Quick Answer
- Keyword research means mapping real search demand — not building a one-time list and filing it away.
- The repeatable process is: seed → expand → cluster → prioritize.
- Intent matters more than volume. A 200-search/month keyword with clear buyer intent beats a 10,000-search/month keyword where nobody is ready to act.
- Google Search Console is the most underused free tool in keyword research. Most teams ignore it.
- Clusters — not individual keywords — are what you plan content around.
- Cannibalization quietly kills rankings. Check for it before every new piece you publish.
- In 2026, ranking for AI search (ChatGPT, Perplexity, Gemini) requires the same keyword discipline as ranking on Google — you just add a layer of direct-answer formatting.
Keyword Research Is Demand Mapping, Not a Spreadsheet Exercise
Keyword research is the process of finding the exact words and phrases your target customers type into Google — and using that data to decide what content to build.
Most businesses treat it as a one-time task. They pull a list, hand it to a writer, and move on. The list goes stale in six months and nobody touches it again.
That's not a system. That's a guess.
Real keyword research is ongoing demand mapping. You're not just asking "what do people search for?" You're asking "where is demand growing, where is it shrinking, and what does a searcher actually want when they type that phrase?" The answers change. Your process needs to run on a cycle — quarterly at minimum, monthly if you're in a fast-moving vertical.
The firms and businesses we work with that get compounding organic results don't just "do keyword research." They have a repeatable process that updates their content plan automatically when search demand shifts.
Here's that process.
Step 1: Seed — Start with What You Actually Do
The seed stage is simple: write down every service, product, outcome, or problem your business addresses. No tools yet. Just a plain list.
For a law firm, that might be: workers' compensation, denied claims, permanent disability, temporary disability, employer retaliation. For an HVAC company: AC repair, furnace installation, emergency heating, HVAC maintenance.
Then add the problem language your customers actually use. Not the technical term — the way they describe the pain at 11pm when they're searching from their phone. "Employer won't pay my workers comp." "AC stopped working in the middle of summer." That language is where high-intent traffic hides.
Seed sources that most people skip:
- Your own sales call transcripts and intake forms
- Live chat logs (what exact words do people use?)
- Google Search Console queries you're already ranking for (more on this below)
- Review language on Google Business Profile and G2/Yelp
- Reddit, Quora, and niche forums where your customers vent
Each of these is a seed keyword waiting to be expanded.
Step 2: Expand — Turn Seeds into a Full Keyword Universe
Once you have 20–40 seed terms, you run them through tools to find every related term, question, and modifier that exists in the search ecosystem.
The tools that actually move the needle:
Google Keyword Planner — Free. Accurate volume data because it comes straight from Google Ads. Enter your seeds and it returns hundreds of related terms with average monthly searches and competition levels. The data skews toward paid search, but organic search demand mirrors it closely enough to be actionable.
Google Search Console — This is the most underused free tool in keyword research. GSC shows you every query that triggered an impression for your site over the past 16 months. Sort by impressions. You'll find keywords you're ranking on page two or three for — real demand you're already touching but not fully capturing. These are your fastest-win targets because Google already associates your domain with them.
SERP analysis — Type your seed keyword into Google and read the page. Look at the "People also ask" section. Look at the related searches at the bottom. Look at what the top three results actually cover. This tells you what Google thinks the keyword is really about — and that's the signal that matters.
Paid tools (worth it at scale): Ahrefs, Semrush, and Moz all show search volume, keyword difficulty, and who else ranks for a term. If you're managing more than 20 pages of content, the data density justifies the cost. If you're a small operation, start with Google's free tools and layer in paid access later.
After the expand phase, you'll likely have 200–1,000+ terms. That's not a content plan — that's raw material. The next step turns it into one.
Step 3: Cluster — Group by Intent, Not Just Topic
A keyword cluster is a group of related terms that share the same search intent and can be covered on a single page.
This is where most keyword research breaks down. Teams sort keywords by topic — all the "workers comp" keywords together, all the "disability" keywords together — and then wonder why they have 40 pages cannibalizing each other.
Clustering by intent is different. You ask: "If someone types this, what do they actually want to find?" The answer determines which keywords belong on the same page.
The four intent types:
- Informational — "how does workers comp work in California" / "what is keyword research." The person wants to learn. They're not buying yet.
- Commercial — "best workers comp attorney Los Angeles" / "keyword research tool comparison." They're researching options. Closer to a decision.
- Transactional — "hire workers comp lawyer free consultation" / "buy Ahrefs subscription." They're ready to act.
- Navigational — "Nordanyan Law contact" / "Ahrefs login." They already know where they want to go.
Search intent tells you why someone is searching, not just what they're searching for — and matching intent is what separates pages that rank from pages that don't.
When you cluster, you're grouping terms that share the same intent answer. "How do I do keyword research," "keyword research process," and "steps for keyword research" all belong on the same page — this one. They want the same thing. Building three separate pages for them would be a mistake.
How to cluster in practice:
- Export your expanded keyword list to a spreadsheet.
- Add an "intent" column. Label each row: info / commercial / transactional / navigational.
- Add a "topic" column. Group keywords that are really asking the same question.
- Each unique topic+intent combination = one page in your content plan.
You'll typically see your 500-keyword list collapse into 30–60 actual content opportunities. That's the right level of specificity to build a content plan around.
Step 4: Prioritize — Pick the Battles You Can Win
You can't publish 60 pieces of content at once. And even if you could, you'd want to sequence them intelligently.
Prioritize by three factors:
1. Business value. A keyword that converts to a signed client is worth more than one that drives traffic. Weight commercial and transactional intent keywords higher. For the law firm clients we work with, "workers comp attorney free consultation Los Angeles" is worth 50x more than "how does workers comp work" — even if the informational term gets 10x the searches.
2. Difficulty vs. domain authority. Keyword difficulty scores (Ahrefs calls this KD, Semrush calls it KD too) estimate how hard it is to rank for a term based on the authority of the pages already ranking. If your domain is new or low-authority, targeting KD 70+ keywords out of the gate is burning budget. Start with KD 20–40 terms where you can compete now, build authority, and climb into harder terms over time.
3. Search volume. Volume matters — but it's the third filter, not the first. A 150-search/month keyword with clear transactional intent and low difficulty can generate more revenue than a 5,000-search/month informational term that never converts.
The prioritization matrix:
Priority: Tier 1 (publish first) · Intent: Commercial or transactional · Difficulty: Low-medium (KD ≤40) · Volume: Any
Priority: Tier 2 (publish next quarter) · Intent: Informational, high business relevance · Difficulty: Medium (KD 40–60) · Volume: 500+
Priority: Tier 3 (long game) · Intent: Any · Difficulty: High (KD 60+) · Volume: High
Priority: Drop · Intent: Navigational or zero business relevance · Difficulty: Any · Volume: Any
Build your editorial calendar from Tier 1 down. Measure results at 90 days. Reprioritize.
Reading the SERP: The Check You Run Before Writing Anything
Before any piece of content goes into production, run one SERP check. Takes five minutes.
Type the target keyword into Google in a private/incognito window. Look at:
- What kind of pages rank? Blog posts, product pages, Reddit threads, YouTube videos? If video dominates, you need video or a video-embedded article — not just text.
- What length do the top results run? Check approximate word counts on the top three. If top results are all 500-word quick answers, a 3,000-word guide won't necessarily win — and might hurt dwell time.
- What do the H2s cover? Scroll without clicking and read the headers from each result. Google surfaces the sections it thinks matter most. Your article should address at least the same territory — and ideally cover something they don't.
- Are there ads? Heavy paid ads at the top signal commercial intent even if the keyword looks informational. Means buyers are there.
- What's in "People also ask"? These are your FAQ section questions, handed to you by Google.
The SERP tells you what Google already believes about a keyword. You can argue with it — but you'll lose.
Turning Clusters into a Content Plan
Once you have prioritized clusters, you build a hub-and-spoke content plan.
The hub is a pillar page — a long, comprehensive piece that covers the broad topic. Each spoke is a cluster article that goes deep on one sub-topic and links back to the hub.
For an SEO services page (services/seo), a hub-and-spoke might look like:
- Hub: What is SEO and how does it work (the broad topic)
- Spoke 1: How to do keyword research ← you're reading it
- Spoke 2: How to build topical authority with content clusters
- Spoke 3: Technical SEO checklist for 2026
- Spoke 4: How to use Google Search Console for SEO
Each spoke links back to the hub. The hub links out to each spoke. Google sees a tightly related cluster of content, interprets it as topical authority, and ranks the hub higher than a standalone page would rank on its own.
The cluster model also answers the question "how much content do I need to rank?" The answer isn't a page count — it's depth of coverage. One hub + four spokes that comprehensively cover a topic beats 15 thin pages that each cover part of it.
Avoiding Cannibalization
Keyword cannibalization happens when two or more of your own pages compete for the same keyword, splitting authority and hurting both.
It happens gradually. You publish a blog post about workers' comp claims. Six months later, a different writer publishes another one. Neither ranks as well as a single well-built page would have.
How to check for cannibalization:
In Google Search Console, filter the Performance report to a specific keyword and look at which URLs are getting impressions. If two or more URLs show up for the same query — you have cannibalization.
You can also use a site: search on Google: site:yourdomain.com "keyword phrase" to see which pages Google is indexing for a term.
How to fix it:
- If one page is clearly stronger, 301-redirect the weaker one to it.
- If both have unique value, differentiate them so they target different intent (one informational, one commercial).
- Add internal links from the weaker page to the stronger to consolidate authority.
Run a cannibalization audit before every new page you publish. Check whether you already have a page targeting that keyword. If you do, update and strengthen the existing page instead of creating a new one.
Keyword Research for AI Search in 2026
In 2026, keyword research also means optimizing for AI search engines like ChatGPT and Perplexity, which pull answers from pages that directly answer a specific question.
The good news: the process is the same. Seed, expand, cluster, prioritize. The keyword universe is identical.
What changes is the formatting layer. AI engines pull from pages that:
- Answer a specific question in the first paragraph (answer-first structure)
- Use the question as a header or near a header
- Give a direct, quotable answer before supporting it with detail
- Cover the full topic cluster on the same page or site (topical authority signals)
The pages that rank on Google page one and the pages that get cited by ChatGPT are increasingly the same pages. Build for Google's ranking algorithm using the process above. Then add the answer-first formatting layer — short direct answers before explanations, clear headers that mirror how people phrase questions, FAQ sections that cover the long-tail queries in your cluster.
That's the full system. Not two separate workflows — one workflow that satisfies both.
What a Real Keyword Research Workflow Looks Like
Here's how this process runs in a real content operation:
Week 1: Pull GSC data from the past 90 days. Export all queries with 50+ impressions. Identify keywords where you're ranking positions 4–15 (fastest wins). Add them to the cluster map.
Week 1–2: Run seed terms through Keyword Planner. Export. Add to cluster map. Label intent. Group into clusters.
Week 2: Run SERP checks on Tier 1 targets. Note competitor gaps. Finalize editorial calendar for the next 90 days.
Month 1–3: Publish against the calendar. Track rankings weekly in GSC. Flag any movement (up or down) in a shared doc.
Quarter end: Pull fresh GSC data. Add new queries. Remove terms you've now fully captured. Add new Tier 1 targets. Repeat.
This isn't a complex system. It's a repeatable loop. The compounding happens because each piece of content builds topical authority that makes the next piece rank faster.
FAQs: Keyword Research in 2026
How do I do keyword research?
Start with a seed list of terms related to what your business does. Run those seeds through Google Keyword Planner and Google Search Console to expand them into a full keyword universe. Group the results into clusters based on search intent. Prioritize clusters by business value, difficulty, and search volume. Publish content against that prioritized list on a quarterly cycle.
What's the best way to find keywords?
The best source most businesses ignore is Google Search Console. It shows you exactly what queries are already driving impressions to your site — real demand you're almost ranking for. Combine that with Keyword Planner for volume data and SERP analysis for intent signals. That combination beats any paid tool for 90% of businesses.
What is keyword intent and why does it matter?
Keyword intent is the reason behind a search. Informational intent means someone is learning. Commercial intent means they're comparing options. Transactional intent means they're ready to act. Matching your content type to the right intent is what determines whether your page ranks — because Google ranks pages that satisfy the searcher, not just pages that contain the keyword.
How many keywords should I target per page?
Target one primary keyword per page. Build the page around a cluster of 5–20 related terms that share the same intent — these will naturally appear in the content and in your headers without forcing them. Trying to rank one page for keywords with different intent fragments the user experience and usually results in ranking for none of them.
What is keyword cannibalization?
Keyword cannibalization is when two or more pages on your own site compete for the same keyword. It splits the authority that would have concentrated on one strong page. The result: both pages rank lower than a single well-built page would have. Fix it by merging or redirecting the weaker page, or by clearly differentiating intent between the two.
How often should I do keyword research?
Run a full keyword research cycle quarterly. Between cycles, monitor Google Search Console weekly for new queries gaining impressions — these are fast-moving signals that don't wait for a scheduled review. If you're in a high-velocity vertical (legal, finance, health), monthly is better.
Do I need a paid keyword tool?
Not to start. Google Keyword Planner and Google Search Console are free and genuinely powerful. The case for paid tools (Ahrefs, Semrush) becomes clear when you're managing 50+ pages of content, need competitor data at scale, or want automated rank tracking. Before that point, the free stack is sufficient.
How does keyword research work for AI search engines like ChatGPT?
The keyword universe is the same as for Google. What changes is formatting. AI engines cite pages that give a direct answer in the first paragraph, use clear question-based headers, and cover the topic cluster thoroughly. Answer-first structure combined with a strong FAQ section gives you coverage in both traditional search and AI-generated answers.
Ready to build a keyword and content system that produces measurable organic results? Book a strategy call — we'll review your current keyword coverage, identify the fastest-win gaps, and show you exactly what a content pipeline built on this process looks like for your business.