Start from the customer, not the tool
The classic mistake is opening a keyword tool and cherry-picking whatever has volume. The right starting point is elsewhere: the words your customers actually use. Reread your latest quote request emails. Replay how people describe their problem to you on the phone. Write down the exact phrasing: nobody says "aluminum window solutions", they say "replace a sliding glass door".
That customer vocabulary is gold: it is what people type into Google, and it is what AI assistants paraphrase when someone asks them for a recommendation. Turn it into a first list of 10 to 20 phrases, including the cities you serve.
Expand with free tools
- Google's suggestions. Type each phrase into Google and note what autocomplete offers, then the "People also ask" and "Related searches" blocks at the bottom of the page. These are real queries, curated by Google itself.
- Search Console. If your website is a few months old, the Performance report lists the queries you already show up for, including ones you would never have thought of. It is your most profitable seam: you are already there, you only need to climb.
- Google Keyword Planner. Free with a Google Ads account, even if you never spend a euro. Volumes come in ranges, which is more than enough to compare queries against each other.
How do you pick your battles?
Not all queries are equal, and volume is a trap. "Plumber" has huge volume and you will never rank for it against the national platforms. "Plumber and heating engineer Antibes" has a hundred times less volume and a hundred times more value: the competition is local, and the person typing it is looking for a contractor, not a definition.
Three criteria to prioritize, in order:
- Intent. Does the query signal someone ready to buy ("quote", "price", "emergency", a city name) or someone doing research? Both have their place, but the first kind pays the rent.
- The actual competition. Type the query and look at who occupies the first page. Small businesses like you: winnable. Only national websites, directories and marketplaces: pick a more specific variant.
- Volume, last. Between two winnable, well-intentioned queries, take the one with more searches. But never the other way around.
Organize: one query, one page
The rule that structures everything else: one main target query = one page. The query goes in that page's title, h1 and first paragraph, and nowhere else as a target. Close variants ("sliding glass door replacement" and "replace a sliding glass door") live on the same page: Google has understood synonyms for a long time.
In practice, a three-column table is all you need: target query, assigned page, current position. That is your editorial battle plan: every query without a page is content to create, every page without a query is content to refocus.
The costly mistakes
- Targeting generic national queries when you are a local player: months of effort to stay stuck on page 3.
- Creating one page per variant of the same query: that is the recipe for keyword cannibalization.
- Choosing queries nobody types because they use your internal jargon instead of customer vocabulary.
- Doing the exercise once and never revisiting it. Queries evolve, and Search Console shows you every month what is emerging. A quarterly pass is enough.