Artificial Intelligence is reshaping how search engines evaluate and prioritize content. In this episode, SEO experts Bill Hunt and Bill Scully talk about the new role of SEO in an AI-driven landscape.
They highlight the shift from keyword-based tactics to strategies that focus on semantic structure, user intent, and real-time content modeling. You’ll also hear why SEO needs to be integrated with product, UX, and engineering teams for long-term success.
If you’re involved in SEO, content strategy, or digital marketing, this episode will help you rethink how to approach search in today’s AI-powered world.
Key Takeaways
- AI is changing the way companies approach SEO.
- Google is applying a multi-dimensional filter to content value.
- Businesses need to rethink their content strategies in the age of AI.
- Understanding user intent is crucial for effective SEO.
- The rules of building content have changed significantly.
- SEO professionals must evolve their mindsets and strategies.
- Collaboration across teams is essential for SEO success.
- Content must be structured to meet AI requirements.
- Businesses should focus on delivering valuable information to users.
- Traffic from traditional SEO methods is still significant.
Catch the Full Episode
Youtube: https://youtu.be/8kJOcNKJGuU?si=7lklp03_ti90V2r4
Spotify: https://open.spotify.com/episode/65GcrZ9BqNRFPC0HSNgB6g?si=4iEBp1hdQWaeKgIC8bPBSg
Apple Podcast: https://podcasts.apple.com/us/podcast/seo-in-the-age-of-ai-future-proof-strategies-for/id1755789311?i=1000718682177
Show Note Links
The Google Indexing Purge Update
Stop Retrofitting. Start Commissioning: The New Role Of SEO In The Age Of AI
Why SEO’s Fundamentals Still Matter in the Age of AI
Speacial Thanks to:
Transcript
Bill Scully (00:05)
Welcome to episode 36 of the International Web Effectiveness Podcast, where Bill Hunt and I, Bill Scully, focus on the strategies and techniques needed to improve your web presence and manage global websites. In the last episode, we took a close look at what goes into deciding whether your company’s website should be multilingual or tailored to multiple countries. We broke down the main factors to consider—and why those choices matter when reaching a global audience.
Today, we’re shifting our focus to something just as critical: how the era of AI is changing the way companies approach SEO. We’ll explore what businesses need to rethink and how to stay competitive as search engines get smarter.
But first, I want to talk about a recent article by Adam Gent in the Indexing Insight newsletter. I subscribe to it, and I’d highly recommend it to anyone.
This particular article was about what’s being called the “Google Index Purge Update.” There was a lot of great information there. According to his findings using his tool—also called Indexing Insight—it appears that Google is actively applying a multi-dimensional filter to assess content value, not just content presence.
In other words, Google is no longer treating indexing as a neutral act. It’s behaving more like a curator, applying machine learning filters. I always go back to the idea of how AI might run multiple searches in the background, trying to determine the user’s intent—and maybe even predict the next question.
Bill, did you get a chance to read this?
Bill Hunt (02:08)
I did. I thought it was excellent. It actually motivated me to finally finish an article I’ve been working on—it’s a few thousand words and, honestly, more of a rant. What Adam showed revealed a couple of key things in his data. Since he’s using the Search Console API to pull data from his clients’ sites, he’s able to track changes closely.
Roughly 25% of the monitored pages in his system were removed from the index. It was eye-opening to see how many previously indexed pages had been dropped. He goes into detail trying to understand why Google might be removing them.
Like you mentioned, I believe Google is applying far more factors now in deciding whether it even needs a page indexed. And I’ve been saying this for a long time—at some point, Google has to make a judgment call: “I don’t need all this content.” It doesn’t need to index 500 versions of the same thing. I think this update shows that Google is becoming much more selective.
Bill Scully (03:31)
Yeah, I think this all ties back to the cost of indexation. Google’s running a business. With all the AI-generated content out there, a lot of it’s just copycat material. They’re trying to determine what’s truly important—and ignore the rest. I imagine they’re applying some kind of 2%/98% rule.
Bill Hunt (03:53)
Exactly. And when you scale that to Google’s level, it becomes massive. Most of us think within the bubble of our own sites. If your site has 10,000 or even a million pages, imagine how many more Google has to process across the entire web.
That’s what Fabrice often talks about—the enormous frontier of content out there that Google may never even get to because it’s bogged down with what it’s already crawling. What’s changing now is that Google seems to be using new criteria:
Do we need this content?
Do we already have something just as good elsewhere?
Does this content generate any revenue opportunities?
Is anyone even using it?
There’s old content that may no longer be relevant. These are the kinds of filters I tried to highlight in my article. We’ve even been developing a training module for developers to help them think like a search engine—how would you decide whether something should be crawled or indexed?
Adam’s work reinforces this. When he looked at the content that was removed, it correlated closely with pages that weren’t getting engagement or had little value. His Indexing Insight tool is a great way to monitor this. It shows what we all should be tracking—like whether content is indexed, how often it’s crawled, and how Google interprets those actions in Search Console.
He does a great job breaking down terms like “discovered but not indexed” or “crawled but not indexed,” and explains what they really mean. It’s an essential read. And I think it ties directly into what we’re discussing today.
The evolution of AI and search starts with indexing—or even with ingestion into a large language model. Speaking of that, slightly off-topic, but related—Cloudflare recently announced a new tool that can detect bots and block them unless you’re willing to pay for access.
Bill Hunt (06:50)
Yeah, it caused a wave of reactions across the internet. Some were cheering, like “Finally! Keep those bots out!” But others were more cautious: “Wait, don’t we want bots to access our content?”
Now we’ve got an intermediary acting as a gatekeeper for roughly 20% of web traffic, essentially saying, “We can block access or set up a toll booth.” That’s a big shift.
Bill Hunt (07:19)
It was just interesting the way that unfolded. It reminded me a lot of my usual rant about crawling and indexing. I created a model that we’re using in this training session—basically asking: what if OpenAI creates their own rubric to decide, “Do we even need this content?”
Is it already available from someone who hasn’t gated it? Is it available from someone who’s not suing us?
Do people even want it? I think it was a 17-factor rubric to evaluate whether it’s worth the time and cost. And what’s that toll worth? I made an analogy to surge pricing, like what we have in New York. If you come during peak times, you pay more—same thing here.
Bill Scully (07:48)
Right.
Bill Hunt (08:15)
It’s just like what we’re experiencing here in Connecticut with our electricity. God forbid you run the AC on a 100+ degree day—you pay more for that electricity. But anyway, that tees us up perfectly for the next part of the discussion.
Bill Scully (08:16)
Yeah, yeah. A really interesting article—we’ll have a link to it in the show notes. Also, I came across a note that Instagram is apparently no longer blocking Google, even though Google was already crawling their content.
From a gating standpoint, that idea might work well for publishers, but how do people actually pay? What’s the process? If I were running a crawler and needed access, how would I go about paying for that? There’s just so much complexity—plus, not everyone is using Cloudflare.
Bill Hunt (09:24)
Exactly. That’s the whole thing—can you bypass it? Sure, Cloudflare has 20% coverage, and they named a bunch of companies using it, but then you’ve also got Akamai, and others who don’t use a CDN.
Bill Scully (09:27)
Right. And there are plenty of people hosting the same information not on Cloudflare. So, the content is out there.
Bill Hunt (09:44)
Exactly. That’s kind of my take—it’s not the panacea that people are making it out to be.
I got a few negative comments like, “Why do you always have to poo-poo everything?” Apparently, we can’t have nice things without me bringing people back to reality. But I think it’s worth having that conversation.
It’s like the rant I wrote on fair use from a legal perspective—Meta actually won a case because, interestingly, the courts found there wasn’t a mechanism for publishers to sell access to AI agents. So maybe that ruling triggered Cloudflare to do what they’re doing now.
One of my arguments was: now that there’s a mechanism, could this potentially overturn what the judge said about fair use? At the time, there was no toll to pay—no way to buy access. But now?
Bill Scully (10:47)
Right. It could reverse the decision.
Bill Hunt (11:06)
Exactly. These are early days. That’s the craziness—the whole gist of this discussion. Things are changing rapidly in this new AI era. It’s like the Wild West again.
I think we have six or eight things we want to walk through today—stuff I think people really need to be paying attention to.
Bill Scully (11:30)
Let’s get into it. You just kicked off a new series of executive thought leadership articles for Search Engine Journal. The first one’s titled “Stop Retrofitting, Start Commissioning: The New Role of SEO in the Age of AI.” We’ll link it in the show notes.
What I really liked was the approach—shifting from being reactive to proactive.
You’re kind of a writing machine. I don’t know anyone who can have an idea and publish an article by the next day. What inspired this one? I don’t know how long it took you, but you always seem to have something live 24 hours after we talk.
Bill Hunt (12:30)
Honestly, I just stopped caring what people think. That was always my blocker. I’ve got hundreds of draft articles—like the one I mentioned earlier—I’ve sat on that one for two years.
But this particular piece was triggered by two or three things.
One was a call I had with someone where I was trying to explain how their content needed to change. And they were like, “Well, we’re tech B2B. We don’t need to change. This is how we’ve always written, and we’re sticking with it.”
But the world is changing around them.
Especially in B2B, we tend to gate a lot of content. It’s bait—lead gen, right? We’ll have this amazing white paper, give away two paragraphs, and then put the rest behind a form. All the good stuff is hidden.
It’s the same thing newspapers do with paywalls. And I was just saying, we need to find a better balance. Maybe not today, but definitely in the near future.
The second trigger was this HVAC engineer I know. He told me about a job on a local high-rise project. There was a guy—not quite a foreman—whose official title was commissioning agent. I asked, “So, does he get a commission for what he does?”
He said no. His job is to ensure that the HVAC system being installed matches the architectural plans and specifications. He’s basically a modern building inspector, but hired by the client.
He checks whether the venting, machinery, and systems—some purchased five years ago—are still compliant with current regulations. He also ensures the HVAC integrates with alarm systems, monitoring tools, elevators—basically, systems on systems on systems.
And I thought, Wow, that’s what SEOs have to do. If we’re allowed to do it right, our job is to make sure everything works together.
Bill Scully (14:50)
If they do it right.
Bill Hunt (14:55)
Right—if they’re involved at the beginning of a site rebuild, a migration, or really any major project. But often, we’re the poor souls brought in after the fact. That’s what sparked all of this. I used to say, as you mentioned, the old way was reactive. We’d get a note saying, “Hey, we went live last Friday—or two weeks ago—and now we’ve seen a huge drop in Google traffic. Can you help?”
And I’d think: You launched a site without any SEO preparation? That’s why we now push to be involved way in advance. Especially now, with AI changing everything.
Another catalyst was trying to explain to someone why consolidating their site from 40 pages—each dedicated to a specific feature—down to just two pages, where every feature is buried in bullet points, was going to be a disaster. Not just for SEO in the traditional sense, but also with how search is evolving. Like you said: query fan-outs. What’s the next question? And the next one?
By condensing all that rich content, they made themselves invisible. I told them: Wait a couple of weeks. Then go into an AI search engine and look up a feature related to your software. See what it returns. Then ask: “Why aren’t we part of the consideration set?” And the AI will tell you: Because we couldn’t find any reference to your key points.
So, the rules of building content have changed. To use a metaphor: we bought equipment five years ago when we were funded to build the building. But the world has moved on.
Bill Scully (16:27)
Yeah.
Bill Hunt (16:51)
So what’s the delta between what we planned for back then and what’s needed today? That’s the question. Businesses need to rethink everything. And SEOs—mere mortal SEOs—need to rethink it too.
We’ve talked about this in our Web Effectiveness Mandate for over a year now. That’s the core of it.
Going back to your query fan-out example, the old way was: open a tool, find popular keywords, and write content—sometimes even if it wasn’t directly relevant—then throw it up on a page and hope we’d rank in the top 10 blue links.
Now, that entire dynamic has shifted. Mike King has been vocal about this—pissing people off, honestly—by pushing for radical change in how SEO fits into the digital strategy. His Relevance Engineering approach is about completely rethinking content creation.
And it ties right back to what you said—query fan-outs. I love DeepSync for that. You can set it up to track thinking patterns. It will identify what a user likely wants to know next. You can literally watch it in action. It’s wild.
I joked in a post recently that I think I found my new job—writing snarky AI responses for people who actually read those things.
Bill Scully (18:25)
Yeah. You’re going to watch it run.
Bill Hunt (18:46)
Exactly. You can follow the chain of questions. Sometimes, I’ll ask: “Why did you give me this result instead of that one?” And it feels like the AI is thinking, “This guy wants to reverse-engineer my algorithm.”
But that’s really the game now—understanding how the system interprets intent. We should do a live session one day where we explore this.
Just by putting “best” in front of a search term, or specifying “for small business,” the system starts creating definitions. What does ‘best’ mean? What does ‘small business’ mean? It approaches the query holistically.
And most of us aren’t preparing content in a way that reflects that kind of semantic parsing. Everyone’s using the word “chunking” now. AI looks for chunks of content.
If I had to define it, I’d say it’s like the content you’d write to get a featured snippet: a clear question and a concise answer. Like an FAQ. Now imagine the AI pulling in your entire FAQ section at once to construct a response.
Bill Scully (20:05)
Yeah.
Yeah.
The high-level concept is similar to a topical map—the Kori framework. When I first started, I took the course three times. Each session was about 10 to 12 hours long. It was still confusing at first, but it gradually made sense.
I joined a cohort too. We’ve been meeting twice a week for the past three months. It’s starting to click—like The Matrix. Suddenly you’re seeing things in a whole new way.
And I’m not exaggerating, Bill—so many of the “aha!” moments people are having now are things Cory’s been talking about for years. He’s built on the work of others—like Bill Slawski and other patent researchers—but it’s all coming together now.
They’re focused on matching attributes, identifying entities, and understanding real intent. Then asking: If that’s the intent, what would be the next question? Does your content answer that next question? It just builds from there.
And if you do this right, you’re not just looking at isolated pages—you’re building a semantic network across your site.
That’s usually not the case with standard product websites. But for those offering solutions, alternatives, how-to content—it makes all the difference.
Done right, it works. A few years ago, I went deep into customer journey mapping. That’s how I started generating better content ideas.
Bill Hunt (22:08)
Right. Exactly.
Bill Scully (22:33)
And this is the micro-level version of that. When someone does a search, Google is essentially tracing the customer journey and returning the most relevant answers.
Bill Hunt (22:44)
That’s really the key, right? It’s about the intent. I actually wrote an article about this yesterday, following a webinar I did last week. I didn’t get to say everything I wanted during the session, so I tried to include it in the article. I believe there are three key shifts happening right now—and that really gets to the crux of it.
The first shift is this relentless focus on answering user intent. Why are users making this query? I have a series I haven’t posted yet that was originally going to revisit things I said 20 years ago—ideas nobody cared about back then. I softened the tone and called it “what’s old is new.” The fundamental question remains: when someone makes a query, why are they asking that? What are they really looking for?
Like you mentioned—the customer journey and these query chains—we were modeling this at IBM back before 2000. I think I mentioned it before: I worked on a project for Sony. This was probably around 2000. We couldn’t find a single pathway where the user journey from searching for a TV down to a specific model was actually represented.
Bill Scully (23:48)
Yeah, we’re—
Bill Hunt (24:14)
—and all the queries in between. Someone could literally type “TV” anywhere on the site, but nowhere was that mapped to a Sony model. And I’m not trying to pick on Sony unfairly here, but what about all those steps in between? Users start to realize there are different kinds of TVs—plasma, flat screen, smart TVs. Things we take for granted today.
Bill Scully (24:16)
The pathway as the navigation.
Bill Hunt (24:43)
Exactly. The dynamics are different now. Users wonder: do I want this feature? That function? And we’ve always talked about those micro-moments—when one query leads to the next and the next and the next. Very few have mapped those chains properly. Corey’s done a great job with that—he deserves props. But honestly, these are things Mike and I talked about back then. We had entire chapters about searcher intent in a book we published in 2005.
So that’s the first shift. The second is this: companies are now focused on removing friction and delivering information. If you want mind share, market share, and user share, your platform has to be functional. Why did Google displace all the others? Because it gave people what they wanted—quickly and effectively.
Bill Scully (25:46)
Yeah, they’ve really protected the searcher. I mean, their whole business model is built around the number of ads shown. They never want users going anywhere else to search for information, and they’ve worked hard to keep that control.
Bill Hunt (26:05)
Exactly. And that leads to my third point. Back in the day, if you asked, “What time is it in Reykjavik, Iceland?” Google wouldn’t give you the answer directly. They’d list links to websites that had the answer, and you had to click through.
Eventually, someone realized, “Hey, this is calculable—we can just show the answer directly.” And suddenly, 10,000 websites that did nothing but post things like “What time is it in Reykjavik?”—based on your server’s IP—lost their purpose. Google began asking: what other friction can we remove?
The third point is about monetization and data harvesting. They realized that if they could gate users—keep them inside Google’s ecosystem, or now within the ChatGPT ecosystem—they could extract more value. With advancements in AI, computing power, and speed, it became easier to monetize this behavior.
One example I always bring up: I once searched for “Thai restaurants near me,” and Google suggested one 78 miles away.
Bill Scully (27:55)
You—
Bill Hunt (27:56)
Why did Google show that as a suggestion? And why did that restaurant show up instead of one just a mile away? Because that far-away restaurant used schema markup with longitude and latitude, making it easier for Google to calculate location. The closer restaurant had its address inside an image—meaning Google couldn’t easily parse it.
So, Google promoted the one 78 miles away over the one nearby because it had calculable data. But think about what that required: they needed my IP, my longitude and latitude, the restaurant’s coordinates, and the computing power to measure the distance. Today, that’s easy—but it wasn’t always.
Now, because computing is faster and cheaper, all these things are possible. Google can gamify search, hold users in place, and shape the experience. But here’s the bigger question: the sites that lost 30–40% of their traffic because Google shows answers directly—did they actually deserve that traffic in the first place?
Bill Scully (29:20)
Right.
Bill Hunt (29:22)
I’m not going to argue that today. But what I’m getting at is these consumption patterns. We need to make sure that we’re opting… So your point that a product website is textbook perfect, where we probably have four or five bullet points, maybe a paragraph of our claim to fame, and then that’s all we have. This is why retailers and affiliates do well — because people actually want more than that. They want reviews.
This is why Reddit does well — because they want real people’s opinions. They might be gamified, but they want genuine opinions. For example, I mentioned at PubCon that I was looking for a specific sort of functionality around a Lenovo laptop. Lenovo wasn’t in the top 10 results because they didn’t answer the question. Everybody else did because they took the data points to the next level. They got out of their way of having bullet points. So that was the point there.
Then the next one was really this idea: how does all this happen? We need to have grown-up conversations — cross-functional integration. So now the product team — let’s go back to your product example — the people creating the product information need to talk to the SEO team to understand how AI is treating this content. They need to talk to the tech team.
Something as simple as this — I talked to a guy recently who said, “I had a bunch of my products drop out of Google search results.” I said, “Give me one of the products.” I went to the page and looked at his schema. He was using product schema, and it said “out of stock.”
I said, “If you’re telling Google that your product is out of stock, why would they show that to people looking for the product? Because if they show the schema saying it’s out of stock, they’re never going to click to begin with.” He said, “But they might.” It’s the mantra of every store owner in the world: “But they might.”
Bill Scully (31:40)
Yeah.
Bill Hunt (31:53)
And the people — I mean, Google may index it, but they know somebody sees it and says, “Well, can we take that schema off?” So you still don’t have the product. How can they buy a product you don’t have, but the traffic might buy something else? So again, that’s our very selfish, myopic view of the world instead of saying, “What if I do have a similar product? How do we make those connections between A and B?”
Bill Scully (31:55)
Yeah.
Bill Hunt (32:24)
All right, and that’s why these teams need to talk. The merchandising team needs to talk to the product team, to the web team, to the team injecting schema. I mean, the idea — what we were talking about in the webinar with Milestone the other day — is this idea of multimodal.
I know you deal with this with clients who want images of their products to show up, images of how their products are used, or videos showing their products being used.
If you just think about this — my son told me about a query for a project he’s working on. He can’t use the word “used,” you have to say “pre-owned.” The query was “pre-owned Rolex Submariner Boston.” We unpacked that query. “Pre-owned” means he doesn’t want a new one; he wants a used one. “Rolex” is a specific brand; “Submariner” is a specific product name or brand within a brand. And “Boston” means he wants to know who has it in the Boston area.
So everybody outside Boston proper has been excluded. Except in paid results, because the guy in California thinks somebody in Boston is going to buy an $80,000 watch online. Maybe. But again, Google’s now stacking the deck.
If you do that query, look at what comes up: a block of paid ads, a local pack, then four rows of shopping carousel ads — that’s a shopping feed. So are you in the shopping feed? Are you in the local pack? Are you in paid? Well, there was a paid ad block with shopping ads, then two organic listings co-mingled after the four rows of the shopping feed, showing more similar products.
So if we step back and think, “How do I get in this whole mix?”
Bill Scully (34:04)
Yeah.
Bill Hunt (34:19)
Yeah, there was probably a left rail of all the meta filters, right? Like price range, color, weight, material.
Bill Scully (34:32)
Yeah.
Bill Hunt (34:33)
I’ve got to talk about images, colors, sizes — all of that.
This left rail is actually a great example. I was talking about it in an article coming up. When Google first introduced those pills at the top, I mentioned at a conference that the fact they have these pills means they know secondary, third, and fourth queries of how people are drilling down. If not, they look at query variations to know what people want.
The fact that they have that left rail, with all those meta attributes, it’s like, come on.
Bill Scully (35:36)
Yeah.
Bill Hunt (35:39)
We need to have a grown-up conversation with the people who produce the content, create the pages, and write about it, because that game has now changed.
We need to make sure we’re triggering the switches to be in the left rail. We need to make sure everything is synchronized.
When I looked at this result, there was only one store that appeared in both. Yet there were all these different modes, if you will.
So we have to think about this multimodal approach, multidisciplinary approach.
Coming back to multimodal and multidisciplinary, another big thing people have been talking about is this: you’ve got to be found, you’ve got to be authoritative, you’ve got to be the marketer on steroids.
What’s interesting is if you do that query to validate the statement that you need to be more authoritative to be in AI search results, all the articles Perplexity references come from PR firms saying you need more PR.
So again, it comes down to consensus bias skewed by people who benefit from making you more findable and marketable. I’m not poo-pooing that at all; I totally agree.
You mentioned Instagram — we have to be doing the Instagram thing, and other social channels, so that if the context warrants it, we’re there.
Again, it’s a team sport, which you and I have talked about forever.
Bill Scully (37:33)
Yeah. I mean, unfortunately, it’s a lot of work. If you’re selling parts cleaner chemicals, you need to cover: Will this consistently work within my process? Is it safe for my team to handle? What’s the total cost of ownership? Is it going to eliminate the source of our failure? Is it really a solution? Is it something that will replace the chemical I’m currently using? And you almost have to write for all of the different personas.
Bill Hunt (38:18)
Right. The analogy I would give is somebody walks into your physical store and talks to Bob or Sally at the counter. We need to somehow suck the knowledge out of Bob and Sally’s brain and make it available for people to access. Like you said, this can be, in one regard… I think I’ve made this sound like a daunting task, like you could just close up shop and be done with it. We can’t do this. That’s not what I’m saying. We just need to be smarter about it.
I think smaller businesses may be slightly more agile than bigger ones because if people are looking for a solution and you have solutions, you could offer what you’d normally spend ten minutes explaining in the store. For example, what if you use something like Riverside or Camtasia? You come in and ask people a list of common problems. “Hey Bob, I’ve got green algae. The motor on my pool pump sounds funky.” How would Bob answer? Then you take that transcript, clean it up, and put it on the web. It costs money in theory, but if someone else does it, why can’t you?
Bill Scully (40:06)
Mm-hmm.
Bill Hunt (40:16)
You set your people down with these tools and have them do a brain dump, then shape it into some form. I’m not saying do it for every little detail, not 30 years of Bob or Sally’s experience, but what are the top 10 questions people ask at your counter? What’s the top 10 emails you get?
Bill Scully (40:22)
Yeah. And what’s their follow-up question?
Bill Hunt (40:41)
Exactly. That’s the chain of custody you were talking about earlier. If I ask this, does it trigger the next one? How do we structure this to navigate people through it? And don’t get me started on how poorly built chatbots are on websites. I just tried one, and the terminology they used about canceling my Norton antivirus subscription didn’t even match their own language. I had to Google how to cancel or downgrade because their fancy, probably million-dollar chatbot didn’t have that info. Now, is that on purpose? Maybe they didn’t want me to unsubscribe or downgrade. Foolish on their part, because I had a nasty rant about that.
Bill Scully (41:16)
You—
Bill Hunt (41:39)
Exactly. This is where we’re going. The last point I want to make is about evolving skills and mindset. We need to think about this content more deeply, about how these processes work. We don’t know everything yet—nobody does—but we have enough info, especially stuff Corey and others are putting out, to know we need to start thinking differently.
And finally, we need to think about measurement. How do we measure this? Traffic alone won’t be a good metric. Are we being mentioned? I’ve criticized tools recently about personalization because my query and your query for the same topic will be different, so maybe they’re not good metrics. But there are directional signals. Are we getting traffic from AI engines? Melissa Jensen from Cisco posted about extracting text URLs from Google results showing snippet segments. That shows AI bots are visiting.
But if you just signed up with Cloudflare and got blocked by default, you wouldn’t see that traffic. We need to look at this and expect that we might lose some traffic, but can we tweak our answers in AI or chat tools to improve engagement? I know you’ve done things where you include a call to action buried in the answer, like “click to see the bigger list,” to boost engagement. Getting creative about that is key.
Bill Scully (44:00)
Yeah.
Bill Hunt (44:04)
That’s the genesis of the article. We have to play as a team, evolve our mindset, understand why this is happening. And where it makes sense for your business, you have to start moving in that direction. But don’t forget, Google still drives a huge amount of traffic through the old way. So don’t just say, “Google’s done, I’ll put all my money into paid media or TikTok influencers.” There’s still a lot of traffic to be earned.
Bill Scully (44:45)
Yeah, yeah. I mean, all of these things go back to—I don’t know what episode it was—but we talked about a “VP of answers.” I really liked that idea: someone whose job is to ferret out all this stuff and at least have a list you can work through when you have time.
Bill Hunt (45:11)
Totally makes sense.
Bill Scully (45:12)
All right, so shift from reactive to proactive, emphasize semantic structure over keywords, build AI-first consumption patterns—crafting clear, self-contained sections.
What are they calling passages? Right. So, you want to answer questions succinctly in short areas and then follow up with the related questions. You want to integrate across disciplines and teams, and measure what matters. I think a lot of people will need to think differently about KPIs.
Until everyone can get their AI KPIs into the search tools—some are doing it now, some are charging a lot of money—and there are other options out there, but it’s not unreasonable to look at impressions. I think that’s a good thing: the more impressions, the more Google likes you, and that’s something to focus on.
So, how would you tell people to evolve their skills and mindsets?
Bill Hunt (46:43)
Just really take in what we’ve been talking about and think about what’s changed, and how we adapt to this. Things like really looking at structured data—not just schema structured data, but structuring content exactly like you said: start at the top level, then move down into the details. Think about whether we need to write differently.
Share data with your teams about how this content is being scored and viewed. I don’t think people should get caught up in cosine similarity and all the math behind it—that’s overkill for most. But just ask yourself: is this saying what needs to be said?
The biggest change is thinking holistically. Are we creating synergies? The number one thing people should do today is use some of these tools to see how they show up in context and out of context. How are you being presented? Then ask: why wasn’t I there? Why wasn’t I in the results?
That answer will often tell you what you need to change—how you structure your data, prepare your content, and think going forward. Because if you don’t comply, you won’t be seen. Someone else will be.
And I think we need to get that message to management—the ones who can free up resources and change hearts and minds—that this is truly a team sport. That’s where we have to go next.
Bill Scully (48:48)
Great. Thanks, Bill. All right, If you’ve made it this far, we’d really appreciate it if you’d give us some ratings. We’re on LinkedIn, Facebook, and X. What we haven’t mentioned before, but you might want to know, is that we also film and post these on YouTube. We hope you’re finding value in our conversations. Please rate and review wherever you can.
We also want to hear from you. Your feedback motivates us and keeps us making these. Bill and I are here to respond—we want to know who’s listening. Just ping us—that would be great. Thanks, Bill. Until next time, talk to you. Bye.
Bill Hunt (49:42)
Thanks.
We’ll see you.