To do keyword research for SEO, you find the words people type when they need what you offer, check what Google already ranks for those words, and decide which page on your site should answer each one. In practice it’s eight steps: pick seed keywords, expand them into a list, label the search intent, judge difficulty, prioritize, cluster keywords that share a page, map each cluster to one URL, and turn each URL into a content brief. You can do all eight with free Google tools.
Most guides stop at step five. They hand you a spreadsheet sorted by volume and wish you luck, which leaves the hardest decisions (what to merge, which URL owns what, what the writer actually needs to know) entirely on your plate. That’s backwards. A list nobody turns into pages ranks for nothing.
This page is the hub for every keyword research guide on the Skyranko blog. Each step gets a short summary here and a link to the post that covers it in depth.
The Keyword Research Process at a Glance
Here’s the whole keyword research process on one screen. Skim it, then read the steps that match where you’re stuck.
| Step | What you do | Output | Free tool that helps | Deep guide |
|---|---|---|---|---|
| 1. Seeds | List 5 to 15 broad terms for what you sell or cover | Seed list | Your own service list, Search Console | Seed keywords |
| 2. Expand | Grow each seed into specific phrases and questions | Raw keyword list | Autocomplete, Keyword Planner, Search Console | Keyword gap analysis |
| 3. Intent | Label what each searcher wants | Intent column | The live results page | Search intent types |
| 4. Difficulty | Judge whether your site can compete | Difficulty notes | The live results page, any KD score | Keyword difficulty |
| 5. Prioritize | Score value, effort and fit | Ranked list | Google Trends for seasonality | Keyword research template |
| 6. Cluster | Group keywords one page can win together | Clusters | SERP overlap checks | SERP based clustering |
| 7. Map | Give every cluster exactly one URL | Keyword map | Search Console, your sitemap | Keyword mapping |
| 8. Brief | Turn each mapped URL into writing instructions | Content brief | A doc or plain text file | Content brief template |
What Changes the Process for Your Site?
The eight steps never change. How you run them does, and four things decide that before you open a single tool.
- Site age. A site with months of Search Console data starts from its own queries. A brand new domain starts from seeds and competitors, because it has no data of its own yet.
- Business model. A blog, a store, a local service business and a SaaS product want different keyword types. The step order holds, but the targets look nothing alike.
- Budget. Free Google tools cover every step. Paid tools mostly save time on expansion and competitor checks; they don’t make better decisions for you.
- List size. Under 200 keywords, you can cluster and map by hand. Above that, you’ll want a clustering tool or a script.
If your site already gets impressions, start with the branch for existing sites below. If it doesn’t, follow the main path as written.
How to Do Keyword Research for SEO, Step by Step
This is the main path. It assumes a small business or blog with a working site and a clear offer, which describes most people who ask me how to do keyword research.
Step 1: Start With Seed Keywords
Seeds are the short, broad terms that describe what you sell or cover, like “roof repair” or “project management software”. You rarely try to rank for a seed itself. You use it as the input that grows into hundreds of specific phrases.
My rule is simple. Every seed should map to something you actually sell or a subject you genuinely plan to own, because seeds that drift away from your offer grow into long lists of traffic that never converts and posts you’ll later regret publishing. The full method, including where seeds come from when you have no data, sits in my guide to seed keywords.
Step 2: Expand Seeds Into a Keyword List
Now you grow each seed. Type it into Google and read the autocomplete suggestions, which Google’s own autocomplete help page says “reflect real searches that have been done on Google”. Then scroll to People Also Ask and the related searches at the bottom of the results.
Next, run the seeds through Google Keyword Planner. It sits inside Google Ads, costs nothing to use, and asks you to finish account setup with billing details before keyword ideas unlock. Its volumes are rounded estimates built from 12 months of history, and very low volume keywords don’t show at all, according to Google’s Keyword Planner help. So treat the numbers as direction, not truth.
The third expansion source is your competitors, and I show how to pick the right ones and pull their lists in how to find competitor keywords. A keyword gap analysis shows terms they rank for and you don’t, which is the fastest way I know to find demand a market has already proven.
Step 3: Label the Search Intent
Every keyword gets one of four labels: informational, navigational, commercial or transactional. Don’t guess from the words. Search the keyword and look at what ranks, because the results page tells you what Google thinks the searcher wants.
In my experience, this is where most wasted content comes from: a buyer-intent query answered with a 2,000-word blog post, or a beginner’s question answered with a sales page, and neither one ranks however well it’s written. My guide to search intent types covers mixed-intent results pages and how to pick a page type for each one.
Step 4: Judge Keyword Difficulty Honestly
Keyword difficulty is a 0 to 100 score tools use to estimate how hard a first-page ranking will be. Each tool builds it from different data, so one keyword can look easy in one tool and hard in another.
I use the score as a first filter. Nothing more.
The real check is opening the results page and asking whether sites like mine rank there. When I see a top ten full of national brands, government sites and publishers with thousands of referring domains, I move on and look for the narrower phrase underneath it. The full breakdown of how Ahrefs, Semrush and Moz each calculate it lives in my post on keyword difficulty.
Step 5: Prioritize by Value, Not Volume
Volume is the column people sort by first. It misleads them most. I rank keywords on three things: how close the searcher is to buying, how hard the keyword is for my site, and how well it fits what we actually sell.
Two traps sit here. The first is skipping a keyword because a tool shows zero searches, even though real people type it; my post on zero volume keywords explains how to check demand the tools can’t see.
The second is ignoring seasonality. Google Trends helps here, but remember that it shows relative interest scaled from 0 to 100, not search counts, as Google’s Trends help center explains. Two terms can both score 100 and still differ wildly in real searches.
If you want a sheet to score all this, my free keyword research template has every column filled with examples.
Step 6: Cluster Keywords That Share a Page
Clustering groups keywords that one page can rank for together. “How to fix a leaking faucet” and “leaky faucet repair” almost certainly belong on the same URL. Two posts would just split your effort, and Google’s language matching systems are built to see past that kind of wording difference, which is the core idea behind semantic SEO.
The most reliable method compares the pages Google already ranks. If two keywords share several of the same top ten URLs, Google treats them as the same need. My guide to SERP based keyword clustering walks through the overlap math, and the comparison of keyword clustering tools helps if your list is too long to do by hand.
Step 7: Map Every Cluster to One URL
Keyword mapping assigns each cluster to exactly one page on your site. One keyword. One owner. The map tells you which page to update, which page to create and which keywords are hands off.
Skip this and you’ll eventually publish two pages chasing the same search. That’s keyword cannibalization, and it’s far cheaper to prevent than to fix. The four questions I use to place every keyword are in my post on keyword mapping.
Step 8: Turn Each Mapped Page Into a Brief
Research ends with a brief. Not a spreadsheet. A content brief tells the writer which query the page answers, who the reader is, what the first paragraph must say and which facts need a source.
Honestly, this is the step I’d least want to skip. A strong writer with a vague brief writes a nice article that misses the query. Mine is free to copy in the content brief template.
Branch 1: Your Site Already Has Search Console Data
If your site gets impressions, flip the order of the first two steps. Start with the queries Google already shows you for, then add seeds to fill the gaps.
Search Console keeps 16 months of data, according to Google’s Analytics help documentation, and every query in it is real. Rare queries get hidden for privacy. Still beats any estimate.
I look for query groups with impressions and no dedicated page. Those are new pages waiting to be written.
If a query already has a matching page, improve that page instead of writing a new one. If the query lands on the wrong page, note it for the mapping step. If no page fits, it goes into the new content list. The full workflow, filters and all, is in my guide on how to use Google Search Console for keyword research.
Branch 2: A Brand New Site With No Data
A new domain has no queries to mine, so it leans on seeds, competitors and the results page. Authority is near zero too. That hits step 4 hardest.
My default for new sites is to chase specific long-tail questions and low-competition phrases first, even when the volume looks tiny, since a handful of pages that actually rank will teach you more about your market than fifty pages stuck on page four. Head terms come later, once the site has pages that rank and links that point at them. I’d also build one tight cluster at a time rather than scattering posts across ten topics, so the site looks like it knows one subject well.
Which Version of the Process Fits Your Business?
The steps hold for every site. The keywords you end up with don’t. Here’s how the business model changes the targets, with the guide that goes deeper for each.
| Business type | What changes | Typical target keywords | Deep guide |
|---|---|---|---|
| Blog or publisher | Mostly informational intent; questions beat head terms | “how to”, “why does”, “what is” questions | Keyword research for blog posts |
| Online store | Each keyword needs the right page level: category, product or guide | Category terms, product names, “best” and “vs” | Keyword research for ecommerce |
| Local service business | Most searches carry no town name; tools undercount local demand | Service plus town, “near me”, implicit local terms | Keyword research for local SEO |
| SaaS product | Small, high-intent keywords close to a signup pay first | Alternatives, comparisons, integrations, use cases | Keyword research for SaaS |
Which Google Tools Help at Each Step?
You don’t need a paid subscription to learn how to do keyword research. Here’s what Google itself gives you at no cost, and what each tool can’t tell you.
| Google tool | Best for | Watch out for |
|---|---|---|
| Google autocomplete | Expanding seeds into real phrasing | Predictions shift with location and trends; no volume |
| People Also Ask | Question keywords for blog posts and FAQs | Answers repeat across many queries |
| Google Keyword Planner | Rough volume and new ideas | Needs a Google Ads account with billing set up; rounded estimates |
| Google Search Console | Real queries your site already appears for | Only shows queries you already rank for; rare queries hidden |
| Google Trends | Seasonality and comparing two terms | Relative 0 to 100 scores, not search counts |
| The results page itself | Intent, difficulty and clustering checks | Results vary by location, so check from the right place |
Paid tools earn their price on big projects, mostly for competitor data and difficulty scores at scale. For a small business doing its first round of research, I’d start free and only pay once you know what you’re missing. My roundup of free keyword research tools lists the free tiers beyond Google’s, with their limits.
Edge Cases That Change the Steps
- Multilingual sites. Build seeds and intent labels per language, from native search data. Translated keyword lists often miss how people actually search in that language.
- Brand new product categories. Nobody searches for a name they’ve never heard. Research the problem your product solves, not the product.
- Very small local markets. Volume tools may show zero for everything. Lean on your Google Business Profile data, your categories and the local results page instead.
- Big sites with hundreds of existing pages. Do the mapping step before expansion. You need to know what you already own before adding anything new.
Decision Matrix: Where to Start
| If your site is | And your list is | Then start here |
|---|---|---|
| New, no Search Console data | Under 200 keywords | Seeds, expansion, manual clustering |
| New, no Search Console data | Over 200 keywords | Seeds, expansion, a clustering tool |
| Established, with impressions | Any size | Search Console queries first, then seeds for gaps |
| Established, with many overlapping posts | Any size | Mapping and a cannibalization check before new research |
| Local business | Any size | Profile categories and local results, then expansion |
Common Mistakes I See in Keyword Research
The biggest? Treating research as a one-time job. Search behavior shifts, competitors publish, and your own site starts ranking for things you never planned. I rerun the Search Console part every few months on any site I manage.
The second is chasing volume over fit. A keyword with big numbers that doesn’t match what you sell brings visitors who leave. The third is skipping the results page. Every tool number is an estimate, and the live results page is the only place you see what Google actually rewards.
The last one is sneaky. People build a perfect list, then let two writers target the same cluster from different angles. That’s why steps 7 and 8 exist.
When Does It Make Sense to Get Help?
DIY works fine for a small site with a clear offer. It gets harder when you sell dozens of overlapping services, run several locations or need hundreds of keywords mapped across an existing site without cannibalizing it.
That’s the work my team does through our keyword research services, priced at $99 to $499 per project, and you can try it on 5 keywords free first. If you’re comparing a keyword research company or two before deciding, my roundup of the best keyword research service providers lays out what to check.
Frequently Asked Questions
How Long Does Keyword Research Take?
For a small business site with one clear service, a first pass takes me a few focused hours. A store with hundreds of products or a site with an existing archive takes much longer, because the mapping step grows with every page you already have.
Can I Do Keyword Research for Free?
Yes. Autocomplete, People Also Ask, Keyword Planner, Search Console and Google Trends cover every step in this guide. Paid tools save time on large lists and competitor data, but they don’t replace checking the results page yourself.
How Many Keywords Should One Page Target?
One primary keyword plus the close variants that share its results page. If two keywords bring up mostly different top ten results, they usually need separate pages. Clustering in step 6 settles this for each group.
How Often Should I Redo Keyword Research?
I’d refresh Search Console findings every three to six months and do a full pass about once a year. Do it sooner after a big site change, a new service line or a drop in traffic you can’t explain.
Is Keyword Research Still Worth It With AI Search?
Yes. People still type and speak queries, and the words they use still tell you what they need. Matching pages to those needs is the same job, whether the answer shows up as a blue link or inside an AI summary. Google’s own SEO Starter Guide still tells site owners to think about the words a user might search for.
Last updated: October 2026 by Mizanur Rahman
