Keyword research with free tools — a step-by-step guide
Most small businesses pick the words for their website or blog without ever asking how a customer actually types the query into Google. Instead they use the company's own professional language — because that's the service name, that's what the brochure says, that's what a competitor already wrote. The result is a page that exists, sometimes a well-written one, that nobody searches for — because the real searcher uses completely different words.
Fixing this does not need a paid tool, a monthly subscription, or a marketing course. Google itself shows you the real words, for free, in at least three places. Here is how to use each one.
Why "I think this is how they search" doesn't hold up
The symptom is familiar: a service page goes live with corporate language in the heading — "comprehensive digital solutions", "full-spectrum web strategy". Traffic either doesn't come, or comes through completely unrelated words.
The mechanism is simple: a real searcher names the problem, not the service. They don't type "full-spectrum web strategy" — they type "why isn't my site selling". They don't type "order an SEO audit" — they type "why can't I find my site on Google". The wider that gap between your language and theirs, the fewer people ever reach the page, even when the topic is right.
The fix is to stop guessing and start measuring. The three sources below cost nothing and do exactly that.
Google's own autocomplete — in the searcher's own words
How it works
The suggestions that appear as you type are not random. By Google's own explanation, they draw on popular and trending searches as well as the individual searcher's own history (Google's help page: how autocomplete predictions are generated). Every word on that list was typed by someone real — not invented by you, written by someone else.
To cut down on personalisation, run this in an incognito window — otherwise the list reflects your own search history, not your target customer's.
The alphabet method
Type your core term, then add letters one at a time after it: "web design a", "web design b", and so on through the alphabet. Each letter surfaces a different set of suggestions. Repeat the same exercise adding "how to", "why", "which", "cost of" in front — you'll get queries with a completely different intent, such as "how to order a website" or "cost of a website". In fifteen to twenty minutes of manual work you can collect fifty to sixty real phrasings — not far off what a paid tool returns, just not automated.
Log every phrase you find in a simple table as you go: the query itself, its likely intent (researching versus ready to order), and which page it belongs to. Skip that note and a week later it's hard to remember which word was meant for the blog and which for the service page — so the same word ends up on both, and the two pages start competing with each other.
The free list at the bottom of the results page
At the very bottom of a search result sits a block of related searches, and inside many result pages sit expandable "people also ask" questions. Both work on the same logic: every word or question shown there was already typed by a real user, and pages already answer it — meaning the topic is genuinely searched, not something you assumed.
Type a query, scan both blocks, then take any new term from them and search again. After three or four rounds the same words start repeating — that's the point where the topic is exhausted. Turn the collected list into actual subheadings (##/###) rather than random paragraphs — every question deserves its own section with its own answer.
Your own site's data — Search Console
Both sources above still know nothing about you personally. Search Console already holds the real queries bringing people to your site — the Queries tab in the Performance report groups clicks and impressions by exactly what the user typed (official explanation). If the site is already live, this is the first place to look — not someone else's guess, your own data.
One caveat: a query on that list won't necessarily show your site if you search it yourself — Google explains that results depend on the searcher's time, location, device and recent history. So "why can't I see myself for this query" is usually personalisation, not an error.
Finding the near-misses
Look for queries with high impressions but a position between 11 and 20. Google already treats you as close to the topic but hasn't pushed you to page one. Strengthening the existing page for that exact term — working it into the heading and the opening paragraph — usually produces results faster than starting a new page from zero.
For instance, if "website design Baku" pulls in hundreds of impressions but sits at position 14, that tells you the page has the right topic — the text just doesn't spell out that exact phrase clearly enough. Sharpening the heading, the opening paragraph and one subheading around it is cheaper and faster than launching a new page from scratch.
How Azerbaijani search behaviour differs
There's a local reality worth knowing: a share of commercial search in Azerbaijan happens in Russian, particularly in Baku and the larger cities. When you check "hazır sayt" or "sayt sifarişi", run the same method (autocomplete, Search Console) against "разработка сайта" too — these are two separate audiences through two separate doors, and one does not stand in for the other.
The second difference is spelling. Some searchers, out of habit or keyboard limits, type Azerbaijani letters (ə/ş/ç) in their plain-Latin form. That doesn't mean the page itself should drop correct spelling — it shouldn't — but when collecting queries, check both the correct and the simplified spelling, because Google treats them as distinct strings.
What to do with the finished list
The collected list can look overwhelming. Split it into three groups: words for the service page, words for blog content, and words that fit nowhere — drop the last group entirely. Every word should live in one place, not get crammed onto a single page.
Collecting the list once and forgetting about it doesn't hold up for long. Seasonal queries (pre-holiday searches, the start of the school year) and new phrasing from competitors age the list quickly — treat the half hour you spent on this as a quarterly habit, not a one-time task.
The best keyword list never comes out of your own head — it comes out of a search box that already exists.
All of this can be done by hand, but it takes time, and the technical side — page speed, mobile rendering, indexing — is a separate job. Our search visibility service handles both halves together — picking the words and getting the page technically ready to carry them — so the list you collect doesn't end up on a page that can't act on it.
