How has AI changed keyword research?

Industry experts explain know keyword research is changing and what tools they (still) love

Nov 09, 2026

Hello, and welcome back. Jessie here, back from a weekend spent walking Toronto’s ravines and perusing a farmers’ market. Of course, a market in the fall is just an excuse to buy an enormous variety of squashes. A tiny butternut was transformed into dinner (a spicy kale, bean and squash soup).

This week: How has AI changed keyword research? We asked industry experts for their thoughts on how keyword research is changing, and the SEO tools they rely on now.

📅 Mark your calendars: We’ve partnered up with Trisolute News Dashboard and special guest Louisa Frahm to bring you a workshop on how publishers should approach non-commodity content in the AI era. Ask a question.

Add to calendar

This is commonly me, WFB (working from bed), on my many keyword research endeavours. Photo by Windows on Unsplash.

Let’s get it.

THE 101

How was keyword research changed?

Keyword research isn’t dead, but it is different. Keyword research was once about “finding the words people typed into Google and matching your content to them,” says Mark Williams-Cook.

Now, with AI and the query-fan out method, there’s the “wider notion of targeting the topic, not the keyword,” says Steven Wilson-Beales.

Gone are the days of reporting search volume, then writing new content — or optimizing existing pages — to match the specified terms. Now, it’s about mapping broader topics and targeting entities, with a particular emphasis on areas that are strategically important to the business.

Entities and topics are more important than ever. Entity SEO is the idea that publishers should create and optimize content on clearly defined, real-world topics over standalone terms. It’s “things not strings,” to use the phrase Google introduced in 2012.

Our experts encourage publishers to optimize across an entity, and for reader needs within that topic area. “Our job as SEOs is to connect the dots between the keywords and what the person actually wants and means by them,” Oleg Korneitchouk adds.

“Understand the audience’s needs and contribute something useful,” John Shehata says. Editors need to “investigate the connected questions around a story” to create more meaningful content that helps readers, he explains.

That’s a sentiment Louisa Frahm echoes: keyword research is now about understanding “what someone needs, their personal circumstances and what information they’ll seek out next.”

A “less rigid approach” to keyword research is required now, says Barry Adams.

#SPONSORED

Connect the Trisolute MCP Server for conversational search data insights

Ever wish you could connect your search data to your favorite AI assistant like ChatGPT or Claude and have an insightful chat? Now you can with the Trisolute MCP Server.

Get insights into SERP share of voice, AI Overviews, content gaps and more
Access a dozen tools via Trisolute’s robust API offerings
Create custom visualizations using Trisolute data
From a single prompt, pull data from multiple reports

Don’t panic: Trisolute has the data you need.

Snag your free trial!

How have keyword research metrics changed?

Traditional keyword research was largely driven by search volume (the average clicks to a term each month), which is a useful metric to understand overall interest.

However, as AI features transform Google’s ecosystem, more terms are considered zero-click. Chatbots encourage longer, more conversation terms, and so the industry is split on how much volume truly matters.

“Organizations need to let go of metrics like search volume (which was grossly inaccurate in the first place) and look at metrics like topic coverage, share of voice, and brand salience,” says Barry.

Volume has never been a particularly useful metric, adds Matthew Bain. While it can be helpful when preparing for service pieces, tentpole events or recurring moments, it’s insufficient for breaking news.

“Historical data tells you little about the interest [a keyword] might generate,” Matthew says. He’s more interested in how quickly interest in a topic emerges, and its related search terms.

“When a story breaks, interest is still taking shape: someone unknown can become a central figure within hours, long before keyword tools register any search volume,” he adds.

Volume underestimates emerging demand, too. “There’s value in recognizing an audience’s need before it becomes a high-volume keyword,” says Louisa.

Relying on traditional metrics means you’re likely to miss important angles. Looking at how fast interest is climbing — aka the velocity — reveals the subtopics and coverage lines readers care about most.

John agrees that velocity metrics matter more than volume. Velocity considers “how quickly interest is rising, when it may peak and which related questions are emerging.”

Long-tail keywords — terms that are multiple words and more specific than the head term — are often missed in traditional SEO tools. However, they’re becoming more common in the AI era as user behaviour shifts toward longer, more conversational queries.

“Keyword tools miss more of the long tail now,” says Mark. “Search is more conversational, so the raw keyword volume data doesn’t capture the follow-up questions and fan-out queries people actually use.”

Those queries are often follow-up questions asked by readers, after the main keyword (for example, “dana white” is a main term, whereas “did trump offer dana white press secretary” is a long-tail keyword that traditional SEO tools might miss).

“Volume is about potential,” says Oleg. He says to focus on “queries tailored to your audience that you can answer better than anyone else.” Use search volume as a proxy for interest, then answer more specific questions your readers are searching.

What tools are you using now and what are possible pitfalls?

Most experts still rely on traditional SEO tools. “My core stack hasn’t changed that much,” says Mark.

Google Search Console, Google Trends, Screaming Frog, Semrush, Ahrefs and others are still the workhorses for SEO.

“Those are the old reliables and they still answer the important questions,” Mark adds.

“Google Trends is especially useful for an editorial mindset: identifying when interest is rising, how language is changing and which related questions are emerging,” says Louisa.

Matthew agrees. Google Trends, along with social listening tools, plus looking at the output of other publishers, are key to “understand how interest develops.”

Many traditional SEO tools have a data lag, while social listening tools provide almost instant feedback. Oleg also uses comments and questions posed by users on publisher stories posted on Reddit to deepen his research.

John uses NewzDash to identify trending queries that trigger AI Overviews, as well as monitor publisher and competitor visibility.

Tracking referral traffic from AI Overviews is available in Marfeel’s tool for real-time traffic and historical data. That enables publishers to know what “type of content that gets them traction for citations in AI summaries,” Matthew explains.

New purpose-built tools like Profound, Peec AI and Waikay are used for AI visibility and prompt tracking. Steve uses a variety of prompt trackers, but cautions that there are many possible pitfalls in conducting keyword or prompt research for AI.

“It’s just too easy to spin up hundreds of synthetic prompts that might give you an insight you want — but is it the insight you really need?” Instead, Steve always brings all stakeholders together during this process to “identify questions that actual humans are asking.”

Relying too heavily on synthetic prompts can result in another pitfall: publishers losing their unique voice. “Another danger is losing your distinct expertise by relying on the same automated suggestions as everyone else,” Louisa says.

It’s also unwise to think how people search on Google is the same as how they prompt in AI. It’s not a one-to-one. “Search queries are often centred on an entity or event,” John says. “AI prompts may ask for an explanation, an update, a comparison or a source. We need to research both behaviours.”

The bottom line: AI is everywhere — including in keyword research and tooling. Remember that the fundamental value of keyword research is understanding what your readers are interested in, then responding to queries and questions where you have authority.

#SPONSORED - The Classifieds

Get your company in front of more than 14,000 writers, editors and digital marketers working in news and publishing. Sponsor the WTF is SEO? newsletter!

THE JOBS LIST

Audience or SEO jobs in journalism. Want to include a position for promotion? Email us.

The Economist is recruiting an Audience Editor in (London UK).
The New York Times is hiring a Vice President, Audience Insights (New York City, New York).
Qwoted is hiring an UK Newsroom Partnerships Executive (London, UK).
Politico is hiring a Social Media Editor (Arlington, Virginia).