Free research
How do I do keyword research for free?
Paid keyword tools are convenient, not compulsory. Everything you need to build a first keyword plan is published by Google and costs nothing but an afternoon.
How do I do keyword research for free?
You can do genuine keyword research using only free Google tools: Google Search Console's performance report shows the queries your site already appears for, Google's autocomplete and related searches show how people phrase things, Google Trends shows relative interest and seasonality, and Google's Keyword Planner gives volume ranges inside a free Google Ads account.
The honest limitation is precision: Keyword Planner shows bucketed ranges rather than exact monthly volumes unless the account is actively spending, and no free tool tells you how hard a keyword is to rank for. Reading the results page yourself covers that gap better than any difficulty score does.
The free stack, and why it is enough
Four free sources and one habit cover the work that paid tools package up: your own search data, Google's own suggestions, relative interest over time, and volume ranges.
Paid tools are worth money for one reason above all others: they estimate how difficult a keyword is and they show you a competitor's whole footprint in one screen. Both are conveniences. Neither is data that Google publishes, and both are models rather than measurements.
What the free tools give you is better in one important respect. Search Console shows your actual impressions and clicks from Google, not a third-party estimate of them. For a business with an existing website, that is the single most valuable keyword source in existence and almost nobody reads it properly.
The habit that ties it together is simply searching the query yourself and looking at what comes back. That takes thirty seconds and settles questions a difficulty score only guesses at.
Start with the queries you already appear for
Before looking for new keywords, read the ones Google has already matched you to. Search Console's performance report holds up to sixteen months of query data for your site: the exact phrases that showed your pages, how many times, how often they were clicked and roughly where you ranked.
The most productive pattern in that report is the near miss. Queries with plenty of impressions, an average position somewhere between eight and twenty, and a low click-through rate are phrases Google already associates with your site but has not promoted. Those are cheaper wins than anything new, because the page exists and Google already considers it relevant.
Read the full breakdown: Start with the queries you already appear forHide the full breakdown: Start with the queries you already appear for
The second pattern is the surprise. Every site appears for phrases nobody would have written on a list: a misspelling, a colloquial name for a procedure, a neighbourhood you did not think you served. Those are real customer language, captured for free, and they are usually better copy than anything a tool suggests.
If you have a Google Business Profile, its performance view adds a second free source: the searches people used to find your profile, split into terms rather than guesses. For a local business, that is the closest thing available to hearing customers describe you in their own words.
The procedure, in order
Work through these in sequence. Each step feeds the next, and skipping to the volume numbers is the most common way to waste the afternoon.
Verify the site in Google Search Console and open the performance report
Set the date range to the last sixteen months, switch to the Queries tab and export it. This is your existing footprint and the only source here that is measurement rather than estimate.
- Sort by impressions to see what Google shows you for most often.
- Filter to positions roughly eight to twenty to find pages one good edit away from the first page.
- Use the Pages tab, then filter by a specific page, to see which queries each page is actually collecting.
- Compare the last three months against the previous period to see which phrases are rising.
Harvest Google's own suggestions
Type the core phrase into Google and read the autocomplete list. Then run the search and read the questions block, the refinement chips along the top and the related searches at the foot of the page. These are Google telling you how the subject is actually phrased.
- Try the phrase with who, what, why, how, cost, near me and best in front of and behind it.
- Type the phrase followed by each letter of the alphabet to pull different suggestion sets.
- Open the questions block and let it expand, because expanding one reveals more.
- Do the same on a phone, since suggestions differ by device and location.
Check interest and seasonality in Google Trends
Trends does not give volume. It gives relative interest over time and by region, which answers different and often more useful questions: is this rising or fading, and does it spike at a particular time of year.
- Set the region to your state or metro rather than the whole country.
- Compare two or three candidate phrasings against each other to see which one people actually use.
- Look at the five-year view for seasonality, then the twelve-month view for current direction.
- Read the rising queries panel, which surfaces phrasings that are growing fast from a small base.
See the remaining steps: The procedure, in orderHide the remaining steps: The procedure, in order
Get volume ranges from Google Keyword Planner
Keyword Planner sits inside Google Ads and is free to use with an account, including one that is not running campaigns. Use Discover new keywords, set the location and language deliberately, and paste in your shortlist.
- Expect bucketed ranges rather than exact numbers: an account that is not actively spending sees bands, not precise monthly volumes.
- Use the Start with a website option on a competitor's URL to pull the themes Google associates with them.
- Read the ranges as an ordering rather than a forecast, because they are averages that hide seasonality.
- Ignore the suggested bid unless you are actually planning to advertise, in which case it is a useful signal of commercial intent.
Search each shortlisted query and read the results page
This is the step that replaces a paid difficulty score, and it is more reliable because it is the live result rather than a model of it.
- Note whether a map pack appears, which tells you the query is being treated as local.
- Note the page type that dominates: guides, service pages, directories or national publications.
- Note whether the businesses ranking look like yours, which is the honest read on whether you can join them.
- Note whether an AI summary appears, because that changes how many people click anything at all.
Group the survivors into clusters
Put every remaining phrase into a group that shares one intent. Each group becomes one page, with a hub page where a topic is big enough to need several.
- Name each cluster by the question it answers, not by the keyword with the highest volume.
- Merge phrasings that want the same page, and split any group that contains two different intents.
- Mark which clusters already have a page, which need one, and which need an existing page improved.
Write down the baseline before you change anything
Record current impressions, clicks and average position for each target query from Search Console, with the date. Without a baseline you cannot tell later whether anything worked.
- Keep it in one sheet with the page assigned to each cluster.
- Re-check at ninety days, not weekly, since query-level data is noisy over short windows.
- Watch impressions before positions, because impressions usually move first.
That is a complete first pass and it costs nothing. A paid tool makes steps two and five faster; it does not make them different.
A list is not a plan until it is grouped
The output of research should be a small number of topics with pages attached, not four hundred rows in a spreadsheet.
Grouping does two jobs at once. It stops you building near-identical pages that compete with each other, and it produces the internal linking plan for free, because the pages inside a cluster obviously belong linked to the hub and to each other.
A practical test for whether two phrases belong on the same page: search both and compare the first page of results. If roughly the same pages rank for both, Google treats them as the same question and they belong together. If the results look different, they are different intents and need different pages.
Keep clusters small enough to finish. Three well-built pages in a cluster beat twelve outlines, and an abandoned content plan is the most common outcome of an over-ambitious first pass.
Volume is not traffic, and the gap is growing
SparkToro's 2026 analysis of a Similarweb clickstream panel measured how many US Google searches end without anyone clicking a result at all.
SourceSparkToro, zero-click search analysis 2026 (Similarweb clickstream panel)
A search volume figure is a count of searches, not a promise of visits. Judge a keyword by whether the people who do click are worth having, and treat any projection that converts volume straight into traffic as fiction.
What each free source gives you, and what it withholds
None of these is complete on its own. Used together they cover most of what a paid subscription bundles.
| Item | What it gives you | The catch |
|---|---|---|
| Search Console performance report | Real queries, impressions, clicks and average position for your own site | Only shows queries you already appear for, and anonymises rare ones |
| Autocomplete and related searches | Genuine customer phrasing, including questions you would not have guessed | No volume at all, and suggestions vary by location and device |
| Google Trends | Relative interest over time, by region, plus rising queries | Relative, not absolute: it never tells you how many searches there are |
| Keyword Planner | Volume ranges, related terms and commercial signals, free with a Google Ads account | Bucketed ranges unless the account is actively spending, and grouped near-identical terms |
| The results page itself | The live answer on intent, format and who you would be competing with | Manual, one query at a time, and influenced by your own location |
| Google Business Profile performance | The search terms people used to find your profile | Local and profile-specific, so it says nothing about the wider market |
Bing Webmaster Tools adds a free keyword research view as well, which is worth a look if you want a second opinion on phrasing without paying for one.
Research only counts once it becomes pages
The most common failure here is not bad research. It is a good list that never turns into anything.
Decide up front how many pages you can genuinely produce in a quarter, then take that many clusters from the top of the list and ignore the rest until they are done. A plan sized to your actual capacity is worth more than a complete one.
Publish, wait, and then use Search Console to see which queries each new page collected. That feedback is the real second round of keyword research, and it is better than anything you could have guessed at the start.
Expect the compounding to be slow. Local SEO compounds: expect movement over weeks and months, not days, and anyone promising instant first-page rankings from a keyword list is overselling.
What free research cannot tell you
Worth knowing before you conclude a keyword is easy, because none of this is visible at zero cost.
- How difficult a keyword is to rank for. No free tool scores it, and paid difficulty scores are vendor models, not Google data.
- A competitor's full keyword footprint, which is the main thing a paid subscription actually buys you.
- Exact monthly volume, since Keyword Planner buckets ranges for accounts that are not actively spending.
- How volume splits between mobile and desktop, or between one neighbourhood and the next.
- Whether an AI summary appears for the query in other people's results, which varies by person and by day.
- What a click is worth to your business, which comes from your own booking and revenue data rather than from any search tool.
None of those gaps stops you making good decisions. They stop you making precise-sounding forecasts, which were never reliable anyway.
Related answers
What to do with the list once you have it, and where the traffic actually comes from.
What are keywords and how do I choose them?
Intent, clusters and the filter questions that decide which phrases deserve a page at all.
What are keywords and how do I choose them?How do I get more website traffic?
Where visits actually come from once the keyword plan exists, and which lever is worth pulling first.
How do I get more website traffic?What is Google Analytics?
The other half of measurement: what happens after the click, and what GA4 actually counts.
What is Google Analytics?The local SEO guide
How keyword work fits alongside listings, reviews and proximity for a business with a service area.
The local SEO guideWant a second pair of eyes on what your site already ranks for?
We will read your Search Console data and point at the near misses first, because those are the cheapest wins on the list.
Straight answers.
Is Google Keyword Planner free?
Yes, with a Google Ads account, and you do not have to run a campaign to use it. You do have to complete account setup, which is the step that puts most people off.
The trade-off is precision. Accounts that are not actively spending see bucketed volume ranges rather than exact monthly figures, and near-identical terms are grouped together under one number.
Do I need a paid tool like a rank tracker or keyword suite?
Not to start. For a single-location business, Search Console plus Google's own suggestions covers the first plan comfortably.
Paid tools earn their keep at scale: many locations, many competitors, or when you need to see a competitor's whole footprint quickly. Remember that difficulty scores and authority metrics are vendor models rather than Google data.
How do I find keywords for a brand new website?
You have no Search Console history, so start with language instead of data. Write down the phrases customers use on the phone, then check them against autocomplete, related searches and Trends.
Then look at who already ranks for those phrases and what kind of page they use. Keyword Planner's Start with a website option, pointed at an established competitor, is the fastest free way to see what themes Google associates with your market.
How long does free keyword research take?
An afternoon for a first pass on a single-location business, assuming Search Console is already verified. Verification itself can take a day if you need a developer to add a DNS record.
Expect the second pass to be quicker and better, because it uses real data from the pages you published after the first one.
Are the search volumes in free tools accurate?
They are estimates, and they are rounded. Keyword Planner reports ranges rather than exact counts for accounts not actively spending, and it groups similar terms together.
Use them to order a list, never to forecast traffic. A large share of searches now end without a click at all, so volume and visits are not the same measurement.
Should I research keywords for AI assistants differently?
Mostly you research the same needs in a more conversational form. Questions people would say out loud, with context attached, matter more than short phrases.
Beyond that, be careful what you accept as fact. How assistants choose and cite sources is not fully documented, so treat any tool claiming to measure AI keyword difficulty as an estimate of something nobody has published.
Where this comes from.
Primary documentation and published research behind the guidance on this page.
- Google Search Console Help: About Search Console (opens in a new tab)
- Google Search Central: SEO Starter Guide (opens in a new tab)
- Google Ads Help: About keyword matching options (opens in a new tab)
- Google Ads Help: About the search terms report (opens in a new tab)
- Google Search Central: Creating helpful, reliable, people-first content (opens in a new tab)Google's published self-assessment questions for content quality and experience.
Talk to the team
A short call, a look at how the business currently shows up, and a straight answer on what we would do first.