I spent the first morning of brightonSEO San Diego in sessions on keyword and prompt research in the AI era. Good talks, smart speakers, full room. And about twenty minutes into the first one, I caught myself thinking: wait… isn't this just what we've been doing? For a moment, I was worried I wasn't going to learn anything new. But after sitting with what was talked about, I realized that's kind of the whole point. It's really not anything new if you've been doing marketing, especially search, already.
I did learn plenty in the other sessions, a lot of it technical. But the reassurance that this part isn't new if you're doing it right? That was its own kind of lightbulb moment.
Back to the basics
The pressure to understand AI search and how it's changing, well, everything, is omnipresent. It feels like everyone is talking about it anytime you open LinkedIn or hop on a client call. And it's led to us all looking for the shiny silver bullet that fixes everything. Which is completely understandable. No one really wants to hear that there isn't an easy answer to AI search. It's just doing the work of knowing your audience.
And the work, at its core, is the same stuff it's always been: knowing who you're talking to, what they're trying to solve, and where they're talking about it. That shiny new AEO silver bullet everyone keeps talking about on LinkedIn is a distraction. It pulls our attention away from the one thing that has always held up, which is understanding the people we're trying to reach.
What the talks actually said
Over and over, speakers landed on some version of the same idea: start with the customer, not the tool. One session put it plainly that volume is not a strategy. To be fair, a lot of old-school keyword research was volume-first,and/or built around competitor gaps. And in my experience, every time I've leaned on that over real audience research, it hasn't worked as well. The version of keyword research that holds up has always been the one rooted in people.
Another session walked through how a team approached a brewery brand by going to the client's location and learning the customer through data and conversation and drinking lots of beer, thinking in terms of moments rather than keywords. A moment is the thing happening in someone's life that leads to the search. For the brewery, that meant birthdays, graduations, retirements: the occasions their beer shows up at. For my client, it's the problem someone is stuck on and can't solve alone, plus whatever led up to it.
They were describing audience research, and doing it well. That's the audience-first keyword research we've always done. At the end of the day, it's knowing your customer and what matters to them.
What knowing your audience looks like in practice
I've seen this a lot lately with a SaaS client of mine. They have had a lot of new features drop recently and to be honest pinpointing exact search volume is not easy. Search the feature name or the branded term and you'll find close to no demand, which makes sense: the feature is new, so nobody has the vocabulary for it yet.
If I were taking a volume-first approach, I would stop there and call it a dead end. But the demand isn't missing, I know it's there. It's just not shaped like a keyword yet. People are out there on the world wide web shouting about issues and describing the problems these features solve in their own words. The pain has existed long before the product language does. My client has just finally given it a name and a solution. So in doing research, the keywords didn't exist, but the problems definitely did.
So the work isn't finding the keyword. It's finding where your audience is having these conversations, learning how they describe the problem, and meeting them there. For me, a lot of the time that happens on Reddit or other online communities. I'll read whatever I can to put myself in the audience's shoes, because it's the closest thing to hearing them describe the problem unprompted.
For this particular client, it led me toward adjacent, jobs-to-be-done-style research around a new feature. Instead of stopping at no volume around head terms, I dug into the job someone is trying to get done with this feature or the problem they’re trying to solve or the outcome they want. That led to finding other signals that indicate demand, like how many times something gets downloaded that was meant to solve this problem. Or how often a particular workflow or template get used or shared as a solution. All of these are audience insights that showcase true demand.
What is actually different when it comes to keyword research
We do know that some things in search and audience behavior have changed. Queries are longer and more conversational, prompts carry more context from the searcher, visibility looks different (citations, mentions, sentiment… instead of just rankings), and measurement is messier and still being figured out.
Those are all real challenges and issues in today’s search landscape, and they change how we track and deliver the work, which makes the work we do feel different. But at the end of the day, that doesn't really change how we decide what matters. That still starts with empathy: knowing your audience inside and out.
The takeaway
If you already do audience-first research, you're not behind. You're probably closer than you think, and the move is to point the work you already do at new surfaces.
And if you haven't been doing that, the best prompt research tool is still talking to your customers, reading their questions, and paying attention to what they actually say.
It's not groundbreaking. It still works.
