Keyword Research for Small Business: A Free, Step-by-Step Process

Most small business websites are written in the owner's vocabulary: a plumber writes about "drainage remediation", while customers type "blocked kitchen sink". Keyword research is the work of finding that difference and closing it.

This guide shows keyword research for small business using free sources: your own knowledge, the Google Search Console data you may already have, Google Trends and, if you want volume estimates, Google's Keyword Planner. It is written for owners, freelancers and the person who "does the website" on the side. You will finish with a spreadsheet that lists the phrases your customers use, notes on what each phrase needs, and one page assigned to each group of related phrases.

It does not promise that a keyword will bring traffic. It makes your pages clearer to searchers and avoids writing several pages for one need.

Short answer: Keyword research for a small business means listing the words customers use, checking what searchers actually want from each phrase, judging how crowded the results are by looking at them, and assigning each group of similar phrases to exactly one page. Free sources include your customers, Search Console, Google Trends and Keyword Planner.

Before you begin

You need a spreadsheet, thirty minutes of uninterrupted thinking and, ideally, access to Search Console. If the site is new and has no Search Console data, that is fine; the early steps rely on your customers. If you want to see your existing data, follow Set Up Google Search Console: Complete Website Owner Checklist first and expect it to take time to collect.

Allow two to four hours for a first pass. Nothing here changes a page until you decide to.

What is keyword research, and what does it not do?

A keyword is any word or phrase someone types or speaks into a search engine. Research means finding the ones that matter to your business and understanding what the searcher is trying to do. Google's own SEO starter guide frames the task this way: think about the words a user might search for to find a piece of your content, and note that people who know a lot about a topic may use different words from those who do not.

The same guide gives two cautions worth remembering. Excessively repeating the same words, even in variations, is tiring for users, and keyword stuffing violates Google's spam policies. And Google says its systems can understand how a page relates to many queries even if the page does not use the exact terms. So research is about understanding your audience, not about counting how often to repeat a phrase. Also, Google does not use the keywords meta tag, so there is no hidden place to list terms.

Step-by-step instructions

Step 1: Write seed topics from your own business

Seeds are the plain-language starting points. Open a sheet with five columns: seed, who asks, what they want, page, notes. Fill the first column from these prompts:

The ten questions customers ask you most often, in their words, taken from emails, calls, reviews and chat messages.
Each product or service you sell, named the way a stranger would name it.
Problems you solve ("leaking tap", "slow website", "tax return late").
Places you serve, and comparisons customers make ("repair or replace").

Aim for 20 to 40 seeds. Do not filter yet, and do not look at any tool; these are the most accurate phrases you will get because they came from real people.

How to confirm it worked: at least half of the seeds are phrases a customer has actually used in your presence.

Step 2: Mine the data you already have

If the site has been live for a while, Search Console's Performance report shows the queries that already produced impressions. Google defines the queries dimension as grouping data by the search query users typed. Group by Queries, then look for three kinds of rows:

Queries with many impressions and few clicks, which can show a page that is visible but not persuasive, or only loosely relevant.
Queries you did not plan for, which can reveal a new seed.
Queries where average position is lower down, which can suggest topics where you are close but not strong.

Remember how Google defines average position: in the chart it is the average position of the topmost result from your site, and in the table it is the average for the specific URL or grouping in each row. That makes it an indicator, not a ranking you can lock in. For a tour of the report, see Google Search Console Performance Report: Queries, Pages and CTR, and GA4 and Search Console Integration if you want behaviour data alongside it.

How to confirm it worked: your sheet has a second block of phrases labelled "from Search Console", each with impressions and clicks copied in.

Step 3: Expand seeds with free tools

Use these without paying anything:

Source	What it gives you	Caveat
Google Search suggestions and "related searches" you see yourself	Real phrasings, and questions people ask	They vary by place and person; record what you see, with the date
Google Trends	Relative interest over time, and comparisons between phrases	Scores are scaled from 0 to 100 relative to all searches, not raw counts; low-volume terms appear as 0
Keyword Planner	Keyword ideas and estimates of monthly searches	Requires a Google Ads account with billing set up; built for advertising
Customer forums and your own site search box	The phrases people use when they are not trying to impress	Anecdotal; collect many examples

Google's Trends help page explains that Trends uses a sample, divides figures by overall search volume for the place and period, and then scales them from 0 to 100. A score of 100 therefore means the highest relative popularity in the range chosen, not a particular number of searches. It also adds statistical noise that is most visible for low-interest queries, so do not read much into small differences. Use Trends as a rough signal of relative interest, for example to see whether a phrase is more popular than an alternative wording, and weigh it against other sources: Google notes that Trends is not a scientific poll and that a spike shows interest, not that a topic is winning.

Keyword Planner describes itself as a tool for finding keywords for Search campaigns and estimating how often they are searched and what they cost to target. Its numbers are advertising estimates. For a small site they are still useful as a rough order of magnitude, as long as you treat them as such.

How to confirm it worked: your seed list has grown into 100 to 200 candidate phrases, each tagged with where it came from.

Step 4: Group phrases by intent

Many phrases mean the same thing, and some that look similar mean different things. "Boiler service" and "boiler servicing cost" might both fit one page. "Boiler not working" is a person with a problem now, and "best boiler brands" is a researcher. Label each phrase with one of these intents:

Know: wants information ("how often should a boiler be serviced").
Compare: weighing options ("repair or replace boiler").
Do or buy: ready to act ("boiler service near me", "book a boiler service").
Find a place or business: a name or address.

Search each important phrase yourself and read the first page of results. If they are all guides and your page is a sales page, the mismatch matters. Search Intent: Match Your Page to What Searchers Want goes through this in detail.

How to confirm it worked: each phrase has an intent label, and phrases that need different page types are no longer in the same group.

Step 5: Judge difficulty by hand

Paid tools show difficulty scores, but they are one company's estimate. You can learn more from ten minutes with the results page. For each priority phrase ask:

Who is on the first page? Big publishers and national brands, or small and local sites like yours?
How thorough are those pages? If every result is a 3,000-word guide with original photos, a 300-word page will struggle to compete on helpfulness.
Do the results include maps, shopping boxes, videos or other features that change what a click means?
Could you add something they lack: local knowledge, real prices, first-hand photographs?

Record a simple rating: realistic now, realistic with real effort, or not worth it yet. Note that these are your own judgements and that Google does not publish difficulty. Also note that a phrase with few searches can still be valuable if the searchers are ready to buy.

How to confirm it worked: every priority phrase has a rating and a one-line reason.

Step 6: Map one keyword set to one page

This step is the heart of the process. Give each group of closely related phrases a single page and write down its target:

Page	Primary phrase	Supporting phrases	Intent	Action
/boiler-service/	boiler service	boiler servicing, annual boiler service	Do or buy	Improve existing page
/guides/how-often-service-boiler/	how often should a boiler be serviced	boiler service frequency	Know	Write new guide
/boiler-repair/	boiler repair	boiler not heating, boiler breakdown	Do or buy	Keep as is

Rules for the map:

One primary phrase per page. Supporting phrases are natural variations, not extra targets.
If two pages would answer the same query for the same need, merge them or differentiate them before you publish the second. Otherwise you may end up competing with yourself; Keyword Cannibalization: Find and Fix Pages Competing for One Query shows how to detect and fix that.
Match the page to the intent: a "how to" phrase gets a guide, a "book" phrase gets a service page.
Plan the site structure to match your groups with Site Architecture and URL Structure and SEO-Friendly URLs.

How to confirm it worked: no primary phrase appears twice in the Page column.

Step 7: Apply the map to titles, headings and copy

For each page, work the primary phrase in naturally: the title, the H1, the first paragraph and a subheading, as in SEO Title and Meta Description Guide and Heading Structure H1 to H6 for SEO. Add supporting phrases only where they fit the sentence; if the page sounds like a list of search terms, rewrite it.

Link the pages together using Internal Linking Strategy, using descriptive anchor text instead of repeating one exact phrase. If the business is local, LocalBusiness Schema Markup can help describe it to search engines, though no markup ensures any display.

How to confirm it worked: each updated page has the primary phrase in the title and H1 in a way you would be comfortable reading to a customer.

Step 8: Measure and revise every quarter

After a few weeks, return to Search Console and compare, page by page, the queries that now appear. New queries in the right group suggest the page is understood. Impressions with low clicks suggest the title or snippet needs work. Add what you learn to the sheet. Seasonal businesses can use their own sales records, and Trends as a rough guide, to schedule updates ahead of the busy period.

How to confirm it worked: the sheet has a dated "review" tab with changes you made and what you plan to check next time.

Final verification checklist

Your sheet lists seeds in customer language, each with a source.
Search Console queries (if available) are copied in with impressions and clicks.
Every priority phrase has an intent label.
Every priority phrase has a hand-made difficulty rating with a reason.
Each page has exactly one primary phrase and a short list of supporting phrases.
No two pages share the same primary phrase.
Titles, H1s and first paragraphs use the primary phrase naturally.
Internal links connect related pages with varied, descriptive anchor text.
A quarterly review is scheduled.

Common mistakes to avoid

Choosing phrases by volume alone. A high-volume phrase with the wrong intent sends the wrong visitors. Match the page to what the searcher wants.
Using your own jargon. If customers never say it, search engines will not see it in queries. Use their words.
Treating tool difficulty scores as facts. They are estimates from one vendor. Read the results page yourself.
Reading Trends scores as search counts. The 0 to 100 scale is relative popularity, not volume.
Targeting the same phrase on several pages. Pick one page; merge or differentiate the rest.
Repeating the phrase unnaturally. Google calls excessive repetition tiring for users. One clear mention in the right places is enough.
Doing it once. Searching habits and your own catalogue change. Revisit the sheet each quarter.

Frequently asked questions

Do I need a paid keyword tool?

No. Your customers' words, Search Console, Google Trends and Keyword Planner are enough for a small site. Paid tools add convenience and larger datasets, but they do not remove the need to read the results yourself.

How many keywords should one page target?

One primary phrase and a handful of natural variations that mean the same thing. If you find yourself targeting two different needs on one page, split it into two pages with different intents.

Is Keyword Planner really free?

Google's help page says you need a Google Ads account and must complete account setup, including billing information, to get keyword ideas. Using the tool is part of that account; check the current terms yourself.

What is a long-tail keyword?

It is a longer, more specific phrase, such as "emergency boiler repair on Sunday", which usually reflects a clearer need. The term is common industry shorthand. Such phrases are often a good fit for small sites because the need is specific.

Official references

Next step: Try the free website sample to see how your own site does on a few of these checks. This guide provides technical information, not a guarantee of rankings, traffic or index inclusion.