Local keyword research: find the words your customers actually type

Local keyword research is mostly a list: the services you sell, the areas you serve, and the way customers phrase both. You can build that list with free tools in an afternoon, and it will be more useful than most $200-a-month dashboards, because it is grounded in a business you actually understand.

The SEO industry has a habit of making this sound harder than it is. Volume scores, difficulty scores, keyword gap matrices. That machinery exists for sites competing nationally on thousands of topics. You are competing in one city, on maybe a dozen services. Here is the whole process, start to finish, with nothing to buy.

Why start with your business instead of a tool?

Because you already know most of your keywords. A tool can only confirm phrasing and fill gaps; it cannot tell you what you sell. So before you open anything, draw a grid: every service you offer down one side, every area you serve across the top. A plumber with 8 services covering 5 areas just mapped 40 keyword targets in ten minutes. That grid is the backbone of everything else in this article.

The one trap to watch for is your own vocabulary. You say "hardscaping", the buyer types "interlock driveway". You say "HVAC diagnostics", the buyer types "furnace not working". Write each service the way a customer would describe it over the phone, not the way it appears on your invoice. When RMCM builds a site, the page list comes from this exact grid before any tool opens, and half the value is catching the jargon a business uses that its customers never do.

How do you find how people actually phrase it?

Type your services into Google, slowly, and read what it gives back. Three parts of the results page are real query data: autocomplete (the suggestions that appear as you type), the People Also Ask box (the expandable questions mid-page), and the related searches at the bottom. None of them are guesses. They reflect what people in your market have actually typed.

The technique takes minutes. Type a service plus your city and pause before hitting enter; the dropdown shows the most common completions. Then add a space and a letter, "plumber toronto a", "plumber toronto b", and watch new phrases surface. Do this for each row of your grid and you will collect dozens of real phrasings, including ones no tool would have suggested, because Google says 15% of daily searches have never been seen before.

Notice what the modifiers are telling you. "Emergency" and "open now" mean urgency. "Cost" means comparison shopping. "Reviews" means trust-checking. Each modifier is a page section, an FAQ entry, or sometimes a whole page you have not written yet.

What does Search Console tell you that tools can't?

The exact queries your site already appears for, straight from Google, free. Google Search Console (Google's free reporting tool for site owners) has a Performance report listing every query where your site showed up, how many people saw it, and how many clicked. A paid tool estimates this. Search Console knows.

Three things are worth pulling out of it. First, queries with lots of impressions but few clicks: you are visible but not compelling, usually a weak page title. Second, queries sitting at positions 5 through 15: these are one solid content improvement away from real traffic, and nothing else in SEO pays back faster. Third, the accidents: queries you rank for without trying. When a page keeps drawing a question you never targeted, that is demand knocking. I check this report with every client, and the accidental rankings have produced more new page ideas than any brainstorm.

Which keywords deserve a page: commercial or informational?

Chase commercial intent first. Intent is the reason behind a search: commercial means the person wants to hire or buy ("emergency plumber toronto"), informational means they want to learn ("how to fix a leaky faucet"). The first has a burst pipe and a credit card. The second is trying to avoid calling you, at least today.

For a local business the commercial phrases are worth far more per search, and they move fast. Google's own research found 76% of people who search for something nearby on their phone visit a business within a day (Think with Google). Informational phrases still have a job, they feed blog content and AI answers, but your service and area pages come first. The chart plots the difference; toggle each group on and off.

The same split shows up in every local niche: the biggest-volume phrases are usually informational, and the money is in the smaller, specific ones. Which raises the obvious question of how small is too small.

How many keywords should each page target?

One topic per page, not one keyword per page. This is where owners overthink it, worrying whether "plumber etobicoke" and "etobicoke plumbing company" need separate pages. They do not. Ahrefs studied 3 million searches and found the average page ranking #1 also ranks in the top 10 for about 1,000 other keywords (median around 400). One good page catches every phrasing of its topic.

Do not let the tiny volumes scare you either. Ahrefs' keyword data shows 92.42% of all search terms get 10 or fewer searches a month. Almost every local query is long tail (a longer, specific phrase with low individual volume) because adding a place name makes it specific. Ten searches a month for "drain cleaning etobicoke" is not a weak keyword. It is ten people a month with a blocked drain in your service area.

So the mapping rule is simple: one page per service, one page per area you seriously serve, each phrased in customer language. That is the whole architecture, and it is why the services-times-areas grid from step one matters more than any tool output. If you are unsure how many pages that adds up to, I broke that down in how many pages a website needs. The one warning: area pages need real local substance. Ten copy-pasted town pages with the city name swapped is a doorway-page pattern Google actively filters.

How does AI search change the words people use?

Phrasing is stretching out and turning conversational, so research built only on two-word keywords is aging fast. Semrush analyzed millions of ChatGPT interactions and found prompts average 23 words versus 3.4 words for a Google search, and roughly 70% of them fit no traditional intent category. People are not typing "plumber toronto" into an AI assistant. They are typing "my water heater is leaking from the bottom and the landlord is away, who do I call in Etobicoke and what will it cost".

That behavior went from fringe to mainstream in a single year. BrightLocal's 2026 Local Consumer Review Survey found 45% of consumers now use AI tools like ChatGPT to find local businesses, up from 6% a year earlier. The practical move for your keyword list: collect the question phrasings, not just the noun phrases. The People Also Ask questions you gathered earlier are exactly what AI assistants get asked, so answer them on your pages in plain sentences. I covered the full playbook in answer engine optimization for local business.

Here is the free toolkit on one screen. Every source below costs nothing and covers a different angle of the same question: what do buyers in my area actually type?

Frequently asked questions

Keyword research is a habit, not a project

The list you build today will drift. Google sees 15% of its daily searches for the first time ever, seasons change what people need, AI keeps stretching how they phrase it, and you will add services. So put a small loop on the calendar: once a quarter, spend thirty minutes in Search Console, retype your core services into Google, and note what changed. Adjust a page title, add an FAQ, fill a gap in the grid.

That is the whole discipline. No subscription, no 4,000-row spreadsheet. A grid of what you sell and where, checked against how real people phrase it, mapped to one honest page per topic. Most of your local competitors will never do even that, which is exactly why it works. Keyword mapping is part of every SEO and local SEO project RMCM takes on. If you want to see what your site already ranks for and where the gaps are, start with a free audit.