Beyond the CMS - A Content Management Podcast | Making AI Search Useful Starts with Better Content: Megan Andrews of Squiz

Useful AI search starts with content people and machines can understand. Megan Andrews, Head of Customer, U.S. at Squiz, joins Chris Bryce to explore the content work behind conversational website experiences.

Megan demonstrates Squiz Content Intelligence, including accessibility findings, topic-level AI-readiness checks and recommendations for contradictory, ambiguous or missing information. The conversation also covers DXP and headless tradeoffs, student feedback, human oversight and the limits of external answer-engine optimization.

What you will learn

How to weigh an integrated DXP against a headless stack and your team's capacity
Why real student feedback matters more than assumptions about AI adoption
How Squiz Content Intelligence groups content into topics and flags weak answers
Where accessibility fixes and AI-readable content overlap, without treating them as guarantees
How to test important user questions and inspect the source pages behind an answer
Why fresh information, human review and content ownership still matter

Chapters

0:00 Welcome to Beyond the CMS Episode 51
1:24 Megan Andrews and Squiz
2:30 DXP vs headless: team capacity and integration tradeoffs
5:57 Who Squiz serves: higher education and government
7:52 Thoughtful AI and humans in the loop
11:23 What students actually want from AI search
12:45 Synthetic personas and real-user blind spots
15:31 Conversational website interfaces
17:10 Squiz conversational search demo
18:30 Why AI makes content strategy more important
21:05 Content Intelligence accessibility walkthrough
23:06 AI readiness: contradictions, ambiguity and gaps
25:21 University of Montana search example
26:50 SEO vs AEO: control what you can control
28:53 Test real questions and trace the source pages
30:57 Related content, freshness and unresolved limits
34:21 Connect with Megan and closing thoughts

About Megan Andrews

Megan is Head of Customer, U.S. at Squiz, overseeing client expansion, renewals and customer growth strategy. Her work spans enterprise technology, SaaS and digital transformation.

Links

Beyond the CMS is hosted by Chris Bryce, founder of Dotfusion.

Creators and Guests

Host

Chris Bryce

CEO and founding partner of Dotfusion, a Toronto B Corp specializing in agentic content operations and AI-ready composable platforms. Host of Beyond the CMS.

What is Beyond the CMS - A Content Management Podcast?

Beyond the CMS is about content operations and the future of content: how enterprises create it, scale it, and get it found by people and AI. Hosted by Chris Bryce, Partner at Dotfusion, the show has just passed 50 episodes of conversations with the people building that future.

It started as a show about CMS platforms, and that conversation never stopped. Guests have come from Contentful, Storyblok, Agility CMS, Squiz, Bloomreach, ButterCMS, StreamX, BILDIT, CrafterCMS, React Bricks, and Magnolia, talking through what headless and composable architecture actually gets you. But the range has grown well beyond platform comparisons. Recent episodes cover Answer Engine Optimization (AEO) and how AI search engines decide what to cite, MCP (Model Context Protocol) servers that connect AI agents directly to a CMS, agentic content operations workflows, accessibility, and what it takes to govern content at the scale of a university managing tens of thousands of edits a week. Episode 50 brought on Olivier Dobberkau of the dkd/TYPO3 Association to dig into semantic search and retrieval-augmented generation (RAG).

The CMS is still the hub. It's just not the whole story anymore. If you're a CMO, IT Director, or Chief Digital Officer trying to figure out how your content gets made, published, and found, by people and by AI, this is where those conversations happen.

#contentoperations #answerengineoptimization #aeo #aisearch #headlesscms #contentful #storyblok #agilitycms

All right.

We are live, uh, for episode 51 now of Beyond the CMS.

I'm Chris Bryce.

I'm the founder of Dotfusion and a person who just loves talking about

content and the operations behind presenting and delivering it, and

I've had the fortune to host wonderful minds who are also thoughtful and

care about the subject matter as well over the last three years or so.

Um, always like to talk a little bit about where the idea came from, and that

was from listening to the Car Guys on NPR, as people would call in and not

know how their carburetor works, and it was like, "I had no idea either."

And, uh, I just like the down-to-earth style of explaining stuff, uh,

thoughtfully to, to people.

Um, yeah.

Well, uh, enough about me and it.

Uh, let's talk about you.

Our special guest today is Megan Andrews.

Uh, we're gonna need to talk about this, but you are the head of

customer at Squiz, which is an awesome title, um, and an awesome platform.

So warm welcome to you, and, uh, let's get right into it.

Maybe just tell everyone now and, and into the future, this lives on

YouTube later, um, about yourself and Squiz, and I know you got a whole

bunch of cool stuff to show us too.

Yeah.

Thanks, Chris.

So excited to be here.

Um, so yeah, I'm head of customer here in the US for Squiz.

Squiz is a digital experience platform, and we've got all sorts

of different types of tools that can plug in and help you run your website.

Um, and I know we're gonna talk about some of them today, which is exciting.

Um, but my background has been primarily in brand and content.

I've worked in lots of different types of industries, so financial services,

craft beer, one of my favorite, um, and now I'm squarely sitting within

tech itself, SaaS, um, technology.

So really fun, uh, to explore.

I think there's always been a lot of creativity within the tech space, um,

obviously, but mashing that up with my brand backgr- ground has been very fun.

Um, and I'm excited to dive into some of the topics we're gonna cover today.

Super cool.

Um, I'm just gonna ask a few, um, questions as we go along that, that

I've collected from our customers and our prospects along the way, and what

I try to do, uh, on the show or the podcast here is explain things in kind of

layman's terms, b- kind of acknowledging the technical nature of stuff.

But one thing I'd like for kind of, uh, the future is your opinion on the

difference between kind of a DXP, uh, and kind of this headless pure stuff

that we talk about because I think that's a de- decision that's a bit

confusing for a CMO or a marketer type person, and we encourage them from the

industry to, to check them all out, and you should know about DXP, and you

should know about headless and legacy.

Uh, how do you kind of describe Squiz itself and its unique spot?

Yeah.

This is something people have to figure out with their teams, their

setup, everything, and there, there's slightly different choices, right?

Um, typically I really try to simplify it in some way.

That, um, in a headless environment, what you often get is a lot of different

things you can build and control, and you, uh, but you've gotta build them.

Mm-hmm. So there's this challenge where you end up quite often with lots of

developers and sort of, um, plugins and things that you're, you're

fitting together around what you need.

Um, and so in that way I imagine it actually takes quite a bit

of architecture of your own.

How do we design this for what we need?

Um, I think of Squiz as a platform that's fairly out of the box,

um, for marketers and developers.

Um, there's a lot that's already set for you.

Um, and you have the ability to layer across as much of the technologies you

want with the systems you already use.

Um- Mm … so you could host everything you need to do within

our digital experience platform.

Um, but if you have your own CMS you're using, that's fine.

You can still, um, use our integrations to plug into a CRM, um, or use our-

Mm … uh, AI tools for conversational search, um, which is keyword search.

So I think that the way I simplify it is the tools are already built and

ready, and they're being updated and built, um, without effort on your side.

Mm.

Um, and I do know that when you're sitting in a seat sometimes, especially

with some of the customers we work with, where they've got small teams and they're

resource constrained and they, they don't have endless budget, um, sometimes those

types of systems are really helpful.

You don't have to- Mm … have a team of six devs who are keeping

every- keeping the lights on with all your tech Yeah, good points.

We, we love Squiz for that reason, uh, and others.

One, we can explain the cost easily.

Yeah.

Right.

You know? Sometimes when you're composing a whole bunch of stuff, there's this

last minute where, you know, if we're building a pure composable thing,

the customer's like, "Well, like what do, what do you mean you don't know

how much the search part's gonna be?"

Uh, you know, for example.

Yep.

Um, and also it's nice to have the kind of systems all bundled and have gone through

the test of time that they work together.

Right. You know, bolting things on often has little rifts in the, in the,

uh, initial connection type of idea.

So, uh- Right … yeah, I think that's a super cool aspect.

Um, you were showing me a bunch of stuff a bunch of weeks ago,

and just before we hopped on, on the call today, that, um, show the

commitment to, I guess, your roadmap in developing the product internally.

Of course, there's a lot of AI stuff going on in there.

Uh, just before we get to that, can you, can you tell everyone

who do you typically serve?

What's the, the right type of client that's, you know, typically

looking for your type of solution?

Sure.

Okay.

Great question because, uh, we have s- what I would say expertise

and deep industry experience in two areas, although, um, it's

enterprise software that could serve- Mm … any type of organization.

Um, so where Squiz specializes in higher education and government.

And, you know, Squiz is headquartered in Australia and has essentially

the vast market share across those two industries in Australia.

Um- Right … and then Funnelback is probably the most well-known, the

search tool, keyword search, um, most, most well-known product across the US.

Um, and similarly, we have a lot of higher education and government customers.

Um, that being said, we also work with a lot more services, uh,

companies and organizations now.

Um, you can imagine, um, larger nonprofits or law firms.

Um, you know, there's, uh, you know, banking institutions.

We have credit unions we work with.

So there are a number of different types of orgs, um, and companies that fit.

I think the, the connecting thread is you need, um, enough technology

to do some sophisticated things.

Um, but oftentimes, you know, you know, when you get into some of the bigger

tech, um, with huge capacity, you're talking about something like Adobe,

um, sometimes that technology can be overwhelming for a team that's, you know,

a little bit smaller in size, and they can't leverage everything that's there.

Um, and it's expensive.

It can be expensive.

So I think there's a sweet spot of like great tools that work well for a team,

um, a middle-sized team that still needs enterprise, um, functionality

across their, their tech suite.

Okay.

So, do you wanna tell me your, the organization's,

um, kind of approach to AI?

How are you adopting it?

What's your position on the future?

That we put it in our title that it may be kind of, um, adding some

excitement to people's careers and their jobs, and et cetera.

Um, what's your position on all of this?

You know, there's, I think, a couple different things that Squiz is doing

that I really appreciate, and it aligns with how I think about AI.

Mm-hmm. That, um, it, it's worth being thoughtful in our approach.

And, uh, you know, I have friends, I'm sure we all do too, and I saw this

interesting article about this, but they'll just throw together something,

some generative AI, like a, a poster for their birthday party coming up.

Or this article I read was really interesting.

It was about, um, this, this person who had a friend who had, like,

showed her a poster he made about his dog who had recently passed away.

But it was all these, like, it, you know, it was AI generated, and it

was all these, like, um, pictures of him, but they were AI, and it was…

He was so taken with this poster.

And she was kinda like, "Oh, no, I don't feel a connection to this."

Um, and it was because she realized that, like, it wasn't some person who had

taken the time to thoughtfully craft it.

It had just been put through a machine and, and put, you know, rolled out.

And so she was talking about the importance of being in touch with,

um, outputs, and the process of creativity and art and emotion.

Uh, so I think about that on some level where, um, there's this piece around AI

that we don't wanna lose the connectivity to ourselves, and the, the human in the

loop is actually a really important thing.

Let's keep humans in the loop, and in mind.

Um, and at the same time, you know, this is where I love what

Squiz is doing with their approach.

Make sure it all makes sense for what we're working on and what we're doing.

Um, you don't need AI to run wild.

Give it great, um, you know, logic and prompts and, uh, direction to help us

with what we actually need to do, and I think that's what I'm seeing with

the tools we're coming out with now.

It's getting, um, prospective customers or students to outcomes

faster in a more personalized way.

Um, and there's, you know, there's no random dog poster being generated along

the way, and you can trust what's coming out in terms of the content itself.

Oh, you're on.

If you don't mind me, um, just sharing a, a story.

So I was at this little cafe the other day and I see this amazing poster

of the items on their menu, and they had this breakfast sandwich, and

it, like, had three eggs on it and about, like, a half a pound of bacon.

I was like, "Are you serious?

This is what…

My holy grail."

"I want that thing."

And I knew in the back of my mind it had that AI polish.

It's, like, kind of this awkward, slightly tart vibe to it.

But anyway, it's like, "Okay, maybe you guys are just rocking the sand."

And I just got this typical kind of like egg McMuffin kind of thing.

And I was like, "Dude, you…

Like, you had to realize that the gap and what that magical thing is, and

what I, what I got is a bit different."

Yeah. Um, so I'm, I'm totally with you.

Yeah.

Um- Yeah.

It's a great sidekick.

Yeah.

I think it's a great…

Yeah.

There's, there's so many ways we could be using AI, so I think the

challenge is always, "But what, what's best, and what do we need to

be doing, and what's not necessary?"

Um, especially, I'll be honest, you know, I've done some student focus groups and

sat with customers who are like, "Hey, can we bring in 14 students to talk about our

website and the functionality they need?"

Um, it's interesting talking to younger folks right now because I

actually put an AI tool in front of a focus group two weeks ago.

Didn't preface it with anything, didn't explain what it was, just

put it in front of them and said, "What do you guys think about this

type of functionality on the site?"

And it was a conversational interface.

You'd think more like ChatGPT style.

Mm-hmm. Like, you could drop in a question.

Um, and, uh, the students, a couple of them at first were like,

"Well, I don't really like AI."

Like, a couple of them just said that straight up.

They recognize and they're like, "I don't know."

Um, and then we started interacting with the tool and they…

And connecting it to, like, the, the challenges they'd had across the site-

Mm-hmm … where they're like, "Well, yeah, sometimes I couldn't find program

informations on the site, so I would just leave and go to Google and do

a search for it and find it there."

Um, and you know, see it in the summary and click to the page.

Mm-hmm.

And so as we started interacting with the tool, then they kind of came around and

were like, "Well, this would actually be a really cool feature to have on the site."

Mm-hmm.

Mm-hmm. So it was this interesting philosophical sort of wall at times, but then when

you realize it can be functional f- and helpful in your own life, I think

that's that thoughtful application of like, "Yeah, actually this

would be really valuable for us."

Mm-hmm.

You just gave me a great insight.

Um, we're working on a project, it's a large migration, a

bit of modernization as well.

20,000 pages of stuff.

Um, we're at about 10 different personas.

Mm-hmm. Um, you know, we realize that we're not going to be able to satisfy all those

personas, but what we've done as kind of a, an experiment, first time for us

maybe, I haven't heard of this before.

But we've…

We're actually using some AI tooling where you can create your own bots, and

bots have souls, behaviors, patterns.

So we've actually created a synthetic profile In the form of a bot and we've

given it you know where it lives, how old it is, all those kind of like- Yep

… things that you'd give to a persona.

And one of the personas is kind of 25-year-old person, and

I'm just wondering, with all the research we've done, if…

I'm curious to see, is it gonna pick up that that persona needs to be

disinterested to some degree in AI?

Um- Yes … good point, and thank you for that, 'cause, uh- Yep … that's

why we need humans in the loop.

Yeah, there is some level of resistance somewhere there.

Yeah. And it's interesting.

We'd…

I talked to a customer very recently who's doing something similar.

They had created similar personas.

I think there's ac- there's, like, platforms out there now that we

can do this, and they were doing some user testing, right- Mm-hmm

on the new site they were designing using these personas, um, which is fascinating.

You don't have to go through the expense of talking- Yeah … to

actual users, or the time, um, but how accurate are they actually gonna be?

Maybe pretty close, but you might miss some…

You might have some blind spots there.

Yeah, yeah.

That's really interesting.

Yeah.

Um, okay, we've got lots of stuff, uh, to talk about.

I wanted to touch on, uh, SEO and AEO, uh, your really cool

content intelligence stuff.

Yeah. Um, are you feeling comfy to fire up your screen and maybe walk us

through some- Yes … amazing things?

Let's fire it up.

Okay.

Okay, cool.

Um, I've got a couple things that I think we can look at together today.

Um- Amazing … one is…

Now, we were just talking before this about how it was election week in the

US, and so this is particularly timely.

Um- That is true … so I don't know how many Americans have even

seen this, but this got on my radar.

Oh, one sec.

For some reason, I've got disconnected from your tab or your screen.

Oh.

If you can just hit some of those buttons down there, we're gonna, uh, fire this up.

And this is- Can you see it?

… going to be a great example that I'm going to share with other folks too.

Okay, all yours.

You're, you're live, and we're seeing your screen.

Perfect.

Okay.

So america.gov,

this is the actual website as it- Yeah … is live right now.

Again, as an American, I didn't even know that this was here.

Um, and it's fascinating to me because, uh, y- I don't know that we're always

used to our government being on the cutting edge of technology, but, um,

this is a completely conversational search-driven natural language query site.

And obviously everyone knows they can drop in here and type something.

What I'm interested in is this little piece here, um, 29,000 websites in one.

Moses, Moses, eh?

Yeah.

It's…

You start to realize, especially as a, a government or a decentralized,

um, large university, this is sort of exactly the challenge.

Mm. There's so many different parts that you have to bring together.

Um, so in this way, you know, it's trying to ease that user friction, where I

wouldn't even necessarily know where to start if I need to find out a benefit

of mine from the federal government.

And this takes all the guesswork out.

You just drop it in there And I think that's probably where

we're headed, is that there's…

We're all so used to finding a canvas that's almost like a

billboard, and we click around and we navigate where we need to go.

We actually think our brains now in terms of keywords and not just normal

sentences because that's how we've been trained with SEO- Mm-hmm … and search.

Um, but this is an evolving and very real example of, um, what, what websites could

be looking like in the next few years.

Very interesting.

Yeah.

Cool. And I will say, like, Squiz has implemented this type of

technology, you know, on our own.

If we didn't have it on our own site, who would we be?

All right, yeah.

Um, so there- there's lots of ways to start to integrate this type of thing.

You know, uh, America.gov

went all in.

On the Squiz site, it's just a simple embedded search bar here.

So you can still have that, like, framework you're used to, the

traditional way of getting around with information architecture.

But I don't know, is my content ready for AI search, Chris?

Let's, let's see.

Um, so right here within the interface itself, it's gonna

give me a- um, an answer.

Awesome. And yeah, so I've got my answer over here, and what we're working on now

is this concept of kind of dynamically generated content that relates to

whatever the search is about, right?

So it's almost like a dynamically evolving website for you.

Amazing.

So, yeah.

Um, so this is the type of thing that we imagine, you know, again,

connecting users more quickly to the outcomes they actually want.

They can see all the sources of where everything came from, you know, do the,

uh, the check to make sure it seems legit, uh, but then otherwise get on their way

with what they're trying to accomplish without having to do too much- Yeah

to get there.

So it's not…

It's an interesting point, 'cause I do have some of these conversations

around, like, the end of content.

Mm-hmm. But it's actually almost the beginning because we need that

content, although it's not…

We- we don't rumble around and search through it the way we used to.

Yep. It's gotta be there, that knowledge base- Oh, yeah

needs to still be there, and that needs to be thoughtful.

It needs to be semantic.

It needs meta.

It needs robot versions, human versions, short form versions.

Very interesting.

Yeah.

I think, um, the content strategy has only become more important

and more interesting to me. Mm-hmm.

The, um…

Instead of thinking in terms of, "Here's a menu item, and here are the six things

that need to be on the mega menu, and then we're gonna funnel people along

to this place," um, you really…

I, I actually think you get to be quite imaginative.

It, it's a little bit liberating to think, "Well, maybe it doesn't actually matter

too much where it lives on the site.

It just needs to be there."

Maybe that's really simplifying it down.

Um, it has to be readable with context, right, for large language models.

But I was talking to a university in, in New York this week, and I

was asking them about enrollment and how things are looking.

Um, they said, "You know, we're, we're kind of stepping up year

over year with enrollment.

We don't have a, a crisis by any means, but our types of institutions,

a small Christian college, there's fewer and fewer of them.

Um, so we're, there are fewer options and at the same time, we don't know if,

how people are gonna find out about us.

Um, what we do know is once they're here, they love it and they stay."

They have like, you know, a 90% retention rate.

And it was so, in that moment I said, "This is really fascinating."

Like, I wonder how much storytelling content could help

you all really drive, you know.

What are stu- what keeps students there?

Mm-hmm, mm-hmm.

What if you had a lot of that type of content that lived on your site, which

maybe you don't normally click through all of those, um, but when you get, it

gets consolidated into someone says, "Why should I go to this university?"

And the large language model can scan across all that great storytelling

and great, give a compelling answer.

That's really interesting in terms- Mm … of a content strategy.

Very cool example.

And is this what you call content intelligence?

Is that the, the product itself?

It, it- So this is sort of like all the things together.

Oh, cool.

So we've got, um, yeah, content intelligence, I would say

is the foundational brick.

Right. So that's a tool, and I can…

I'll sh- I'll actually pop this open right now.

I've got it- Awesome … ready.

So this is the backend look, okay?

This is what it would look like if you're gonna be in the tool itself.

Yeah. Um, now this isn't a, a live example, but it's, um, it mirrors it.

Um, and so what we do with content intelligence is we look at the

accessibility aspects of the site and the AI readiness or content piece,

because they go hand-in-hand, right?

Like y- a large language model can't really read your

site unless it's got, um…

it's all structured the right way, your metadata's all there.

Um, and accessibility is actually a great way to make sure that's true.

Good point.

Um, so you can scan your whole site with that content intelligence tool, and it's

gonna give you an output of everything that's going on in terms of accessibility.

Um, so it's gonna give you suggested fixes.

You can even, um, choose if you want it to be like a content fix or if it's

maybe a developer that's required.

Um, and this is the type of, um, you know, recommendation it's gonna give you.

Like, "Hey, there's 54 pages affected by this.

It's a critical issue.

Developers are gonna need to be involved to fix it."

Um, and it's gonna show you the current implementation, and

then give you a recommended fix.

So, you know, very helpful, useful stuff.

Um, accessibility is top of mind obviously for a lot of organizations

who have to be compliant legally.

Mm-hmm. Um, but this added benefit of being compliant and accessible

makes you very readable for large langu- language models as well.

And can this…

will this autocorrect itself?

Like, can you approve that and then it just does the fix?

It…

Yes, if you're across, uh, the, the CMS that Squiz has as well, right?

Amazing.

Yeah, yeah.

So if you're using that, it's a one button, go ahead and fix it.

Um- Like a self-healing system.

Yes, exactly.

Yeah, right on.

Okay.

Yeah, so some efficiencies to be had there in terms of how

your team manages the site.

Um, I'll take us back to, let's see, the AI readiness side.

So again, still the content intelligence tool, accessibility on one side,

and then AI readiness on the other.

So this is where the tool…

So it's a large language model that, that's, that's scanning the site and

saying, "What do I see across your site?"

It's actually gonna group together, based on your content, what it identifies

as the main topics of your site.

Hmm.

So maybe you're gonna have student life that it found, undergraduate programs, um,

and it gives you the option to then deep scan or monitor each of those topic areas.

And based on what it sees there, it's gonna tell you how many issues it's found,

or if it's just optimized and ready, um, that might be holding you back from being

optimized for large language models.

Nice.

Um, so an example might be, um, and this is something that would

take forever in a human audit.

Um, let's say that you listed your pho- your phone number for student services

in one area of the site, and it's this.

But actually, student services moved and there's a different

phone number and it's on- Oh, yeah.

Yeah, yeah.

Yeah.

So it'll find contradictions.

Uh, it'll also flag when things are ambiguous.

Um, so where the model is like, "Hey, I understand some context,

but you could provide a little more and I can understand it better."

Mm-hmm.

Um, and then areas where, um, you know, there might be something totally missing.

It's just like, "I don't have any information about this."

Um, so it flags all of those for you, gives you recommendations to

fix them, and then eventually you get to a place where you're like, "Okay,

all, all of our site is optimized.

The content's ready."

And that can just stay there and live there, right?

That, that means that any large language model indexing or crawling

your site is gonna understand it.

Um, but that's that l- you know, foundational piece that then you

can parlay into the front-end piece, which we've showed you, which is that,

um, conversational search interface.

Right.

Right. Very cool.

So you know the content's ready to go.

Anything that's being pulled to generate this response is from that

content intelligence optimized content.

Okay.

Yeah.

That is very cool indeed.

Cool.

And there's more.

Um, I will show kind of another, um…

This is, like, a live example, and it's always fun to see those.

Oh, great.

University of Montana is great, and they embrace digital, uh, transformation,

and they do some great things.

It's no surprise that undergraduate admissions or admissions in general is

a, a really valuable part of the site.

Um, you know, what I showed with our SQUIZ site is, like, fancy.

It's pulling in different components.

It's, like, dynamically evolving as the questions go.

This is a little more straightforward.

So what is tuition?

I'm just gonna ask this question And what I like about this is it's gonna

have your standard set of search results that load below, right?

So everyone's knows where they are based on where we've been.

They're like, "I can just go to the standard search results."

But they also have this opportunity to interact with the chat.

So it's layered up top.

You can keep asking questions, um, or you just go right down to

where you need to go down here.

Okay.

I get it.

Yeah.

So- Very cool … again, content intelligence is behind this.

It's driving, it's, you know, all the content on the back end, and then

that front-end implementation is just however you wanna layer it onto your

site and, and provide that experience.

Okay.

Oh, that's great news.

Yeah.

Um, how do you approach, uh, solving for, uh, like SEO and AEO?

It…

Everyone's got a slightly different take on, on that, and I think

it's a little bit more than just shoving some schema data on a page.

Yeah.

Um, yeah.

Where, where's, where's the collective mindset on solving for that?

Yeah, it's, um…

There are, there's enough overlap.

I often think of a Venn diagram with this, where it's like- Mm-hmm

hey, some key principles of SEO are never gonna go away.

Um, all that great metadata and, and content matter, um, these things

will still be true for a while.

Mm-hmm. Um, what is different is I think we know now how SEO works.

There's an algorithm.

Mm-hmm. There's a ranking system.

We can see keywords that are maybe more available to us

to get higher rankings with.

Um, there's, there's actually some actionable things you can do around SEO.

Mm-hmm. Um, AEO doesn't feel like that, right?

Like- Yeah, it's a bit mysterious.

It is a black box.

Mm-hmm. Um, okay, chat, how you finding this stuff?

No idea.

Yeah.

Um, so unless or until there's more transparency around what's

happening there, um, what we say, at Squiz in particular, is basically

control what you can control.

Mm-hmm. So the content on your site, that's your authoritative source, and do

what you can to influence those- Mm … large language models.

But right now we don't know what the full game is to consistently

show up the way we want to there.

Yeah, it's interesting.

I was thinking, like, do we just run, like, a thousand bots all night long

telling GPT that Dotfusion's the best?

Like, I don't know.

Yeah. Maybe it's as simple as that.

Could be.

I think about, like, the, the artists on Spotify who release a song- Mm … and

then have the similar sort of like stream this, the one song to get listened.

Mm-hmm.

Mm-hmm. Eventually they find you and shut you down.

Yes, yeah.

I'd rather go through the struggle of, uh, of keeping the high road, you know?

Yeah.

At the end of the day.

But it's fun to, like, experiment with these things and- Yeah … I

think y- like, you had shared this really cool idea, you know, with me

a while ago about, um, the Squiz tooling will actually create questions

and then, and then assure that…

Like, it's kind of profiling your customer's questions, so- Yes

it can spur you to create content that answers them.

And I thought that was just a, a cool program, 'cause it's like,

yes, AI, think about my persona.

What are the hundred questions that they ideally are asking that they

should find our site, and how do we prepare content to answer those?

Uh- Yeah … like, that was a cool, uh, functionality of the built, I think that's

built-in tooling with your system, right?

Yeah.

It is.

And that part's, that's a really interesting piece.

So I, I'm torn on that one, Chris, because I almost, as a, as a marketing

and content person, I know I would've been looking at that and reading every single

question and response out of the top.

Right, right, right.

You know, thousands that are generated.

Mm-hmm. Like, is this right?

Is this accurate?

And I don't want people to think that that's what they need to do.

Um- What is more helpful, and there is a part of the tool where you can do this, is

come up with the top 20 or 100 questions you know are consistently being asked.

Right.

Yeah. Look at your keyword search, like the, all those terms, the queries.

Like, what are the advisors' questions that they're getting frequently?

And then ask the tool for a response on those.

You can just type it in and ask, and it'll show you what it

would currently respond with. Yeah.

Um, so yes, there's a database it's forming that's a database of facts that

it can reference, and then layer on and test it yourself to know, like, oh,

this response is what we expect, and it's accurate, and we're good to go.

Mm-hmm.

Um, the best part is when you do that, it'll show you exactly

where it got, all the source pages where it got the information from.

And if the response is not passing the sniff test, you can be

like, "Nope, this wasn't right."

Yeah. Which of these four pages didn't, doesn't quite get this right, you know?

Yeah.

Interesting.

Mm-hmm. Um, to what extent, um, and we're just playing with, with some of these

ideas, but we're doing them kind of with a multiple set of tools.

Can, can Squiz itself figure out that when you have a new…

Let's say you're posting an article or a paper or, or a post of some kind.

Can it intelligently kind of understand previous posts that should

be kind of linked to the new one?

Um, that, uh, just something that, that we're just thinking about out loud.

And then when we post the new one, if frequency and updates

are important for a post, can we somehow automate the injection of

a new post link in an old post?

Mm.

I know we're, we're just kind of playing with this kind of stuff

from a curiosity perspective.

Um, and I know you're kind of agentically powered tooling as well.

Yes.

Yeah.

Um, I have an answer to that that might not be fully

satisfying because it's evolving.

Um- Mm … right now the biggest piece for how the tooling

works is content and context.

So it will identify, like a large language model would, oh, these are