Keyword research is still the first thing I do on any new project. It tells me what people actually search for, in their own words. Then it tells me which of those searches I can realistically win. Get it right, and every article you write has a job to do. Get it wrong, and you create content nobody was ever looking for.
Here is the catch. The way we do keyword research has changed a lot in the last two years. Volume used to be the whole game. Today, intent comes first and volume comes second. AI answers now sit on top of many results, so a top ranking does not always earn the click it once did.
So this guide covers the modern approach, step by step. First, I will explain what keyword research is and why it still matters. Next, I will walk through the exact process I use, the metrics that count, and the tools I trust. Finally, I will show you how the AI search era changes the job. Let us start with the basics.
Every keyword hides a goal. Match that goal, or a top ranking still brings no clicks.
What is keyword research?
Keyword research is the process of finding the words and questions your audience types into search. You then measure how often they search each one, and how hard it would be to rank for it. In short, it is market research for search.
Think of it as listening at scale. Every search is a person telling Google what they want. Keyword research gathers those signals so you can build content that answers them. As a result, you stop guessing what to write. Instead, you let real demand guide you.
Here is a quick example. Say you sell running shoes. A tool might show that “best shoes for flat feet” gets far more searches than “flat feet trainers”. Now you know the exact phrase to use, and you did not have to guess.
It also does a second job that people often miss. Good keyword research maps demand to your business. Some searches signal a buyer. Others signal a browser. When you know the difference, you can spend your effort where it pays off.
Why does keyword research still matter in 2026?
Some people claim keyword research is dead. They are wrong, but I understand why they say it. The search results page looks very different now. AI Overviews answer many questions directly. Because of that, plenty of searches end without a click at all.
Studies suggest that well over half of Google searches now end in zero clicks. That sounds like bad news. However, it only kills one lazy habit: chasing huge, vague keywords for raw traffic. The keywords that bring customers are as valuable as ever.
There is a second shift too. AI answer engines like ChatGPT and Perplexity now send real visitors. To earn a mention there, you still need to know the questions people ask. That is keyword research, just pointed at a new surface. I cover that surface in detail in my guide to generative engine optimization.
So keyword research has not died. Instead, it has grown up. The goal is no longer traffic for its own sake. The goal is the right visitor, from search or from an AI answer, on a page that fits what they want.
How has search intent changed keyword research?
Search intent is the reason behind a search. It is the single most important idea in modern keyword research. Google now works hard to understand the goal behind each query. Therefore, you have to understand it too.
The four types of search intent
Most searches fall into one of four buckets. First, informational searches want to learn something, like “how to do keyword research”. Second, navigational searches want a specific place, like a brand name. Third, commercial searches compare options, like “best keyword tools”. Fourth, transactional searches are ready to act, like “hire an SEO agency”.
Each type needs a different page. For example, an informational query wants a guide, not a product page. A transactional query wants a clear path to buy, not a long essay. Match the format to the intent, and you give yourself a real chance to rank.
Why intent now beats volume
Here is the change that matters most. In the past, we sorted keywords by volume and picked the biggest. Today, that approach fails often. A huge keyword with the wrong intent brings visitors who bounce straight off.
So I flip the order. First, I ask what the searcher wants. Then I ask whether my page can be the best answer. Only after that do I look at volume. Because of this, a smaller keyword with perfect intent often beats a giant one with none.
You can read intent straight from the results page. Search your keyword and study the top ten. If they are all guides, Google has decided the intent is informational. If they are all product pages, it is transactional. In other words, the current results tell you exactly what to build.
The fastest intent check Type your keyword into Google and look at what already ranks. Those pages are your brief. If you cannot picture your page fitting naturally among them, the intent does not match, and you should move on.
The keyword research process, step by step
Now for the method. I follow the same five steps every time, in the same order. Many people skip ahead to volume, and that is the classic mistake. Follow the order below, and your list stays useful.
The keyword research process in five steps. The order matters more than any single tool.
Step 1: Brainstorm seed keywords
Start with your own head, not a tool. Write down the broad topics your customers care about. These are your seed keywords. For an SEO agency, seeds might be “SEO”, “link building”, and “keyword research” itself.
Keep this list short and simple. You only need a handful of seeds to begin. After that, the tools will expand each one into hundreds of real searches. Above all, make sure every seed connects to your actual services.
Step 2: Expand your seeds with tools
Next, feed each seed into a keyword tool. The tool returns related searches, along with rough volume and difficulty. Suddenly, one seed becomes a long list of specific queries. I will name the exact tools I use later in this guide.
Do not stop at the tools, though. Google itself is a goldmine. Look at the autocomplete suggestions as you type. Then check the “People also ask” box and the related searches at the bottom. These come straight from real user behaviour, so they are pure gold for keyword research.
Step 3: Judge each keyword
Now you have a big list. Most of it is noise, so you have to filter. For each keyword, I check three things quickly: the intent, the volume, and the difficulty. If the intent does not match my business, I cut it right away.
This step is where discipline pays off. It is tempting to keep every keyword with a big number. However, a big number with the wrong intent wastes weeks. Be strict here, and the rest of the work gets easier.
Step 4: Prioritise your list
After filtering, you still have too many keywords. So you rank them. I score each one on value, effort, and relevance. A keyword that is high value and low effort goes to the top of the list.
New sites should lean into easier keywords first. Because you have little authority yet, giant terms are out of reach for now. Win the smaller ones, build some trust, and the bigger terms open up later. In short, start where you can actually score.
Step 5: Map keywords to content
Finally, give every keyword a home. Each primary keyword should map to one page, and one page only. This prevents a problem called keyword cannibalisation, where two of your pages fight each other for the same term.
I usually group related keywords into clusters. One main keyword becomes a pillar page. The related ones become supporting posts that link back to it. This cluster structure is exactly how I built this blog, and it relies on the foundations in my technical SEO checklist.
Head terms versus long-tail keywords
Keywords come in different sizes, and the size changes your strategy. Broadly, there are head terms and long-tail keywords. Knowing which to chase is half the battle, especially on a younger site.
A few big head terms, then a very long tail of small, specific searches.
Head terms are short and broad, like “shoes”. They pull huge volume, but competition is brutal. As a result, ranking for them takes years and deep pockets. For most sites, they are the wrong first target.
Long-tail keywords are longer and more specific, like “waterproof running shoes for flat feet”. Each one has low volume on its own. Together, though, they make up the vast majority of all searches. Some studies put long-tail queries at around 91% of everything typed into search.
Long-tail keywords win on two fronts. First, they are far easier to rank for. Second, they convert better, because the searcher knows exactly what they want. Therefore, I load up new sites with long-tail keywords and leave the head terms for later.
Let me put numbers on it. One head term might get 50,000 searches a month, yet you may never crack the top ten. Meanwhile, fifty long-tail keywords might get 200 searches each. That is 10,000 visits a month, from pages you can actually rank. So small keywords add up fast.
The five metrics that judge a keyword
When you weigh a keyword, no single number decides. Instead, you read several together. Here are the five I check before I commit a page to any keyword.
Read these five together. Search intent quietly outranks the rest.
Search volume tells you roughly how many people search a keyword each month. Treat it as a range, not a fact, because every tool only estimates. Keyword difficulty tells you how hard it is to rank against the current pages. A high score plus a new site means a long wait.
Search intent tells you what the searcher wants to do. As I said, it is the one that matters most. Business value tells you how close that searcher is to buying. A small buyer keyword often beats a huge vague one.
Finally, trend direction tells you whether interest is rising or fading. Rising terms are where tomorrow’s traffic hides, so I watch them closely. Together, these five give you a full picture. On their own, each can mislead you.
The best keyword research tools
You do not need to spend a penny to start. The free tools are genuinely good, and they cover most small sites completely. Pay for a bigger platform only when research becomes daily work.
Start with the free Google tools. Add a paid platform when you need scale and depth.
Free tools to start with
Begin with Google Keyword Planner. It gives volume ranges straight from Google, which is the source that matters. Next, add Google Trends to spot rising topics and seasonal patterns. Both are free, and both are reliable.
Two more free sources round out the set. Google Search Console shows the queries you already rank for, which is often where the easiest wins hide. AnswerThePublic surfaces the questions people ask around a topic. Between them, these free tools handle the core of keyword research for most sites.
Paid tools for scale
When you need more, Ahrefs and Semrush are the two I reach for. They offer deeper data, bigger keyword lists, and a clear view of what competitors rank for. In addition, they track your rankings over time, which saves hours each month.
A newer group of tools uses AI to speed things up. Some cluster thousands of keywords into topics in seconds. Others build content briefs from the top-ranking pages. I round up the ones worth paying for in my guide to the best AI SEO tools.
| Tool | Best for | Cost |
|---|---|---|
| Google Keyword Planner | Volume ranges straight from Google. The best free starting point. | Free |
| Google Trends | Rising topics and seasonal demand. | Free |
| Google Search Console | Keywords you already rank for. Easy wins live here. | Free |
| Ahrefs | Deep keyword and backlink data at scale. | Paid |
| Semrush | Broad research and competitor gap analysis. | Paid |
How do you find keyword ideas your competitors miss?
Your best keywords are often the ones your rivals overlook. So once you have a solid list, go hunting for gaps. A little extra digging here pays off nicely.
Start with a competitor gap analysis. Paid tools like Ahrefs and Semrush show the keywords your competitors rank for and you do not. As a result, you get a ready-made list of proven targets. Then filter that list by intent, just as you did before.
Next, mine your own Search Console data. It shows queries where you sit on page two, just out of sight. Often, a small tweak lifts those pages onto page one. In other words, the easiest wins are already in your account. Google explains this well in its Search Central documentation.
Finally, look beyond the tools entirely. Reddit threads, forum posts, and YouTube comments show how people really talk. Because of that, they reveal long-tail phrases no tool has caught yet. These raw, human questions are pure fuel for keyword research.
How to do keyword research for AI search
This is the newest part of the job, and it changes fast. AI answers now shape a big share of searches. So your keyword research has to plan for them, not just for the classic blue links.
The method moved from volume-first to intent-first. Volume still matters, just not first.
The old way treated each keyword alone. You picked a phrase, wrote a page, and chased an exact match. The new way is different. Search engines now think in topics and entities, not just strings of text. Therefore, they reward sites that own a whole topic, not a single phrase.
That means you should research questions, not just keywords. Pay attention to how people phrase things when they talk to an AI. Look at the questions in Reddit threads, in support tickets, and in the related questions on Perplexity. These conversational queries are the raw material for AI visibility.
Target question and conversational keywords
People talk to AI tools in full sentences. So your list needs questions, not just short phrases. Think “why does my page not rank” rather than “page ranking”.
Gather these questions from real sources. The “People also ask” box is a great start. Autocomplete helps too, especially when you open a query with why, how, or what. Then answer each question clearly on the page, in its own short section.
Topical authority is the goal here. When you cover a topic deeply and link the pieces together, you signal real expertise. That expertise is what earns both rankings and AI citations. It also leans on strong, credible content, which I cover in my guide to building E-E-A-T into your content.
To be clear, classic keyword research still runs underneath all of this. AI search does not replace it. If you want the full comparison between the two disciplines, I break it down in my post on SEO versus GEO.
Keyword research mistakes to avoid
After years of audits, I see the same errors again and again. Watch for these, because each one quietly wastes effort.
Chasing volume alone. A huge keyword feels like a prize. However, if the intent is wrong, it brings visitors who leave at once. Always check intent before you fall for a big number.
Ignoring the long tail. New sites often aim only at head terms. As a result, they never rank for anything. Start with long-tail keywords, and build up from there.
Letting pages fight each other. When two pages target the same keyword, they split your authority. So map one primary keyword to one page, every time. This keeps your site clean and your rankings strong.
Doing it once and forgetting it. Search demand shifts, especially now. Therefore, revisit your keyword research every quarter. Fresh terms appear constantly, and you want to catch them early.
Skipping the structured data step. Once your keywords map to pages, mark those pages up clearly for search engines. My schema markup guide walks through exactly how to do that.
Frequently asked questions
How often should I do keyword research?
Do a full round when you plan a new site or a big content push. After that, refresh it every quarter. Search demand moves quickly now, so a yearly check is no longer enough. A quick monthly look at Search Console also catches new terms early.
Is keyword research still relevant with AI search?
Yes, very much so. AI answers still draw on the questions people ask, and keyword research is how you find those questions. The surface changed, but the core job did not. If anything, understanding intent matters more than ever.
What is keyword difficulty?
Keyword difficulty is a score that estimates how hard it is to rank for a term. Tools calculate it from the strength of the pages already ranking. A high score means strong competitors, so a new site should usually pick easier targets first.
How many keywords should one page target?
Give each page one primary keyword. Then let it naturally cover a handful of closely related terms. Do not try to force many separate keywords onto a single page, because that muddies the focus and confuses search engines.
Do I need paid tools for keyword research?
No, not at the start. The free Google tools cover the essentials, and they use real Google data. Add a paid tool like Ahrefs or Semrush only when keyword research becomes a regular, daily part of your work.
What is a seed keyword?
A seed keyword is a broad starting term, like “coffee” or “SEO”. You use it to kick off your research. Tools take that seed and return hundreds of longer, more specific keywords built around it.
Want this done for you?
At Admonq we build keyword strategies that target the right searches, match real intent, and map cleanly to content that ranks. If you would rather have experts handle it, let us take a look at your site.
Keyword research is not the flashy part of SEO, yet it decides everything that follows. Find the right searches, match the intent behind them, and give each one a clear page. Do that consistently, and your content stops guessing and starts answering. That is how you earn traffic in 2026, from search engines and from the AI answers that now sit beside them.
