How to Do Keyword Research (Step by Step)
A keyword research process that starts from real questions, not a tool export: find terms, judge them, group them by intent, and hand them to a writer.
Keyword research is finding the exact words your buyers type into search and AI assistants, then judging which ones are worth building a page around. The process: gather real questions, check demand and difficulty, group by intent, and map each group to one page.
That last step is one piece of a bigger SEO content strategy: the plan that decides which pages get written, in what order, and why.
On this page
Where do keyword ideas actually come from?
Four free sources beat a blank search box. Google Search Console shows the exact queries your site already gets impressions for, including ones you rank on page three for and never noticed. Autocomplete and the "People also ask" box show what a real searcher typed next. Your own sales calls and support tickets show the words a buyer used before they ever searched. Competitor pages, read for the questions they answer rather than copied, round out the list.
A paid tool adds two things free sources cannot: an estimate of how many people search a term each month, and a fast way to see every keyword a competitor's page ranks for. Neither replaces the first four sources — it filters and prioritizes what they surface.
Do not skip the free sources even once you have a paid tool. A tool's index is built from clickstream and search-engine data it can see; it cannot see the words your own buyers used on a call yesterday, and those words often convert better than anything a keyword database suggests, because they came from someone who actually bought.
Search Console is worth a second look here. Most teams check it for rank, not for keyword discovery, and miss that the "queries" tab under any page is a free list of every phrase Google already associates with that URL — including ones the page was never written for. A page ranking on page two for a term it never targeted is a signal that term deserves its own page, or at least a subheading.
Pull your existing queries
Search Console → Performance → filter by low position, sort by impressions. These are terms you are already halfway visible for.
Expand with autocomplete
Type your seed term into search and note every suggestion and every "People also ask" question that appears.
Read three competitor pages
List every subheading they use to answer the topic. Subheadings are usually the questions they researched.
Ask sales or support for the actual words
The phrasing a prospect uses on a call is rarely the phrasing a keyword tool suggests, and it converts better when you match it.
How do you judge which keywords are worth targeting?
By three numbers together, never one alone: search volume, difficulty, and how directly the term connects to something you sell. A high-volume term with no buying intent brings traffic that never converts; a low-volume term with strong intent can be worth more per visitor than either.
| Signal | What it tells you | What it does not tell you |
|---|---|---|
| Search volume | Roughly how many people search the term monthly | Whether they want to buy, learn, or compare |
| Keyword difficulty | How strong the current top pages' backlinks are | Whether you can write a better, more complete answer |
| Intent match | Whether the term matches a page you can honestly write | Nothing — this is the one that decides if it is worth it |
Weight intent match highest. A term with modest volume that matches exactly what a page on your site should say is a better target than a bigger term that would force the page to be about something it is not.
Difficulty scores also vary between tools, because each one calculates them from its own backlink index and its own formula. A term scored as easy in one tool and hard in another has not changed — the tools disagree, which is normal, and neither number should be the only reason you skip a keyword. Read the actual top-ranking pages instead: if none of them fully answers the question, the term is more open than any difficulty score suggests, whatever the number says.
What tools do you actually need?
Fewer than most keyword-research guides suggest. A free Search Console account, plus one paid tool for volume estimates and competitor gaps, covers the research for most sites. Semrush and Ahrefs are the two most common choices, and the honest answer is that either one does keyword research well enough that the difference rarely decides the outcome — your process does.
Where a paid tool earns its cost is competitor gap analysis: seeing every keyword a rival page ranks for that yours does not. Running that report against your two or three closest competitors, then filtering for terms with clear buying intent, usually surfaces pages worth writing faster than starting from a blank list.
How do you group keywords by intent?
By what the searcher is trying to do, not by how the words look. "Best CRM software" and "CRM pricing" look unrelated but both signal someone comparing options; "what is a CRM" and "CRM meaning" look different but both signal someone still learning the term. Reading the current top results for a term is the fastest way to check: if they are all comparison posts, the intent is comparison, whatever the phrase suggests. Search intent covers the four common types and how to spot each one from the results page alone.
Once terms are grouped, each group becomes one page. A group that splits neatly into two different intents — one informational, one transactional — is two pages, not one page trying to serve both readers.
Grouping also catches near-duplicate targets before they turn into competing pages. "How to do keyword research" and "keyword research process" read as different phrases but describe the same search, and a site with a page for each will usually see Google pick one to rank and leave the other stuck — or rank both for a fraction of what one page could have earned alone. Merge the pair before either gets written, not after.
How do you turn a list into pages that get written?
Through a brief, not a spreadsheet handed to a writer. A keyword list tells you what to target; it does not tell a writer what the page should say, which sources to check, or which internal links it owes. Writing an SEO content brief turns each keyword group into the document a writer actually works from — target keyword, intent, word band, required links, and the questions the page has to answer.
Track results by group, not by individual keyword. A page rarely ranks for only its target term; it usually picks up a cluster of related variants, and measuring content ROI by page or by group is closer to how search traffic actually arrives than checking rank for one phrase.
How often should you redo the research?
Every six to twelve months for an active site, and immediately after a major product or pricing change. What people ask does not stay fixed — new competitors enter, terminology shifts, and AI-assisted search adds question phrasings that did not exist in last year's list. A keyword map that has not been checked in a year is a plan built for a market that has since moved.
Treat the refresh as light, not a redo from scratch. Pull the last twelve months of Search Console queries, compare against the current keyword map, and add or retire terms where the gap is real.
Retire terms honestly, too. A keyword that stops getting searched, or a page that was built for a product you no longer sell, is worth dropping from the map rather than leaving as a placeholder nobody will ever write. A shorter, current keyword map beats a longer, stale one every time someone has to plan next quarter's content from it.
Questions people also ask
How many keywords should one page target?
One primary keyword and a handful of close variants that share the same search intent. If two topics need different pages to answer well, they are two keywords for two pages, not one page trying to rank for both.
Do I need a paid tool to do keyword research?
No. Search Console, autocomplete and the site search boxes of forums in your niche find real questions for free. A paid tool adds volume estimates and competitor gaps faster, which matters once you are past the first dozen pages.
What is keyword difficulty and can I trust it?
It is a tool's estimate of how hard a term is to rank for, usually based on the backlinks of the current top results. Treat it as a rough sort order, not a guarantee — it says nothing about how good an answer you can write.
Should I target keywords with zero reported search volume?
Sometimes. Long, specific phrases often show as zero volume because too few people search them individually for a tool to measure, but in aggregate they can bring real traffic, and they are usually the easiest to rank for.
Sources
- SEO Starter Guide: The Basics — Google Search Central (checked 22 September 2026)
- Keyword Research: The Beginner's Guide by Ahrefs — Ahrefs (checked 22 September 2026)
- Keyword Research: The Ultimate Guide — Moz (checked 22 September 2026)
Get a keyword map built around what you actually sell
Book a free 30-minute consultation with Ibrahem and Jeremy. We look at your site while we talk, and you leave with a prioritised plan.
Who wrote this
Ibrahem HelalCo-founder
Leads strategy and delivery.
Works on: SEO, GEO
Reviewed before publishing by Jeremy Zava, Co-founder.
- Keyword Mapping: Assigning Keywords to PagesA keyword map assigns one primary keyword to one page across your whole site, so two pages never compete for the same search. Here is how to build one.
- What Is Search Intent? A Plain DefinitionSearch intent is the reason behind a query, not the words in it. Getting it wrong is why well-written pages still fail to rank for their target term.
- How to write an SEO content briefA content brief that actually helps a writer covers the gap to fill, the questions to answer and the proof to include, not just a keyword and a word count.
- Semrush vs Ahrefs: Which One Should You Buy?Semrush and Ahrefs cover the same core jobs — keywords, backlinks, audits, rank tracking — but differ enough in depth and price to matter. Here is where.