keyword research without paid tools is entirely possible, and for a lot of small sites it produces better decisions than a subscription ever did. When you cannot see a volume number, you are forced to look at what people actually type, what currently ranks, and what questions keep coming back. Those are stronger signals than a monthly estimate, and they cost nothing but attention.
The direct answer: you build a keyword list from four free inputs, which are search suggestions, the pages that currently rank, community questions, and your own performance data. You label each entry for intent and evidence, then score and cut. Volume figures are optional throughout.
Why is keyword research without paid tools still worth doing?
Because the number you are missing is the least reliable part of the paid experience anyway. That is the first thing to accept about keyword research without paid tools.
Most paid tools estimate monthly search volume by modelling data they cannot see in full. A query showing 12,000 monthly searches does not mean 12,000 people will arrive. It means a model produced a number that will be wrong by some unknown margin.
What free research gives you instead is direction with reasons attached. Fourteen suggestion variations means people use a phrase many ways. Thin pages from low-authority sites means the bar to compete is low. A question appearing in six threads across two years means the confusion is persistent.
Direction with reasoning beats a number with no context, especially when the number is the part you cannot afford to check. That is the honest case for the method: the free signals answer the questions a small site actually has.
There is a second reason that matters more for small sites. Paid tools encourage a habit of optimising for the biggest number on the screen. Working without them pushes you toward a better question, which is whether you can genuinely satisfy this search better than the pages currently doing it.
Keyword research tells you what people search. The wider discipline covered in content gap analysis also asks what you already cover and what competitors leave unanswered, which is the part that turns a term list into a decision.
What should you collect before you start?
Four inputs, none requiring a tool. Setting these up first is what separates keyword research without paid tools that works from an afternoon of scattered searching.
A seed description. Write one paragraph about what your site does, who it serves, and what problem it solves. Without it, your research drifts toward whatever is popular rather than what is relevant.
A list of your existing pages. Every URL you have published, with its title. The biggest risk is proposing a topic you already covered eighteen months ago and forgot about.
Access to search suggestion interfaces. Google, Bing, and YouTube all run public suggestion endpoints that return phrasing as you type. Reading them is free. Treating their output as volume data is not. This is the most important input in keyword research without paid tools, and the one most people misuse.
Your own performance data, if you have it. Search Console, or whatever analytics shows which queries brought people in. If you have nothing, note that you skipped this input rather than guessing. It is the strongest input available, so the method loses the most when it is missing.
A fifth item is optional: three to five sites you consider your real competitors. Not the biggest sites in your category. The ones that show up next to you for the searches you care about.
How do you mine search suggestions in keyword research without paid tools?
Search suggestions are the fastest source of real phrasing and the easiest to misuse, and they are the input that makes keyword research without paid tools practical at all.
Type your seed term and record what appears. Then work through the variations. Adding a letter after the seed, or a question word before it, surfaces a different set of suggestions each time. Doing this systematically rather than once produces a much richer cluster than typing the seed and copying the first five results.
Four patterns are worth collecting specifically.
- The suggestion as typed. The plain phrasing people use.
- Question forms. Suggestions beginning with how, what, why, when, or which. These map directly to headings and FAQ blocks.
- Comparison forms. Anything containing versus, or, alternatives, or instead of. These signal readers who are choosing between options.
- Year or number forms. Suggestions containing a year or a count, which usually mean the searcher wants something current or something ranked.
Here is the part that matters. Every one of these is a suggestion signal. It tells you the phrasing exists. It does not tell you that 4,000 people a month use it. Write that label next to the entry, because three weeks later, staring at a list of 180 entries, you will not remember which came from evidence and which from a guess.
A note on depth. Running the seed with no modifier gives you the most popular phrasing. Adding a letter gives you adjacent phrasing. Adding a question word gives you the question set. Four passes per seed is usually enough, because going deeper starts producing near-duplicates.
How do you classify intent in keyword research without paid tools?
You read the phrasing and the results. Difficulty scores are a proxy for competition. The actual competition is visible on the results page, and reading it is what makes keyword research without paid tools feel like investigation rather than data entry. It is also the step most people skip.
Intent classification starts with a simple question. What is the searcher trying to do?
- Informational. They want to understand something. Phrasing like “what is,” “how to,” “why does.”
Commercial investigation. They are comparing options. Phrasing with best, top, alternatives, versus. - Transactional. They want to act now. Phrasing with buy, price, download, free, template.
- Navigational. They already know the destination. Brand names, product names, login terms.
- Problem-solving. They have a symptom and want a fix. Phrasing with not working, error, fix, slow.
Then assess the competition directly. Open the results and note four things:
- Is the keyword in the ranking pages’ titles and URLs, or are they ranking by accident?
- Are the ranking pages short and shallow?
- Are the ranking pages actually on this topic, or did they match a phrase?
- Are there low-authority sites sitting in the top results?
Any yes is good news for you. You are looking at the actual pages you would need to outrank, which is more honest than reading a number between 1 and 100. Reading the actual results also shows you why the competition is what it is. You might notice that every ranking page is a listicle when the query clearly wants a walkthrough, and no difficulty score will show you that.
Google’s SEO starter guide covers how to think about the terms people actually search, and its section on understanding your audience is a useful companion to this method.
How do you find the questions nobody answers well?
This is where free research genuinely outperforms a keyword tool, because tools under-report conversational phrasing and question variants. Question research is also the part of keyword research without paid tools that most people get the most value from.
Three places to look, in order of reliability.
Question sites and forums. Search your topic on question platforms and discussion communities. Record every recurring question, and note how old the threads are.
Video comments and titles. Search suggestions on video platforms surface phrasing people use when they want something explained rather than read. Comment sections on popular videos often contain the exact question the video failed to answer.
Your own inbox and comments. The questions you have already been asked are the highest-confidence gap signals you will ever get, because there is no interpretation involved.
For each question, note the phrasing as asked, not as you would rewrite it. The reader’s phrasing is what maps to a search query and to an AI answer engine’s retrieval. Rewriting “why does my sourdough starter smell like nail polish remover” into “starter odour issues” loses the specific language you need. Question phrasing feeds directly into your headings and FAQ section, and it is the part of keyword research without paid tools that most people undervalue.
Support forums and comment threads under tutorials are rich in question phrasing, because the people writing them describe a specific problem in their own words.
What do you do when you have no volume data at all?
You work with relative demand and honest labels, and you decide on structure rather than absolute numbers. Missing volume data is the normal condition here.
Four substitutes for volume data:
Suggestion breadth. A topic with 30 suggestion variations has more surface area than one with 3. That does not give you a monthly figure, but it does rank two topics against each other.
Cross-platform presence. If the topic appears in search suggestions, in forum threads, and in video titles, the demand is spread across multiple contexts, which is stronger than repeated appearances in one place.
Recurrence over time. A question that appears in threads from three different years has staying power. One burst of activity is more likely to be a passing moment.
Your own impressions. Queries where your pages already appear, even low, are demand evidence you can trust more than any model, because the search engine itself decided your page was somewhat relevant.
What you never do is invent a number. A list with an honest “demand signal: suggestion breadth, 14 variations” column beats a list with 14 fabricated volume figures, because the second list makes you feel certain about something you do not know.
The impression substitute is the strongest one available, because it is the only demand evidence in the process that comes from the search engine rather than from your interpretation. Queries where you appear well below the first page, with real impressions, are the cheapest wins available, and closing them means improving a page you already own. Search Console exposes this at the page level, and its performance report documentation explains which query and position figures you can rely on.
How do you build long-tail phrases from what you collected?
Long-tail phrases are the specific, longer versions of a query. They are where free research shines, because they come directly from real phrasing. Most of the value in keyword research without paid tools ends up living in this part of the list.
Build them by combining what you found. A question pattern plus a specific situation produces a long-tail phrase. “How do I” plus “fix a slow laptop” plus “on Windows 11” produces something closer to what people type than any single-tool suggestion.
The test for a good long-tail phrase: could you write a page that answers this completely, and would a person searching it be satisfied? If a single paragraph answers it, it belongs as a heading inside a larger page.
Group them. Three to eight long-tail phrases sharing one subject become a cluster, and a cluster is usually one strong page plus several internal links. In keyword research without paid tools, clusters are how a small site competes with a larger one.
Long-tail phrasing is also the most durable output of the method. Head terms shift with whatever is trending, and long-tail phrasing changes slowly, because it reflects how people describe a problem.
One practical rule: use the reader’s nouns, not yours. If your audience says “hanging indent” and your draft says “paragraph indentation offset,” use theirs in the heading. Matching happens in the reader’s vocabulary.
How do you score and cut the keyword list?
At the end you will have far too many entries. Everything above has been collection, and keyword research without paid tools only becomes useful at the cutting stage.
Score each candidate on five factors, weighted toward what you need:
| Factor | Question to ask |
|---|---|
| Relevance | Does this fit what my site is about? |
| Intent match | Can my site genuinely satisfy this search? |
| Ranking ease | Is the current competition thin, off-topic, or low-authority? |
| Topical fit | Does this strengthen a subject I already own? |
| Traffic quality | Will the people searching this want what I have? |
Then apply four deletions without hesitation.
Delete anything off-topic. A high-demand keyword that does not fit your site is a distraction with good numbers.
Delete anything you already cover well. If a page already answers the intent, the right action is to improve that page, not write a second one.
In keyword research without paid tools, delete anything too broad for your current standing. A head term dominated by large brands is not a realistic target for a small site. Build the cluster first, come back later.
Delete anything you cannot differentiate on. If you have nothing to add beyond what already ranks, your page has no reason to exist.
What survives is your list. It will be short, and that is the point. A short list you act on this week beats a long list you review once and file away. Cutting is where keyword research without paid tools gains its advantage, because you judge fit rather than sort by a number.
What are the common mistakes in keyword research without paid tools?
Four show up again and again.
Treating suggestions as volume. The most damaging error, because it is invisible once written down. A list of suggestion phrases presented as demand data looks credible and is not.
Chasing head terms early. A new site targeting a broad term will not rank, no matter how good the page is. This is the mistake that makes people conclude keyword research without paid tools does not work. Build the cluster around the term first.
Ignoring intent mismatches. If the results for a query are all product pages and you write an explainer, you will not rank. Check what format currently wins before deciding what to write.
Never revising the list. Queries shift, competitor pages get rewritten, and your own coverage changes. Revisit the keyword list each quarter. Most people who try keyword research without paid tools abandon it after the first pass, which is why the list never gets better.
How do you turn the list into a page?
The list is not the deliverable. The page is.
Pick one entry as the focus keyword and make sure the exact phrase appears at the start of the page title, in the URL slug, in the meta description, and in the opening line of the body. Let supporting phrases appear naturally in headings and body text. Then check your own site before publishing. If a page already answers this intent, improve that page instead of adding another. Overlap between your own pages is the most common self-inflicted ranking problem, and it is avoidable at this stage. The process for finding that overlap is set out in cannibalization check.
FAQ
Can you really do keyword research without paid tools?
Yes. Search suggestions, live result pages, community questions, and your own performance data cover the core inputs. What you lose is official search volume figures, which you replace with relative demand signals.
Are search suggestions the same as search volume?
No, and confusing the two is the most common error in keyword research without paid tools. A suggestion shows that a phrasing exists and people use it. It says nothing about how often.
How many keywords should one article target?
One primary keyword. A handful of supporting phrases can appear naturally in headings and body text, but a page targeting several unrelated keywords usually serves none of them well.
What do you do when you have no search volume data at all?
Compare topics on relative signals instead. Suggestion breadth, cross-platform presence, recurrence over time, and your own impression data all rank demand against each other without claiming a monthly figure.
How long should a keyword list be before you start writing?
If it has more than twenty entries and no scores, it is too long. A useful list is short enough that you can pick the top three without deliberation. The rest belongs in a file you review next quarter. The method fails more often from an uncut list than from missing data.
Is this method only for beginners?
No. Plenty of established sites run it because the free signals answer the questions they actually have. The method scales down to a single blog and up to a focused content cluster.
How do you know a keyword is worth targeting?
Three conditions. Real phrasing that people use, current coverage that is thin or off-intent, and genuine fit with what your site is about. If any of those is missing, the keyword is not ready.
Do you still need a paid tool for keyword research without paid tools?
It helps for staying organised when you have hundreds of candidates at once. Below that scale, the free method answers the question better, because it shows you why the competition looks the way it does rather than reducing it to a score.
Conclusion
keyword research without paid tools is not a compromise you tolerate until you can afford a subscription. For most sites it is a better starting point, because it forces you to look at phrasing, competition, and questions rather than a single modelled number. Suggestions show you the wording. Live results show you the actual competition. Communities show you the confusion that never resolved. Your own data shows you where the search engine already considers you relevant.
Collect from all four, label every entry honestly, classify intent, build long-tail phrases from real phrasing, then cut hard. The list you end up with will be smaller than a tool export, and you can act on it this week.