struggle at scale. If you're planning content for 200 keywords a month across a large site with several writers, manually pulling autocomplete and People Also Ask for each one becomes a full time job on its own, and at that point a paid tool's time saving earns its subscription. Free methods also don't show you click-through rate curves, keyword difficulty scores, or backlink gaps between you and page one, though honestly most small businesses were never competing on those metrics anyway, they were competing on writing something clearer than the five other tired posts already ranking. If you're weighing up whether that stage has arrived for your business, or whether you'd get more from someone setting up the process rather than buying more software, that's usually the point where working with an AI consultant for small business to build the workflow once, rather than buying tools you'll half use, pays for itself faster.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
And one more thing worth saying plainly: most businesses doing keyword research don't have a tool problem, they have a "we haven't looked at what our customers type" problem. I've sat in meetings where a team debated keyword strategy for forty minutes without once opening Google and typing the phrase they were arguing about. Free or paid, that's the actual mistake.
Frequently asked questions
Is free keyword research accurate enough for a small business website?
Yes, for most small businesses it's more than enough. You're not competing for national volume terms, you're trying to match the exact phrase your local or niche customer types, and Google's own autocomplete, People Also Ask, and Search Console show you that directly without any modelled guesswork.
What's the biggest limitation of doing keyword research without paid tools?
Speed at scale. Manually pulling questions and phrases for five blog posts a month is fine. Doing it for 200 keywords across a large content calendar becomes slow enough that a paid tool's time saving starts to justify the subscription.
Can Google Search Console really replace a keyword tool?
For your existing site, yes, better in some ways, because it shows you queries you're already ranking for on page two or three, meaning real demand you're half-capturing, rather than a modelled estimate of demand that may not reflect your actual audience at all.
How do I find keywords for a brand new website with no traffic yet?
Use competitor content and forum threads instead of your own data. Pull the headings and FAQ sections from the top three ranking pages for your topic, search your topic plus "reddit," and build your list from the real questions people are asking there before you've got any Search Console data of your own to work from.
Primary sources
Related reading: Ubersuggest: the new all-in-one keyword research tool and The 5 Best Keyword Research Tools for Entrepreneurs.
For the bigger picture, see my full guide to SEO.
