Yes, keyword research is still relevant. Nothing replaced it. It changed. It still shows you the words people type into Google, and Google says its AI answers can run several related searches behind a single question. What it can't show you is the long, one-off question a buyer types into ChatGPT. So I'd keep doing your keyword research, and add a second habit: writing down the questions your buyers ask.
US monthly searches for "is keyword research still relevant"
Mar 2026
10
Apr 2026
10
May 2026
10
Jun 2026
10
Jul 2026
20
Aug 2026
140
Sep 2026
140
Google Ads search volume via DataForSEO, pulled October 16, 2026. Google rounds these figures.
Why keyword research is still relevant for AI answers
Keyword research is still relevant because Google's AI answers still start with searches. Google's page on its AI features says AI Overviews and AI Mode may use a "query fan-out" technique, "issuing multiple related searches across subtopics and data sources." I covered how that works, and what it means for your pages, in why the official page can lose to a forum post. I won't repeat it here.
Short version: a long question goes in, and shorter searches come out. My read is that those shorter searches look a lot like the ones people type themselves. You won't see the AI's own searches as search volume. Still, since they look like what people type, keyword research points you at the right topics.
Google also says the usual SEO basics still count for AI Overviews and AI Mode, and there's no special markup to add. Old-school SEO is still part of what we do at Plate Lunch Collective, and the words people type still belong in your titles and headings. If you're wondering whether SEO itself still pays off, I wrote about whether SEO is dead.
So the job is still there. The tools just can't see all of it.
What keyword tools can't see
Keyword tools count searches that enough people type. They miss the rest.
Google says so on its own Keyword Planner help page. Keyword Planner gives "estimates" of how many times a keyword is searched each month, and keywords with very low search volume don't show up at all. Your Search Console has a limit, too. Google leaves out what it calls anonymized queries, ones that "aren't issued by more than a few dozen users over a two-to-three month period." Those searches still count in your chart totals. You just can't see the words.
That's where the long questions live. Google says that since AI Overviews launched, people have been bringing it "more complex, longer and multimodal questions." Picture a made-up bakery owner on Maui getting asked, "Can you ship a birthday cake to Oahu without it melting, and how far ahead do I need to order?" Odds are nobody else types that exact question, so no keyword tool will list it.
DataForSEO's data for October 2026 shows the same thing. The plain question "is keyword research still relevant" got about 10 US searches a month from November 2025 through June 2026, then jumped to 140 in August and again in September. Phrasings like "keyword research for AI search" and "how to do keyword research for ChatGPT" had no volume Google Ads would report. People clearly want to know. They just ask in their own words.
How far off are the traffic numbers in SEO tools?
Third-party tools are useful, and I use them. But their traffic numbers are estimates too.
The agency Promodo tested that in a study published in October 2024. It took Google Search Console data from 184 websites for the first half of 2024 and compared it with the traffic Semrush, Ahrefs and Similarweb estimated for the same sites. In a December 2024 write-up of the study on Collaborator.pro, the average error was 61.58% for Semrush, 48.63% for Ahrefs and 56.95% for Similarweb. In some cases, Semrush estimated 130,000 visits where the real number was 50,000.
Average error in estimated website traffic, against Google Search Console
Semrush
61.58%
Similarweb
56.95%
Ahrefs
48.63%
Promodo study of 184 websites, data from January to June 2024, published October 30, 2024. Figures from Collaborator.pro's write-up, December 5, 2024.
That study measured traffic estimates, not keyword volume, and it didn't look at AI Overviews at all. So I wouldn't blame AI for those gaps. I read it as a reminder that every number in a tool is somebody's model. Google's guide to AI search makes a related point: "No third-party tool has access to our internal ranking or AI systems."
I'd use the tools for direction. Which topics are bigger. Which are growing. Then I'd check them against your own Search Console, which is the real count for your site.
How to plan pages around the questions people ask
In SEO, this is called keyword intent mapping: sort the searches by what the person wants, then decide which page answers which. I still do it. I just start from the questions now, not from a spreadsheet of words.
Put close variations on one page
You don't need a page for every way to say the same thing. Google's guide to AI search says its systems understand synonyms, so "you don't have to worry that you don't have enough 'long-tail' keywords or haven't captured every variation." It also says that making separate pages for every variation, mainly to game rankings or AI answers, breaks its spam rules.
A made-up example. A pest control company doesn't need one page for "termite inspection cost" and another for "how much is a termite inspection." One page that answers the cost question plainly does the job for both. I wrote more about why thin pages split one topic in the chunking post.
Spell out who you are, where you work and what you sell
I think of it this way: an AI answer can only name you if it can tell which business you are. So say it plainly and the same way everywhere: your name, what you sell, the places you serve. Google lists keeping your Business Profile up to date as one of the basics for its AI features. The rest of that job is entity SEO, getting the facts about your business to match everywhere they show up, on your site and off it.
Answer each person who weighs in on the purchase
In my experience, a bigger purchase rarely has just one person asking questions. Take a made-up roofing job. The homeowner asks what it costs. Their partner asks about the warranty. The neighbor on the HOA board asks which materials are allowed. Each of those is a different question, and each one needs an answer somewhere on your site.
Keyword research tends to find the first question. Talking to your customers finds the other two.
Where to find the questions your buyers really ask
Some people now call this prompt research. Here's where I'd look first:
Your last few months of sales calls, emails and reviews. Write down every question someone asked right before they bought or walked away.
The Queries list in your Search Console Performance report. Longer questions show up there once enough people have asked them.
The "People also ask" box and the AI Overview on the searches you care about most.
ChatGPT and Gemini. Ask them your buyers' questions and see who gets named.
There are tools that track how AI answers change over time, and we compare them in our AI visibility tools directory. They're handy. Just know that the ones I've looked at test a set of questions somebody picked, not what real people type into ChatGPT.
What to measure besides rankings
I still check rankings. They just won't tell you whether you're named in an AI answer. For Google's own AI, Search Console now has a report that counts how often your pages showed up in AI Overviews and AI Mode. I walk through what it does and doesn't show in my post on whether SEO is dead. For ChatGPT and the rest, our measurement page lays out which numbers we trust and which we don't.
Where I'd start
Plate Lunch Collective is an AI SEO agency that gets brands named when buyers ask, in search, on social, and in AI. Planning pages around the questions buyers ask is the first half of our citation-ready content work. Writing each answer so it makes sense on its own is the second half.
If I were doing this for your business this week, I'd start small. Run your keyword research the way you always have. Then pull twenty real questions from your inbox and your sales calls. Search a few of them in Google and in ChatGPT. Look at which ones your site already answers, and which ones it talks around.
The gaps are where I'd spend the next month. If you'd rather we take a look, the AI Search Game Plan shows where you stand in Google and in AI answers, who comes up instead, and what to do first.
Common questions
Is keyword research still relevant in 2026?
Yes. Google says AI Overviews and AI Mode may run several related searches behind one question, and in my view those look a lot like the searches keyword research already shows you. What it can't show you is the long, one-off question a buyer types into an AI assistant. Pair your keyword research with the real questions your customers ask.
What is keyword intent mapping?
Keyword intent mapping means sorting searches by what the person wants, like learning, comparing or buying, and then deciding which page on your site answers each one. It's an old SEO habit that still works. The change is that I'd build the map from your buyers' real questions as well as from keyword tools.
What is prompt research?
Prompt research is figuring out what questions people ask AI tools like ChatGPT and Gemini about your kind of business. I don't know of any public count of what people type into AI tools, so you can't look up prompt volume the way you look up keyword volume. Your sales calls, emails, reviews and Search Console queries are the best starting places.
Do I still need a paid keyword tool?
It helps for seeing which topics are bigger and which are growing, and I use them. Just treat the numbers as estimates. In a 2024 Promodo study of 184 websites, the traffic estimates from Semrush, Ahrefs and Similarweb were off by roughly 49% to 62% on average against Google Search Console. Your own Search Console is the real count for your site.
Should I target long-tail keywords for AI search?
Not one page per phrase. Google's guide to AI search says its systems understand synonyms, so you don't need every long-tail variation, and that making separate pages for every variation mainly to game results breaks its spam rules. Answer the question well on the page where it belongs.
Curious how your business looks on Google and in AI answers?
Plate Lunch Collective is an AI SEO agency that gets brands named when buyers ask, in search, on social, and in AI. If you'd like to talk it through, we're easy to reach.
Hayden Bond, in SEO since 2004, founded Plate Lunch Collective, an AI SEO agency that gets brands named when buyers ask, in search, on social, and in AI.
