EP 274 – Why “Position One” Doesn’t Exist in Local Anymore (Brighton SEO Recap)

A single, universal search rank is a relic of one-size-fits-all queries — so as AI systems weigh a searcher’s location, history, and budget before they even generate an answer, visibility stops being a position to win and becomes a distribution to earn across every path a customer might take to ask.

Mark Kabana’s End of Position One

Around 26:52, Mark Kabana lays out the argument that’s been building across the episode: there’s no such thing as a universal “position one” anymore, because context decides the answer before ranking even starts.

Element	What it answers	Primary lever
Old rank (position one)	“What’s the single best answer to this exact query, for anyone who types it?”	Keyword and query optimization
Context (location, history, budget, company, time of day)	“What’s the best answer for this person, right now?”	Signals AI systems weigh before ranking even begins
Visibility as distribution	“How often does my brand show up across the finite set of conversational paths a customer could take?”	Coverage across paths, not rank on one query

Takeaways

• AI vendors already outnumber local ones on the expo floor. Andrew ranks Brighton SEO’s vendor mix: AI trackers and AI content makers first, AI citation vendors next, “and then local guys” third. (07:44–07:59)

• Near Media’s own unpublished survey says the industry has the emphasis backwards. A follow-up consumer survey found the overwhelming majority of internet use is tied to real-world, local intent — buying products, finding services, deciding where to eat — not the online-only behavior SEO conferences are built around. (08:12–09:40)

• Andrew has tried and failed to correlate AI visibility with direct traffic. Despite the industry assumption that showing up in AI answers should lift direct visits, he says: “we just I can’t tell you that we’ve seen any” — he’s tried repeatedly and can’t find the link. (15:40–16:17)

• Local search is going conversational, not query-based. Mark’s example: instead of typing “hotel with two sinks,” people now just tell ChatGPT what they need and let it hold the follow-up conversation — a shift he says is already changing how he personally searches. (12:48–13:26)

• People already sort AI tools by task, per Morning Consult survey data. Google AI/Gemini gets used as an extension of Google itself; ChatGPT gets the deeper, more personal requests — trip planning, “be my therapist” — a distinction Greg says shows up consistently in Morning Consult’s research. (19:23–20:33)

• Google is racing to close a personal-context gap it knows it’s losing. Mark’s read: Google’s new push to connect contacts, tools, and MCPs is a direct response to people sharing far more personal information with ChatGPT and Claude than with Google itself. (22:17–23:32)

• Andrew’s counter to “no more position one”: the paths are still finite and mappable. For most local verticals he puts the real number of conversational paths closer to twenty than infinite — a dentist’s tree is basically “insurance or not, whitening or not” — which makes the problem tractable rather than open-ended. (30:22–30:58)

• Google’s own review prompts are a free customer-research panel. Mike’s tactic: audit which attributes Google is now asking reviewers to fill in for your category — that’s a direct signal of what customers actually care about, at no research cost. (33:44–34:00)

• Google’s local data moat survives the AI transition. Even ChatGPT, despite its own licensing deal with Yelp, is still scraping Google’s local data to answer questions — evidence that Google’s local graph remains the deepest dataset in the category regardless of which chat interface a consumer uses. (34:14–34:33)

• Agentic, permissioned browsing is already live in mainstream consumer AI — twice, independently, in the same week. Greg watched ChatGPT ask permission mid-session to open a browser and search Google, then Yelp, for body-shop recommendations; Mike had a similar experience getting an AI tool to check business licenses across fifteen locations with a state licensing board. Mike notes Claude has done this via its skills for a while. (34:33–36:35)

Related Reading

Concepts

• No more position one — Mark Kabana’s term for the collapse of a single universal top rank now that AI weighs a searcher’s own context before it ranks anything.
• Visibility as distribution — tracking presence across a finite set of conversational paths a customer could take, rather than one rank on one query.
• The conversational path — Andrew’s IVR-chain analogy: a bounded decision tree of follow-up questions (insurance or not, whitening or not) rather than an infinite open-ended chat.
• Agentic, permissioned browsing — a consumer AI assistant asking to open a browser and search another site live, mid-conversation, before it answers.

Practitioner Notes

• “No more position one” is a direct rationale for user-research video over rank tracking. If a single query no longer has one universal best answer, a rank report can’t show what actually happens for a given searcher — watching a real session, as in the Choosing a Lawyer study, is the only way to see which path a given context actually produces.
• Andrew’s failure to correlate AI visibility with direct traffic complicates the discover-in-AI/validate-on-Google pattern documented elsewhere in the brand series — worth testing against Near Media’s own referral data before treating that pattern as settled; a lack of correlation in his numbers doesn’t prove a lack of causation given how blunt last-click attribution already is.
• The two live agentic-browsing anecdotes are a fast-moving GBP/listings signal. If mainstream consumer AI tools are now browsing live rather than relying only on cached training data, listing accuracy and GBP hygiene matter on a much shorter feedback loop than the industry has been assuming.

Pick your starting point:

00:10 Cold open & guest intros
01:08 Brighton SEO takeaways: automation over marketing
04:08 Where “local” AI actually happens: the Siri/iPhone test
07:44 AI vendors already outnumber local vendors on the floor
08:12 Near Media’s survey: the internet is mostly local intent
10:13 GoogleU, Sheryl Sandberg, and the death of the Yellow Pages
12:48 The conversational shift: “does the hotel have two sinks?”
13:27 Does 99% of AI shopping search end back on Google?
15:40 Can you correlate AI visibility with direct traffic? (No.)
19:23 How consumers actually sort ChatGPT vs. Claude vs. Google 22:17 Why Google is racing for your contacts and MCPs
25:58 iOS 27, Siri, and 2.5 billion queries a day 26:46 The end of “position one”
30:22 Are conversational paths really infinite? Andrew says no
33:44 Free customer research: audit Google’s review prompts
34:33 ChatGPT (and Claude) are already browsing live, with permission

Full Transcript –>

Editor’s note: This is a cleaned reading transcript of “Magic-episode-01.txt” — Andrew Shotland (Local SEO Guide) and Mark Kabana (Places Scout / Yext) with Greg Sterling and Mike Blumenthal. Stutters, false starts, and pure backchannel (“yeah,” “right,” “mm-hmm”) have been removed or folded into the surrounding turn for continuous reading; wording is otherwise left faithful to what was said. A small number of proper names could not be independently verified and are left as heard — see the Corrections list at the end for the full list of what was fixed versus what was intentionally left as spoken.

---

Greg (00:10)
Hey everybody, welcome back to the Near Media Podcast — it’s me, Greg Sterling, and as always, the lovely and talented Mike Blumenthal. Today we have two great guests: Andrew Shotland of Local SEO Guide, and Mark Kabana, of course, of Places Scout, now part of Yext. How are you guys doing today? Good? Enough coffee?

Mark Kabana (00:29)
Interpreting.

Andrew Shotland (00:29)
Fantastic, Greg.

Greg (00:32)
Well, it’s great to have you here. And of course, I asked you to be on this podcast at Brighton SEO — which, three out of the four of us… Mike wasn’t there, but the other three were. Let’s start off today: we’re gonna take the temperature of the industry, talk about trends, where we see things right now with AI and SEO and local SEO. But let’s do a quick recap of your experiences at Brighton first. What are your reflections on it? Anything particularly interesting — or not interesting? Strange, wonderful, disturbing?

Andrew Shotland (01:08)
Besides just getting to see a lot of great SEO people we don’t see all the time, I found the most interesting things I got from attending sessions and talking to people were around automation. There were some very good talks on how to basically automate your agency using AI, which I found particularly useful. Everyone’s talking about being a “marketing engineer” — there isn’t a lot of “here’s how to be a marketing engineer” content, or actually there is a lot, but I thought the people who spoke about it did a particularly good job of helping people get started — Noah Learner, and a guy named Cristiano Winckler, had a really great talk on it. I also found Cindy Krum’s talk on how video is becoming the big AI thing very compelling — in typical Cindy Krum fashion, she just spewed out a million facts at once and showed she’s very smart. It was distinctively different from all the other AI presentations, of which there were many. Go look up what people said about her — I did a whole thing on it on LinkedIn if you want to see.

Greg (01:29)
So, internal operational stuff?

Andrew Shotland (01:31)
Yeah, internal operations. Yeah — because everyone’s talking about being a marketing engineer and stuff like that.

Mike B (02:20)
Yeah, Cindy is always astute at capturing these trends. The interesting thing about video is it’s the only original source on the internet these days — everything else is self-ingested, previously ingested, or spewed content from AI. And with that breakout from Hugging Face, they talked about how they’d intentionally polluted most of the sites in the world with their own content, so that most of it couldn’t be used anymore. So it’s like video becomes the only place where human originality exists.

Andrew Shotland (02:55)
And people, as we’re doing right now — people seem to like watching people do stuff versus reading about it. So, not a bad thing.

Mike B (03:02)
So — I probably shouldn’t be picking my nose at this moment, but… what is it? It’s information gain for the GPT, right?

Greg (03:05)
Probably not. Probably not.

Andrew Shotland (03:07)
Depends — yeah, depends what you’re trying to be visible for.

Greg (03:11)
Sure — Mark, what about you? What were some things that were interesting to you about the show?

Mark Kabana (03:19)
So it’s more into the shift of us talking more about AI than anything else. I can’t tell you how many times I’ve heard the word “agentic” and “agents” throughout the conference, in the market material, in people’s conversations. And I think at some point we start to get a little bit over-fixated on AI itself — especially within this crowd, where we focus on local — and we forget how important local plays in there, and how do we optimize for local within the AI ecosystem. There was a lot of definitely interesting information and talks outlined. However, as far as actionable items in terms of being a local business, I found some of that content kind of underwhelming — somewhat not existent, as much as I would typically see at these conferences in the past.

Mike B (04:07)
Does some of that come from the lack of understanding about how my wife uses her iPhone? I just started using Siri in iOS 27 — and I know Greg’s been using it for a while — but Siri Intelligence. And it strikes me that this is how most people, and where most people, are going to be using AI: something like Siri, which is integrated, available, reasonably smart, solves problems. That, to me, is where local is gonna happen — is my wife, driving to Buffalo, looking for a unique thing and asking Siri where she might find it. Do you think it’s just this sort of gap between our reality? Let me digress a little bit: when we started doing user behavior research, and I watched how people searched, after doing twenty years of local, I was shocked. I thought, “My God, this is crazy.” Some woman looked at twenty-five different hotels because she wanted a full-length mirror for a trip to Italy. She couldn’t go to Italy if she wasn’t looking good, and she couldn’t buy a hotel room if it didn’t have a full-length mirror. I sat through twenty-five minutes of this in user testing, and I was like, “My God, people really do act this way.”

Greg (05:17)
So let’s put a pin in this and come back to it, because that anecdote is a perfect expression of the difference between SEO and AI optimization.

Andrew Shotland (05:29)
I think what’s happening — my understanding is conference organizers like Brighton or SMX have always thought that catering to the local audience is bad economics, ’cause I think they think it’s SMBs or something like that.

Greg (05:46)
Yeah, that was kind of the conclusion of Chris Elwell at Third Door Media, when we tried to do — we tried to do some local shows, and they just kept saying the demand isn’t there, the demand isn’t there. But it’s a misunderstanding — well, at least from an importance, a significance standpoint — it misunderstands the market.

Andrew Shotland (06:07)
It makes some weird assumption that the market is all SaaS and e-commerce — and half the e-commerce is all retailers, and half the SaaS is catering to SMBs. These conferences seem to think everyone’s a Fortune 500 company that doesn’t have locations. Outside of that, it’s not very interesting to them.

Greg (06:34)
What — I attended a few sessions, not in any disciplined way. I kind of wandered in and out of some sessions that sounded interesting, but there was very little explicit local content — like, “this is about local search or local SEO.” The sessions were framed in different ways, not “here’s how to optimize for local consumer intent” or whatever the description would be. There were a handful of sessions — four or five — that touched on it.

Andrew Shotland (07:07)
Yeah, I signed up.

Greg (07:09)
So the conference really just didn’t have much of that content. And yet the vendors, interestingly — we were talking about this in the green room — there were a lot of vendors that were listings management, reputation management, sort of franchise-y listing syndication, and sort of quasi-agencies dealing with local. There was a kind of a split there. It didn’t dominate the vendor floor, but there were enough of them there — it was a significant presence. But that wasn’t reflected in the agenda. Strangely, interestingly.

Andrew Shotland (07:44)
Yeah, I’d say local vendors were like the third most popular. There were AI trackers and AI content makers, link builders, and AI citation vendors — and then local guys, I’d say, is how I’d rank it. I think it’s because they think the customer is agencies. And so they’re trying to cater to agencies and Fortune 500 companies, and then they forget that agencies mostly work with SMBs.

Greg (08:12)
Well, I feel like I’ve been rolling — you know, Mike, in a different way too, and maybe you guys as well — I’ve been rolling a sort of a rock up a hill for twenty-plus years, trying to say this local stuff is really way more important than almost anything else going on on the internet. We just did some survey-based research — one of the questions, it was a follow-up survey from one we did last year for a client — one of the questions was to consumers: characterize how much of your search behavior is directed at finding local business or things to do in your area. Essentially a laundry list of local-intent stuff. It was designed to indicate: nearly a hundred percent, three-quarters, fifty percent, less than that, whatever. And the majority of people said, like, most of what we’re doing on the internet pertains to stuff that’s happening in the world — it’s happening offline. We’re buying products, we’re trying to find services, we’re looking for restaurants. That’s the real behavior — that’s the reality of consumer behavior: people buy a lot of stuff online, but mostly they’re using the internet to help them make decisions about things they do in the real world, or buy things in the real world. And that’s just never been reflected in the SEO industry or conferences. It’s really mysterious to me — it seems self-evident.

Mike B (09:40)
It’s particularly mysterious, given that Google has recognized this, their algorithm recognizes this, Mark and the green room recognize this — that if you don’t have brand, or you’re small in your local market, you’re not going to get the conversion on Google, you’re not going to get the rank on Google. Google recognizes it. The conferences don’t, but Google does.

Andrew Shotland (10:01)
Is it possible that it’s just — local is so fragmented, right? — that it’s hard for these conference guys to aggregate enough local things?

Greg (10:13)
It’s not — I mean, Mike with LocalU, and me, long, long ago, with the Kelsey Group and other things — we programmed conferences —

Mike B (10:21)
Names not to be mentioned.

Greg (10:24)
— okay, wait, wait, wait — the Kelsey Group is no more —

Andrew Shotland (10:26)
Did you just violate a consent decree or something?

Greg (10:31)
— it doesn’t exist anymore. So, quick footnote: very early on in Google’s selling of AdWords, they were hosting — it was GoogleU, and this was like teaching people how to do AdWords. They’d bring people into a venue, and Sheryl Sandberg got up there and said, “Hey, this is what we’re doing — we’re doing advertising, it’s directional media, just like the Yellow Pages.” That was the analogy she used to explain Google AdWords to business owners. At the time, the Yellow Pages was still relatively healthy — the big Yellow Pages publisher in the United States, now called Thrive, formerly Yellow Book, bought a lot of the existing companies and then sold off its print directories to private equity. There’s a couple of publishers still left, so we can finally say RIP Yellow Pages now — but it’s twenty-five years later that the industry is really dead. Anyway, that was a bit of a digression, but there’s a lot of stuff to talk about.

Andrew Shotland (11:35)
Thanks, Gramps.

Greg (11:38)
There’s a lot of stuff to talk about in local.

Mark Kabana (11:38)
Well, I think a lot of what they missed, Greg, was the fact that — with local and everything we’re talking about — they’re not really talking about the shift itself: where is that behavior shifting, what are consumers doing differently, how do we as marketers address those shifts, and what do we need to do to actually win and be considered and ranked inside of AI search. That’s a lot of what I’ve seen missed. It’s a lot more about automation and agents rather than consideration, intent, context — obviously the biggest change we’re seeing right now, because that absolutely changes everything. So I think a lot of the conversation would be better if it was geared towards those shifts, and how do we as marketers address them.

Andrew Shotland (12:22)
I think every conference would benefit if ninety-nine percent of the content was someone getting up for five minutes and going, “Here’s three things that I did that worked, and here’s exactly how I did it.” And goodbye.

Greg (12:35)
Well, I think they were trying to do that — each session was twenty minutes long. They were trying to do a version of that: condense it into actionable tips and then move on to the next session. I don’t —

Mark Kabana (12:48)
One of the biggest shifts going on is the conversational shift. Like you were saying, Mike — performing twenty-five minutes of searching to figure out if a hotel has a mirror. One of the special ones for me: does the hotel have two sinks? I’m not battling with one tiny thing — I always like the photos. How big is that sink in the room, before I book it? Stuff like that matters. And now, instead of searching that —

Andrew Shotland (13:10)
Wait a minute — how much sinkage does one man need? What are you doing with that second thing?

Greg (13:14)
And, precisely —

Mike B (13:15)
Needs one for the beard and one for the brushing.

Mark Kabana (13:18)
But now, the shift is that I’m doing that via conversation instead of through Google search, because ChatGPT can do all that for me.

Andrew Shotland (13:26)
Maybe you just need a bidet.

Mike B (13:27)
Although, interestingly — Brendon Kraham from Google just said that ninety-nine percent of searches that start on ChatGPT for shopping — or, I assume, he meant Claude as well — end up on Google to close. This was a statement he made yesterday or the day before, at some conference. He’s a Googler.

Andrew Shotland (13:42)
Make sense?

Greg (13:44)
So, yeah — that’s directionally true, but not literally true. I mean, people do use Google after AI, for sure, a hundred percent.

Andrew Shotland (13:55)
We were talking about the New York Times versus OpenAI transcript that came out —

Mike B (14:01)
Copyright case, yeah.

Greg (14:02)
Copyright. Yeah.

Andrew Shotland (14:04)
— another thing that came out in there was that a Microsoft executive for Bing said: no one clicks on it in the AI. Or maybe ChatGPT said it, something like that.

Mike B (14:15)
Right — click-throughs drop by ninety percent or something.

Andrew Shotland (14:18)
No — that was from the search engine. They said in the AI, when you’re talking to the LLM — no one, they said nobody. “We can put links up there as much as we want, no one clicks.”

Greg (14:29)
Because you’re only gonna click through — in a conventional search result, you have to click through to get the information. Now, because it’s all delivered to you, packaged up to respond to your specific question, there’s no reason for a click until the very, very bottom of the funnel — if you wanna go buy something or verify something.

Mark Kabana (14:50)
Really, until the action’s performed — that’s kind of the way I look at it. A lot of times now I skip search entirely, because ChatGPT summarizes everything, and context plays such a big role. Going back to the two-sink analogy in a hotel room: my ChatGPT knows when I say, “Hey, I need to find a hotel room” — and, that’s Lisbon, where I’m planning on going after the “Cynic Conference” next month.

Mike B (15:10)
And this one has three sinks —

Mark Kabana (15:12)
Yeah. It knows to recommend a hotel room with two sinks. And if it recommends it to me, I’m going to most likely skip search entirely — I’ll just go to that website, it’ll give me a link, and I’ll book it directly. So that’s another fundamental shift: the human isn’t doing a lot of the searching anymore. The AI is summarizing it, letting us make a decision based on our needs and context, and perform the action completely outside of Google.

Greg (15:40)
So I wanna come back to this point Mike made about his —

Andrew Shotland (15:42)
This is what drives me a little nuts.

Greg (15:45)
Yeah — go ahead, Andrew.

Mark Kabana (15:47)
What is it, Andrew?

Mike B (15:48)
He’s just being driven nuts in general.

Andrew Shotland (15:51)
I think we’re having a little lag — so I was saying, by my bandwidth… what Mark said, like — well, I just go right to the website. Supposedly your increase in AI visibility would now lead to an increase in direct traffic. And we just — I can’t tell you that we’ve seen any. I’ve tried to correlate that like crazy, and I can’t. I can’t do it. I have not seen it.

Mike B (16:17)
What about conversions, though? Are you seeing conversions dropping, or business dropping? That’s really the question, at the end of the day. In other words, Mark or I would pick up — I would pick up the phone and call. Google doesn’t even give you the phone number in local results anymore, right? So —

Greg (16:32)
Unless you’re an advertiser.

Mike B (16:33)
— unless you’re an advertiser. And so I would do the query on someplace — Claude, ChatGPT, Google, and now Siri — and I would just pick up the phone and call. That would be my preferred interaction. And so the conversion is the same — the path is just a little different. It’s not that last-attribution — last attribution is Google, right?

Mark Kabana (16:55)
Right — and I think that’s just because some of the changes Google’s made that annoy us as users. Now we find we have a different path out of Google, and out of these massive search result pages with all the ads on them, to find information. We don’t have to waste time sifting through search results anymore when the AI can summarize all the information we need — especially with context coming into play. It already knows what we need, what we want, and can make that recommendation outside of Google. And, Mike — that could be a good segue into the ads screenshot you have.

Mike B (17:28)
Yeah — well, just one note on your comment: one of the reasons I use Google is to find stuff on Amazon.

Greg (17:34)
Right, that’s right.

Mark Kabana (17:35)
Well, let’s say we don’t use Google. Like, I use Google AI mode a lot — if I need a quick, dirty, dry version of an answer, I’ll use AI mode. But if it’s more important to me, and I want to spend the time, I’ll use ChatGPT all day over Google.

Mike B (17:52)
Although, I would say I do that a lot too, but the reason is my browser default. I do a query in my browser URL bar and it defaults to Google. If that defaulted to Siri, Claude, or ChatGPT, those quick answers could come from almost any place for me.

Greg (18:10)
Well — you can change it. Oh — you can’t change the default on Chrome, I guess. That’s right.

Mike B (18:21)
Or on Safari, on my phone — you can’t change the default URL to go to Siri or ChatGPT. You can only change it to Bing or DuckDuckGo. So — but, go ahead.

Andrew Shotland (18:35)
I’m curious, for the Near Media boys — what I find is, I have very specific things I use ChatGPT for, specific things I use Claude for, and Google for. We’re hyper-online, hyper-tech pros. I’m curious, in your research: are you finding consumers think, “this is a ChatGPT action”? Or are they just — like what Mark’s saying — “No, I just search”?

Mike B (19:04)
We didn’t have the technology prior to this to answer that — our current run’s technology was sort of limited in terms of where people could go and what we could record. But we’ve expanded our technology capacity, so the next run we’ll be able to answer that question better. But you have to pay to get it.

Greg (19:26)
Pay to play, just like Google. I have an answer for you, Andrew, but it’s based on Morning Consult survey data. They’ve found that people are making these distinctions and see different use cases for the different AIs. This isn’t super clear-cut, but there are indications people are discriminating between them — using Google AI for this, ChatGPT for that, and some number of people using Claude. It’s based, to some degree, on brand perception and user experience. The one thing I remember from their report is that people tend to use Google AI / Gemini / AI Mode as an extension of the way they use Google — they see it as an expansion of Google. Whereas people use ChatGPT very differently — much more personal information, much deeper: “here’s what I’m trying to do, I’m going on a trip,” or “be my therapist, ChatGPT, tell me how to…” whatever the problem is.

Greg (20:33)
I think that people — and in my personal experience, I bounce between Claude and ChatGPT all day long. I clearly prefer Claude for certain kinds of tasks, mostly work-oriented stuff. Then ChatGPT — I’ll do quick answers or factual information I’m looking for, or as a Google alternative, like Mark was saying — I use ChatGPT pretty much as a Google alternative. And then I’ll go to Google when I need something local, or need to find an article — like Andrew was talking about, the copyright litigation with the New York Times and OpenAI, so I need to find the latest thing, and I’ll go to Google and find that. So, anyway — a very long-winded way to say: yes, I think as people become more sophisticated, they’re making these distinctions between “this is good for this, and this other thing is good for that.”

Mike B (21:31)
How many of them, though, have — like I have — Siri, paid ChatGPT, paid Gemini, and paid Claude? How many people have that? I mean, I assume you —

Mark Kabana (21:42)
Yes.

Mike B (21:44)
— have multiple paid, right?

Greg (21:45)
Yes.

Mike B (21:46)
Exactly. The four of us do. But in the real world, they’re using the free versions, or —

Andrew Shotland (21:49)
We’re the problem.

Greg (21:51)
Well, those numbers are discoverable — I don’t know them off the top of my head. ChatGPT has a billion users, and some subset of those are paid, same with Gemini, same with Claude. But it’s the minority.

Andrew Shotland (22:07)
My guess is we’re talking about the free users. So — are they sticking with one, or is it like Google for search, ChatGPT for local, or who knows?

Mark Kabana (22:17)
So I wanna make a point here, ’cause I think we’re kind of missing exactly what this all means. I think Google’s starting to realize that consumers who use AI are sharing a lot more personal information with other tools than with Google. They’ve kind of fallen behind in that space — like Greg said, we share a lot more personal information with ChatGPT and Claude and our other tools, and we do not do the same thing with Google. Google is recognizing that, and now they’re really pushing — they had a CTA the other day: “hey, why don’t you allow me to share all your contacts, recommendations, connect all your tools.” They’re really making a push now to get you to add all your connectors, MCPs, and your contacts into Google itself, because Google is realizing they’re losing ground to these other tools — they have not become an AI like they should; it’s still like an old search engine. Of course they have Gemini, and that’s why they’re making a shift into AI Mode and actually becoming more of an AI itself. But in order to do that properly, you have to have context, because every search is different based on the person’s behaviors or context. Google needs to catch up in that space — they need to get people to share the same information with Google that they share with ChatGPT and other AI tools, in order to provide them with the right results.

Mike B (23:35)
I did do an experiment, though, where I asked each of them to categorize me as a marketing persona and identify my demographics — age, sex, and other information.

Greg (23:46)
How many sinks — how many sinks you like, that sort of thing.

Mike B (23:49)
Yes — that kind of was, yes, how many things I like. That’s correct. And — surprisingly, ChatGPT was the most accurate, because I’d asked it the most questions over the previous year — identifying my predilections in terms of purchasing, type of quality I wanted, my age, and income. Google wasn’t far off. But anybody using these tools should go ask that tool to characterize you, see what it says about you — it gives you a sense of who’s gathering what information, and what they can really determine about you. And it’s a lot.

Greg (24:27)
I think Google started this shift in its UI/UX in response to ChatGPT — ChatGPT emerged, Google had this “code red,” and started this whole process that’s led us to where we are today, and beyond. I was telling this to — I think we were talking about this, Mark, at the event — users really like the chat interface. They really like the natural-language, unstructured conversational experience, and the ability to ask follow-up questions. That’s clear in survey after survey after survey — they prefer that to the traditional search experience. They don’t trust the AI results quite as much as Google, but they’ve definitely embraced the chat experience. Google historically said, “Well, we don’t do any personalization except for local” — that was the only category, they weren’t doing any personalization. More recently, they’ve said they ARE doing personalization. And to Mark’s point — what they recognize is that they’re getting so much more information about you as a user, it’s a gold mine they never had before. They had queries. But now they’ve got this holistic, enormous —

Mark Kabana (25:47)
Context. It’s really context. Context is what they’re —

Greg (25:52)
— all this context, and they have — as you said, Mike, you’ve figured out how much they know about you — they can infer everything, yeah —

Mike B (25:58)
Based on tangential queries — product purchases, intellectual inquiries, health, cars I was interested in, all this sort of stuff. Everything but who I am — and it inferred a tremendous amount. That said, I see iOS 27 and Siri as a great disintermediator of all three of them. Given that fifty-three percent of the US population is using an iPhone, and they had two-point-five billion daily queries prior — I think it’s going to shift this balance for a large percentage of people: most of what they’re getting from the free ChatGPT, the free Claude, they’ll be able to get from Siri perfectly adequately. And I think that’s going to totally change this landscape — they all have to be afraid of it.

Greg (26:46)
Well, I think, hypothetically, it could really impact Google.

Mark Kabana (26:52)
I mean, I think the bigger question is how is it going to change Google. And I’d like to share a little bit more about that, based on the study we recently did with Joy. I think one of the biggest impacts of this — as Google is trying to get users to share more information and context, so it can personalize results to exactly what they need — is that there’s really gonna be, at some point we’re seeing it right now, no position one anymore. That idea of a universal position one is really starting to break down. Position one still exists for any individual search or individual searcher, but position one for whom, where, under what context does that position one matter? With traditional search, we mostly thought about that as a query. But with AI, the system could potentially know my location, preferences, what I asked five minutes ago, what I liked before, my budget, what time it is, who I’m with — all this good stuff, and actually what I’m trying to accomplish. So all that context can affect that answer before the ranking even begins. And I think that’s one of the biggest changes we’re gonna see: these rankings are gonna start to matter based on context and individual query. A good example: if two people ask, “what is the best Italian restaurant in San Diego?” — the right answer doesn’t necessarily have to be the same for each of those individual people. For me, it means something probably nearby my hotel, that has good wine, not too loud, available tonight, similar to restaurants I liked before. But for somebody else, that same question might mean it’s kid-friendly and inexpensive. So what that change means is: as we’ve always traditionally thought about rank, rank is now moving to an observation under a particular set of conditions, rather than some permanent property of a business. And the whole process of search is moving from “what ranks for this query” to “what fits this person, in this situation, right now.” I think that’s the biggest change we’re gonna see happen in Google in the next upcoming months, as Google’s able to obtain more context about the user and learn more.

Andrew Shotland (28:50)
I want to add to what Mark says — he’s spot on. The way we describe AI stuff to clients: there are, basically, infinite conversation paths it can take. It always ends at the end, and it’s like, “did you mean this, or are you more interested in that?” — kind of gets to what Mark’s saying. I think of it like an IVR chain, like a tree: press two if you want this, press three if you want that. So — I think maybe Yext should do this: instead of looking at “hey, you’re number one for a recommended dentist in this area,” you map out the most common conversational paths and ask, “how does your brand stick around on that path?” — across X number of queries or prompts. And I think that’s really what we’re trying to figure out.

Mark Kabana (29:42)
Right — and it’s shifting towards visibility becoming a distribution, not a single number. It’s distributed across all those various paths you’re talking about. So, in order to even be ranked at this point, you have to be considered — because we have context. And that’s the most interesting thing I see here: one ranking is essentially a sample. It doesn’t describe the whole market, even though the query is similar, because context comes into play and shifts what results get considered for that query. And that’s individual to the user. That’s why we say there’s really no position one anymore — because position one changes based on context, from user to user.

Andrew Shotland (30:22)
But I’ll challenge that just slightly. I actually think there’s a limited set of conversational use cases for most local businesses. It may be a hundred, but I think it’s more like twenty. You’re a dentist, you do teeth whitening — what is the conversation? Do you have insurance? It’s not like an infinite range of “how white do you want your teeth.” So I think there’s a lot of local use cases like that where the conversational paths are actually relatively finite. And now it’s up to smart people like you guys to figure out what those paths are, and be like, “okay, can we be relevant for all of them?”

Greg (30:58)
Well, let me piggyback onto that, because it brings us back to the sinks-and-mirror conversation. What you guys illustrated are these very specific requirements — pieces of information people want to know before they book a hotel: the full-length mirror, the sink configuration. People have these requirements, they’re just not often fully expressed, because — you couldn’t necessarily do that; you had to do a general search, then go to a website, then dig through the website to see if that information was there. But what’s really important, I think — and some of our research reveals this — is you need to get that information out. This is sort of context and sort of not. You need to get that information out to Google, to the LLMs, in order to be discovered — because if somebody’s going to do something really specific, or take a conversation path that gets to a really specific point — like, “give me a hotel in this area,” “give me a hotel that’s rated such-and-such,” “give me a hotel that’s rated such-and-such and under this price threshold, and is next to…” — right, all of that —

Mike B (32:07)
And that has a nice bed and soft pillows, and two sinks, and a mirror.

Greg (32:12)
— and then your wife gets to, “I need to know it has a full-length mirror.” Well, that’s a kind of a deal-breaker that most of these hotels and data aggregators and search engines wouldn’t necessarily think about — I mean, you might think about it if you saw the keyword, the query logs — but this is one of the big changes versus search: you could sort of optimize against a narrower set of things — good reviews, all the fields filled out, hours, and so on. I’m not an SEO, but — versus AI, which I think gives people many more options, many more conversation paths. The way you have to address that is by getting every single piece of information your customers want to know out there. You’ve got to understand what your customers want — Mike’s wife wants a full-length mirror — and get that information out there so the AI —

Mike B (33:13)
It wasn’t my wife. It was some batshit crazy researcher we had to watch the video for. Not my wife — my wife is not that crazy. It’s okay.

Greg (33:19)
Sorry, sorry, sorry, sorry.

Andrew Shotland (33:21)
So — I think you’re right, Greg, but I’ll counter it slightly: if “full-length mirror” is an option that, let’s say, fifty percent or twenty percent — whatever the number is — of customers find valuable, yeah, I agree. I think it’s impractical to deliberately go after every little use case, unless it’s just a data thing — like, we just need to have that attribute available. That’s too —

Mike B (33:44)
But this is why Google has both focused on reviews for the last fifteen years, and has now started asking for additional attributes in reviews — they know what people are asking. So when you look at what they’re asking other people to fill in on their reviews about a place, you can count on those things. So, in terms of where to find this information, at least at a broad level, I’d say: do a review of your local industry and see what questions Google is asking. Which also gives them a leg up in these answers as well, because — we’ve talked about this a number of times — they have the best local graph, and they keep making it deeper and wider. That local graph gives them a huge advantage, to the point where ChatGPT, even though they have a deal with Yelp, is still scraping Google’s local data.

Greg (34:33)
So I had a really interesting experience yesterday, apropos of this. I have an old Honda CR-V, 2007 — it’s about 20 years old — and it’s in pretty good shape, except the paint on the roof has been oxidized, the clear coat is gone, and it looks kind of gross and bad. So I’m trying to figure out if I can repaint the roof and how much that’s gonna cost. I’m asking ChatGPT to recommend a body shop, somebody that can do this affordably — ’cause the car is only worth like, maybe thirty-five hundred bucks — and I don’t want to spend two thousand dollars repainting the roof.

Mike B (35:06)
Isn’t that fascinating, though, that a twenty-year-old car is worth three thousand dollars?

Greg (35:10)
Well, it’s in good shape. It’s in good shape. But, anyway —

Andrew Shotland (35:12)
The CR-V is a freak car — we have a CR-V, and people knock on our door like, “Will you sell me your CR-V?” And I’m like —

Mike B (35:19)
Right — I have a ten-year-old one, I looked it up, it’s still worth twelve K.

Andrew Shotland (35:23)
I don’t understand it.

Greg (35:25)
Yes — so now we segue into car talk here. But, anyway, anyway —

Greg (35:29)
So, what ChatGPT did — and I have never seen this before — is it asked for permission to search Google. It asked me to authorize it to invoke a browser and go search Google. It said there’s a lot of noise in the search results — I forgot the verbatim language — “Can I go search Google?” or something like that. I authorized it, and it opened, showed me the little browser, and went through this whole thing. It did the same thing for Yelp. I’ve never seen it do that before.

Mike B (36:04)
Claude’s been doing that for a while, through their skills.

Greg (36:07)
I’ve seen the demos, I’ve seen its capabilities, but I’ve never had that experience with ChatGPT.

Mike B (36:13)
I had a similar experience the other day, just as a note — I was doing some spam research. I needed to know if fifteen locations were registered with the Colorado licensing body. It went out, clicked on every page, did a search for every one of these fifteen with my permission, and looked up whether they were actually licensed with the Colorado authorities.

Andrew Shotland (36:33)
So — thanks for having me. This was great.

Greg (36:33)
There’s a lot more to talk about, obviously. We’ll do a part two in the future. In summary, there’s a lot going on — it’s a really great time to be an SEO, even though people don’t appreciate local. Thank you guys for being on the podcast. Everybody, thanks for listening — we’ll do part two, see you next time.

Mark Kabana (36:54)
Thanks, guys.

Andrew Shotland (36:55)
Thanks.