SEO’s New Frontier With Duane Forrester

Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
SPEAKER_00 (00:02):
This is the unknown secrets of internet marketing.
Your insider guide to thestrategies top marketers use to
crush the competition.
Ready to unlock your businessfull potential.
Let's get started.

SPEAKER_03 (00:17):
Howdy, welcome back to another fun-filled episode of
The Unknown Secrets of InternetMarketing.
I am your host, Matt Bertram.
Uh, what an exciting time welive in.
It's uh 2026, and it's like thebirth of the internet has
happened again with AI, and AIis working its way into
everything, and AI has reallyaffected marketing in a big way.

(00:38):
We were probably one of thefirst disrupted industries in
that.
And um, I wanted to bring on aspecial guest.
This is going to be kind of likea masterclass, guys.
So uh get your get yournotebooks out there.
Uh, I have the one and onlyDwayne Forrester uh on online.
So, Duane, welcome to the show.

SPEAKER_01 (00:56):
Matthew, thank you.

SPEAKER_03 (00:57):
And for anybody that's uh been been in the
industry a long time, sportsguy, um, you know, he's he's cut
he's been doxxed and he is outthere uh uh and he is standing
by his name and his work, andhe's done some great things for
some big companies, uh likeBing, Yelp, um, you know, sports
betting.
Was it MGM?
Was that was that who you'rewith or who was it with?

SPEAKER_01 (01:19):
So I started my career with Caesars Palace
Casino.

SPEAKER_03 (01:22):
Caesar Palace, that's what it was, right?

SPEAKER_01 (01:23):
And then um moved on to sports betting, which was not
them, um online, and um ended upleaving that company, went to
Microsoft for a decade, moved onto YXT, and you know, I've been
deep into AI since then, andit's been eye-opening,
invigorating, frightening,exciting.

(01:47):
Um, there are several moreadjectives that go in that line,
and and and I'm sure yourlisteners have several of their
own to add there as well.

SPEAKER_03 (01:55):
Yeah, that thank you for the correction.
Yeah, yX.
I'm looking at the back of yourbook here, right?
Um, being, of course, and uhreally schema.org when you
helped launch that, I thoughtthat was uh huge.
And that's where um, you know, Istarted hearing your name.
And so um I am I'm just superexcited to have you on.
Uh, I've been going through yourbook, I got all kinds of

(02:18):
questions or kind of deep divesof things that I think would be
helpful for other peoplelistening because I can tell you
when we started talking aboutvector embeddings and chunkings,
um I certainly started to feellike I was outside my scope of
like, okay, I'm gonna have tolearn some.

SPEAKER_00 (02:36):
So that feeling.

SPEAKER_03 (02:38):
So some some some new things because this is not
what I was taught or or anyother words.
I was like, what is that?
So I need to do that.
Like, what is that?
And so um, you know, I one ofthe things to kind of tee up
this conversation, and we cantake it wherever we we talked
for for a while prior to this.
Um, but uh uh in his book onpage 149, um uh basically we

(03:02):
were talking about what it takesto be an SEO today and like
where that that's moving to andthe new skill sets you have to
have.
And and and really, you know,when I look back at being an SEO
at working with a largerorganization, you don't have a
lot of control, right?
Like so you have so you're yourcertain lane or your certain
inputs, but you can't impacteverything.

(03:23):
And when you get in to whereit's at today, it it has to be
um part of the DNA of thecompany to move the needle.
Um, and and I've seen a lot ofSEOs get frustrated at bigger
companies and try to kind ofteach everybody everything and
then they move on, and then thenext company doesn't understand,
and like with AI, with entities,right?

(03:46):
Yeah, like there's no like youcan go do this over here and
this doesn't impact it overhere.
And so I think the role of anSEO, whether the title gets
changed, which I know there's alot of debate about that, but
like needs to be elevated andthey need to have more control,
and and they they really haveall the data, which is shocking
to me when you talk to liketraditional uh mediums or uh

(04:09):
marketers, like they love allthe data.
And so so I think SEOs need toelevate themselves to kind of uh
have full visibility of what'sgoing on.
Um but one of the things yousaid, and I want to read this
for everybody, uh the the therequired dedicated expertise
today requires optimizing forretrieval confidence and

(04:31):
chunking structure, not the sameas optimizing landing pages for
human conversion.
Understanding vector embeddingsand semantic similarity scoring
is not the same as running akeyword gap analysis.
Tracking brand presence andzero-click AI answers is not the
same as monitoring rankings andtraditional search.

(04:52):
So basically, which I'm seeingit with all the tools, you gotta
throw out what you knew to acertain degree and come up with
maybe new KPIs, or as you kindof put it, there's an additional
layer on top of good SEO.
So it AI doesn't protect youfrom bad SEO that you have tech
debt you gotta make up.

SPEAKER_01 (05:13):
Yeah.

SPEAKER_03 (05:14):
But you know, but but now you gotta you gotta you
gotta go that extra mile.
There's that extra um uh likesphere of influence that you
have to tap once you do thebasic stuff or the foundational
stuff.

SPEAKER_01 (05:26):
I'll I'll I'll put it to you this way, Matthew.
The title of the book is themachine layer, and it's not
accidental, right?
Like, you know, you you've readit, you're part way through the
book.
Um, you know, at some pointyou'll have to tell me if you
actually think it's worth themoney you spent on it, because
I'm I'm dead curious.
The authors don't we don't likewe don't get to get that
feedback, you know.
Um, but one of the key pieces ofthis is, and I talk about it in

(05:50):
the book, right?

SPEAKER_03 (05:50):
It's like so Dwayne, leaving you a review, okay, on
Amazon, and then and then doinga semantic citation with us
together for the eat experiencewith me in the picture of the
book is actually worth it.

SPEAKER_01 (06:08):
So you see everyone, you see what we're doing here,
right?
Like, you know, um, but it'sit's like and I I frame it in
the book like you know, you'regoing from high school to
university, you're going fromyou know, um, a graduate degree
to a master's degree.
Um, you can put whatever exampleyou want on it, but but you're
leveling up with what we'redoing today.

(06:30):
Now, look, you know, thespokespeople at the engines will
tell you differently.
The engines themselves seem tobe putting out different
information right now, possiblycontradicting.

SPEAKER_03 (06:40):
Yeah, I still have to do that.
And it's a moving target, right?
Like it and it keeps, yeah.

SPEAKER_01 (06:44):
This is the point, okay?
It is a moving target, it is anabsolutely a moving target.
So, um, like I built somethingin this book that I wished was
here two years ago becauseeverything is moving so fast,
and and this is a challenge.
Like, like I did a survey, Ilooked at jobs that were posted

(07:07):
for SEOs, and I took, I don'tknow, the better part of 300
jobs, and I looked across everyone of those jobs, and I, you
know, basically came up with 2%of those jobs are talking about
AI in relation to SEO.
And I sat and I thought tomyself, so if these other people

(07:27):
are hiring someone, let's justsay their intent is that two
years from now that employee isstill with them.
Um, where exactly is this persondeveloping these skills around
understanding the AIenvironment?
And I'm purposely avoidingtalking about optimizing for AI

(07:47):
because there's there's a veryclear step of understanding how
LLMs operate, what they do, whatthey don't do, that is it is
different than the easyunderstanding of search.
Like everyone alive today knowswhat search is.
Even though a billion activeusers daily use ChatGPT, they
could not accurately tell youwhat it is and what it does in

(08:11):
the background.
You need to know you've gonethrough this, Matthew.
You've taken a couple ofclasses, you're about to get
some certification that I'm kindof a little bit jealous of.
Um, like you're but you're doingthese things, and you're doing
these things on a veryfast-moving train.
So, like, it doesn't stop atevery station.
If you want to get on, it'sdifficult.

(08:33):
Like, you've got to run and leapand hold on, and eventually you
find your set, your yourfooting, and you're like, okay,
now I can understand this.
Then you start to see how itapplies to everything you knew
before that, which brings meback to a very important point
that you made.
Um, you can't suck at SEO andexcel with the AI environment.

(08:56):
And I'm speaking like as abusiness, if your SEO presence,
if your execution of SEO is notgood, better, or best, if it's
mid-level or lower, you are notgoing to perform in AI answers.
It's that simple.
It'll be a rare viral momentthat you get called forward.

(09:17):
Whereas people who actually havea very good solid foundation,
very good solid reputations,relationships, they are going
deep on things.
Um, I released a um, I releaseda um we'll call it a recently.

SPEAKER_03 (09:35):
Is that what you're talking about?
The substack?

SPEAKER_01 (09:37):
So, yeah, this I I released a substack, but I gave
my my readers a specific processfrom a white paper that came out
in November that was veryspecific about looking at
product in AI and ways toinfluence the product being
brought forward.
And they identified a bunch ofthem.

(09:57):
I created an eight-point planfor it and said, hey, if you're
a subscriber, you get it, it'sfree, here you go, right?
So it's there if anybody wantsit.
Um, but my point behind this isnot that anybody should go get
that.
My point is we're now startingto see the very beginning of a
white paper based on amethodology that's shared and

(10:18):
there was a test and there'sempirical data behind it.
I don't know whether it's rightor wrong, good or bad, if it's
repeatable.
I'm telling everyone, you shouldgo try this and tell us if it
works.
Because I can tell you rightnow, without that, a lot of
people are guessing their wayforward.
And and if you're being toldchunking doesn't matter, geo

(10:40):
equals SEO, SEO equals geo.
Look, I'm here to tell you assomebody who worked inside one
of the largest search engines.
Um, it's my experience, andbased on my knowledge, that is
not accurate, that is not thedirection you want to lean in.
You need to understand thisyourself and how it applies to
you because it is different.

SPEAKER_03 (11:03):
So I want to ask somewhat you can you can give me
uh an open-ended answer, but Iwant to ask somewhat of a
pointed question because um Idon't know the answer, right?
So I I know that like qualifyingfor like fan out terms when
you're ranking, you want to rankin that top 100, right?
So you rank in that top 100.

(11:23):
And I did see some data early onthat there was a correlation
with the um the AI bots or thechat GBTs or of the world, like
whatever you want to call um thethe the the machine layer um of
which you know defining thatlike a search engine is not like
a big LLM and like what thedifferences are, I think might

(11:46):
be might be useful because Ithink that there's some some
kind of correlations of uh someof the formulas and things that
they do, but it is is uniquelydifferent.
But I thought okay, so you youyou're in the top 100 and you're
showing up, and then you'relike, okay, AI answers.
And then there's like this kindof direct correlation between
the different uh chat bots orwhatever of the answers they

(12:09):
showed.
And then I I believe it was itwas either HRS or SEMrush came
out with a big study that said,you know, 21% or something like
that of all the callbacks areare not even indexed.
Okay.
And so we know that Google andother search engines are not
respecting like do fall, likethey're sucking in all the
information and trying to make adetermination about it.

(12:30):
And I'm assuming that the youknow, the the AI is doing the
same thing, it's got access toeverything.
And it's like and it's makingassumptions.
So one of the things you eventalked about too was how do you
measure influencing throughthrough a framework or something
like that, how the AIsunderstand and process it, but
without giving you the citationfor what they're doing.

(12:52):
And and and I I I like that, youknow, that that was very
interesting to me.
And that's you know,trademarking and you know,
there's there's there's ways tokind of claim that territory,
but that's not something that Iwas ever taught or was taught
about when we're talking aboutSEO.
And and now you're going, well,as an entity strength, like the

(13:13):
trust signals that are comingup, how do I, how do I know if
I'm gonna show up there or not?
And and you got these tools thatare now you know scraping it
daily and telling you like inthe average, but it's like you
gotta under you gotta understandto your point how these things
work because to give you a tipand a trick on how to influence

(13:34):
it and optimize it, which itseems like it's a moving target.
So what's working now is notgonna work.
They're getting smarter.
So if you're spamming everybody,like that's only gonna work for
so long, and then you're gonnabe holding the bag going, like,
I don't know what happened.
Like it's worth it to now todayto just do the work the right
way and kind of claim that landand get in that long-term data

(13:55):
set.
Um, but but then I'm even seeingwith the grounding with like
Chat GBT when it started saying,like, hey, like so so it has to
be very, very current uharticles that are something like
uh 10 months or is what I read.
And so there's all thesedifferent factors that you know,
I mean, but then now the AIs docall really authoritative stuff

(14:21):
that are hugely old, right?
And and Google does the samething.
So there's a lot of likecompeting information, and it's
like, and it's so so I mean, Iwould love like okay, to show up
in the AI answers, you know,what's what's like a frame in
which people need to just thinkabout it, maybe in kind of a
short snapshot.

SPEAKER_01 (14:43):
Okay, so um look what you're describing is is you
know the norm, right?
Like in the absence of clarity.
Um, first off, let me describeit this way.
We are back in the Wild West.
Uh, that's where SEO started.
And if you've been an SEO formore than five years, you will

(15:03):
recognize the feeling.
You'll recognize the feeling ofyou've gone from essentially
walking down a corridor withartwork and lighting and plush
carpeting and chairs to stop andrest in, because this is the
structured world that you livedin as an SEO working with a
known entity, which is yourdirection.

(15:24):
And somewhere along the way, youopened a door and you stepped
out into a desert, and there isabsolutely nothing.
And some people thrive in thatenvironment.
They say, This won't kill me.
I have what I need between myskills and the raw entity of
this desert, I will be able tosurvive.

(15:46):
Other people look at that andgo, I need a checklist where
like I don't have everything Ineed.
Like it's and that is completelynormal.
That feeling is real, okay.
And in fact, that that feelingis the core premise of the book,
is about going from fear toactualization.
So, yeah, you're uncomfortablenow, but I'm gonna give you a

(16:08):
bunch of frameworks you can useto build comfort around.
And that's you'll get there asyou read it.
Um, but but the reality is thatlook, there's a lot of stuff
that you should have been doing.
We touched on like where you'redoing good SEO or you do
mediocre SEO.
Okay.
Um, one way that I would cutthat is do you have structured
data in place?

(16:28):
And you know, a lot of peoplecan raise their hand and they're
gonna say, Oh, yeah, I got it.
And I'm gonna go, okay, did youuse every piece of structured
data that you could across everyarea of your website that you
could?
Oh, well, no, I only used ithere.
You know what?
Mediocre SEO.
Good SEO starts at you've doneall of the work, not some of the
work.
And that's really what we'reseeing, right?

(16:48):
Is this leveling.
Now, I'm picking on structureddata because it's an easy one
and people understand it.
Okay.

SPEAKER_03 (16:54):
Well, I also want to throw it out there that people
are even like I know I know youropinion on it, but and I would
love to to deepen that becauseI've heard people waffling
saying, Well, I don't know ifthat's really important or not.
And and I've heard you say it'sabsolutely a hundred percent
important, but like I've seen iton LinkedIn even recently.
People are like, Oh, like, Idon't know, it's that, and I'm

(17:16):
like, it it's a it's a definedstructure that they understand,
and it can, and you're tellingthem, hey, this is what I want
you to believe, and then theycan go verify it.
It gives them like a startingpoint to to cross-verify it.
Like, I'm like, why would thisnot be important?
You're basically telling it,this is what I want you to
think.
Go believe me if I'm telling youthe truth or not.

(17:37):
That's what I'm here.

SPEAKER_01 (17:39):
Um, I have the unique position of having been
there when that was launched andunderstanding why and
understanding how it was usedand for what reasons.
If you want to skip it, I willsay thank you very much for
making my life easier.
Because if your approach is Idon't need it, it doesn't

(18:01):
matter, I'm happy to have lesscompetition.
Thanks for you know, steppingout of the race.
I appreciate it.
Um, but again, you know, you doyou.

SPEAKER_03 (18:12):
So like I what about uh I I just saw something
recently, and maybe it was uh uhuh clickbait, but it was like uh
uh what is lm.txt doesn'tmatter.

SPEAKER_01 (18:24):
No, so here's the thing um the problem with LM
LLM.txt, and I I just put out asubstack uh maybe a couple of
months ago on this, I did a deepdive on it.
Um, the problem isn't what it'sattempting to do or how it does
it or the wording or anythinglike that.
The problem is adoption andtrustworthiness.
So when robot.txt, schema.org,sitemap.xml, when those are all

(18:48):
launched, they were launchedwith explicit backing from the
search providers.
So the platforms themselves hada vested interest in it and
said, We will follow this, wewill agree with this, we want
that.
Not a single platform has comeon board.
This is a private effort onLLMs.txt.
It is someone who hastechnically their own vested

(19:13):
interest involved with it, but Idon't believe that's their
point.
I believe their point is we needsomething different than a robot
TXT.
And if no one's gonna create it,I will create this and give it
to you.
At one point in my career, Iwanted to know if I was being
paid well, so I started theindustry's first salary survey.
I had a vested interest inunderstanding if I was

(19:34):
compensated well, but then Iended up giving a survey to the
entire industry that was usedfor a decade to benchmark jobs.
So there you go, right?
Um, I get people do thingsaltruistically or just, you
know, not necessarily for theirown purposes.
And I think that that's wherethis came from.
Now, do I think that it hasvalue?
Not so much.
Um, none of the engines are onboard.

(19:56):
Google typically you hear thesecycles where they come up with
this, you know.
Oh, we don't back this, they sayit.
And if you look, there's a thereis a cycle, there's a timeline
around which these messages getrepeated to us.
And um, you know, Google likesto do things on their own,
right?
They always have, and for goodreason, like they have the
resources, whether it's people,intelligence, money, foresight,

(20:20):
whatever.
Like they they are well equippedto roll their own and just like
do their own thing.
So it shouldn't be a surprise toanyone that Google comes out and
says, Yeah, we're not gonnasupport that.
And and that's fine.
Um, it doesn't mean that it'snot valued.
And I would be shocked.
Look, if I worked for Chat GPTand they said to me, Um, hey

(20:41):
Dwayne, um, we have our owncrawler, um, it's got to go out
there and gather stuff, like,you know, how should it interact
with the world?
You come from search, you knowcrawlers.
What should we be doing?
I would be telling it to look atLLMTXT, and I would be gathering
every LLM TXT I could as theplatform, and I would be
creating an average signalacross thousands of those

(21:02):
instances.
And then I would say, which ofthese signals benefits me and
which of these harms me?
And then I would be in favor ofsupporting the things that
benefit me and ignoring thethings that harm me.
And so after that, I mightconsider telling the world that
I support this.
Problem being that if I come outand say I support it, everyone

(21:26):
hears that as a binary yes orno.
Even if you explain, here arethe 18 things I support.
If you say anything else, Iignore it.
For years, I battled theargument that um robot TXT was a
controller for crawlers, andeveryone was in this do crawl.
What's the proper do crawl?

(21:47):
What's the proper do index tag?
How do I write this?
And I'm like, you don't.
The assumption is the crawlerwill crawl everything, that's
its job.
It's the only reason it exists.
If you let it go on its own,that's what it will do, it will
crawl everything.
And some of those crawlers arestrong enough to break things,
they will crawl that hard andthat fast and that insistently.

(22:10):
And robot TXT is a way to say,Don't harm me, don't look in
here, don't waste your time,whatever it is.
But it's strictly a blockingmechanism, it is a no object, it
is not a yes object.
And for 15 years, the debate hasraged about oh, you have to have
do crawl, you have to have dueindex, you have to have these

(22:30):
things, and it's like, no, youdon't.
If you just put a robot.txt upthere with nothing on it, but
the file is there, you find allof the clue it needs, right?
Like literally an empty file ispermission.
So so we're in the samesituation.

SPEAKER_03 (22:46):
So just because to tie it all up, what did you ever
think of uh the humans.txt file?

SPEAKER_01 (22:52):
Um, listen, uh to me, this is like we're starting
to veer into territory of shouldwe debate geo or SEO or AIO?
Look, we got more importantthings.
I agree.
I agree just for the record.
Yeah, geo is used across 183industries as an acronym that
means different things.
It is not a good acronym for youas a search marketer to show up

(23:14):
at a meeting and start tossingaround because there is a very
higher than average chance thatsomebody in that room has a
different meaning attached toit, and you will create friction
in that room.
They will either mistrust you,they will be confused by you,
which is a form of mistrust, andtherefore you won't get invited
back, or they won't listen toyou.
Or best case scenario, everybodysmiles at you when they leave

(23:36):
the room, they go, dude, didn'teven know what Geo meant.
And everyone else at the companylaughs about it because they
know what it means to them.
Just be careful with that,right?
Like SEO might get confused fora Korean surname, but nothing
else.

SPEAKER_03 (23:51):
Yeah, I mean, all these semantic anchors in all
these different directions iscreating a lot of noise and a
lot of confusion.
And and I I feel like a lot ofpeople are debating on the the
the title of it, not actuallylike what's actually changing
and the work in it, you know.
Um I would love uh like kind ofgoing back to to to what we
talked about, maybe maybetalking about some KPIs, right?

(24:14):
Like what are the right KPIs?
Because the current KPIs oftraffic, um of uh keyword
rankings are thrown out thewindow in my book.
Uh I don't find much use in themanymore.

SPEAKER_00 (24:27):
Yeah.

SPEAKER_03 (24:27):
I mean, they're they're they're a point, but
they're not enough.
Like them on their own are nothelpful.

SPEAKER_01 (24:33):
So I'm just I'm calling this up right now,
right?
Because this is really importantthat we look at some of these.
A couple of caveats.
And I look, I outline all thisin the book, obviously, and I
talk about this in my substacks,right?
But like just for people thatare listening or watching the
show, um, some of the data thatwe are gonna talk about and we
are gonna say is important isnot available to you.

(24:55):
It is data that is only insidethe actual platform.
So proprietary to ChatGPT,perplexity, claude, Gemini.
And those platforms are notgonna share that, right?
I expect sometime on the otherside of never, Google will start
sharing this in Search Console.
Um, and I don't think that theother entities that I mentioned,

(25:18):
I don't think that they that itwill even hit their radar to
share anything.
So, like forget the theequivalent of Search Console or
Webmaster tools from OpenAI.
Just don't see that happening.
There's no upside to them forfor chasing that.
Um, but these are things that weknow are being used internally
at these platforms, right?

(25:39):
So, for example, semanticdensity score, okay.
Um correct me if I'm wrong,Matthew, but you brought up
semantic earlier, and you werekind of like, is it that big of
a deal?
Is it that different?
Maybe it's not that different.
That was the direction we wereleaning in at that moment in the
conversation as a questionaround as a question, you know.

SPEAKER_03 (25:58):
I I mean, I when Bert came out, and Bert's part
of my last name, right?
Right, you're already dug intoit, and and uh, you know, the
semantic spano is uh is is quiteimportant.
Um, and there's a lot of there'sthere's a lot of uh uh yeah, I I
think yeah, so keep going.

SPEAKER_01 (26:13):
So here is the important part when you hear
somebody say semantic densityand you hear them talking about
semantic search, those are twoentirely different things.
The word is the same because themeaning is the same, but
semantic density is actuallysomething that's looked at
inside these systems.
So within your chunk, are youactually deep in knowledge?

(26:36):
Do you provide everything I needwithin that chunk to understand
that topic, that question, thatanswer, that entity, whatever it
is?

SPEAKER_03 (26:44):
Yeah, we were yeah, we we were talking about like I
think when people are writingcontent, they they kind of
string it out where wherethey're they're given like
little nuggets throughout thearticle because they're looking
at the article in summation, butthat's not how AI looks at it,
it'll cut it off.

SPEAKER_01 (27:02):
Here, here's the problem, okay?
It's easy, it's easy for me andeverybody to sit there and say,
uh, yeah, no, that's not good.
Don't write long form, and andyou know, like you want to avoid
that, you want to chunk andblah, blah, blah.
And then it's super easy for asearch engine to sit down and
say, Don't do that.
That's not a great humanexperience.
And if you look at both of thoseon a bell curve and you put them

(27:24):
both out at the outside edge,you are correct.
Like both of those statementsare true on their own at the
same time.
Um, however, you can find thatblend.
Now, this is where, as a contentwriter, your job is not to sit
down with ChatGPT and say, Hereare my parameters, give me 500

(27:45):
words on this topic, and thenyou take those and edit those.
And at the end of the day, yougo, Wow, I created 11 new
articles today.
First off, you didn't create 11new articles today, unless the
definition of creating is file,save as, save as new title name,
then you created it.
Yes, I agree.
However, in no way is diminishedthe importance of the human in

(28:08):
this loop.
In fact, it underscores theimportance of the human in this
loop, okay?
Because your job is not whetherit's long form, short form,
bulleted points, a list, that'snot your job.
Your job is to understand I havean entity, and on this page, I

(28:28):
have identified nine entities,it's arbitrary number, it could
be any number.
Okay, if the page was a pageabout you, there would be one
entity, Matthew, repeated manytimes, but there would be a lot
of topics in relation toMatthew.
Okay, he is a competitive waterskier, he enjoys modeling in his

(28:53):
spare time, like all of thesethings are entities of their own
in relation to you, okay,association versus another page,
which is all of the greatestpodcasts that have SEO people on
them.
Okay, let's just say that numberis nine.
Each one of those, as they aredescribed, has to live in a

(29:14):
world where the descriptionabout that object has everything
that not only the human needs toknow about it.
Why did it make the best list?
Who are these people?
What is their background?
What are you giving me thatothers aren't giving me?
So on and so forth.
But you also, by doing that andbringing it forward for the
machine, you create a betterexperience for the person,

(29:34):
right?
The example I always go to isum, I go to this example on um
on um product search, okay.
Um, a few years ago, I wanted anew coffee maker.
And um I'm kind of a unique guy,like I want something different,
right?
Like I'm not just gonna run overto Walmart, buy a coffee maker,

(29:54):
and then like you know, a yearand a half later, run over to
Walmart and buy another coffeemaker because that one crapped
out.
Like, I want somethingdifferent, unique.
Problem is when you go lookingfor that stuff, what you start
to learn with coffee makers isdifferent unique means tall
glass pieces, and these thingsdon't fit under a counter, like
on your counter, but not belowthe cabinet, right?

(30:15):
Which then makes me realize Ineed measurements so that I know
how big your coffee maker is, soI know if it fits.
And you, as the manufacturer,give me the size of the shipping
box because that's what thefactory in China supplies to
you.
So that's what you put into yourmetadata about the product,
which then goes into everyproduct feed as the size and

(30:37):
dimension of the product, butI'm not putting your box under
my cabinet.
Uh, you can take three to fiveinches out of that for packing
material, and that's so I needthat data, and people don't do
that, they don't write that way.
Another problem that you fit oryou find is people write for
humans.

(30:57):
We write emotionally.
There's a lot of like trying toget a hook in there, right?
Almost exclusively, these hooksare ignored by machine learning
systems.
They are not emotional, they arenot nostalgic, they are not
interested in feelings, they areinterested in the facts, they
are interested in objectivity,they are interested in um a

(31:20):
concurrence of opinions.
So everybody loves it, everybodyhates it, that type of thing.
Um, but that to them is justmore data.
They don't actually believe youwant an example of this, okay?
Go to Amazon, go dive in on thereviews of a product that you
want to buy and read the AIoverview that Amazon provides

(31:41):
you, okay?
Because they tell their AI, gothrough all the reviews, come
back and tell us what it is.

SPEAKER_02 (31:46):
Yep.

SPEAKER_01 (31:47):
I would say 40, maybe 50% of the time, people
love it because of its height,and other people complain about
how tall it is.
It's like one, two in theirsummary, because a lot of people
complain about it and a lot ofpeople like it.
And you're sitting there going,This is useless because it's a
data engine.

(32:07):
It's just looking over a blockof time, a block of occurrences,
and saying, Love, hate, boatequal, okay, I should put both
of those in.
Here you go.
And then you, as the human, arelooking at it, going, You just
gave me two pieces ofinformation that contradict each
other.
What's the point of theoverview?
Like it, you know, that's yourreality.
So, so when we really get downto it, look, creating content

(32:30):
matters, still matters, willalways matter.
Who you're creating it for, yougot to pay attention now.
And to your point earlier,Matthew, you were kind of
touching on this.
Each one of these platformsbehaves differently, they all
value different things fordifferent reasons.
And the settings that they haveacross their systems, the
weighting and then thetemperature.
Okay.

(32:50):
So weighting would be your leftto right balancing, and then
temperature is your vertical onthat.
And everything is moving at thatsame time.
I can't do vertical andhorizontal same time.
Can't do it.
Um, so but that they're allmoving in that at that same time
in real time when it's beingcalled, and you, as the writer,
have to figure out how to createcontent that positions you to

(33:14):
slide both of those to themaximum so that you will get
included.
And then congratulations, you'vedone something you don't even
know you've done it because howdo you know if it's you they're
talking about?

SPEAKER_03 (33:26):
Well, you know, when I when you said that, like what
that made me think of is like II saw a a platform do this
pretty pretty, I thought it wasa pretty good tactic if you
understood what what the intentis is what you're trying to do
is in the long-term memory or inthe inf uh like the the the
operational heading, not justthe prompt of the AI, the

(33:48):
memory, uh to to get yourproduct in there, but their
recommendations are gonna bebased on all their past
searches, whatever it'sremembered about that person.
And so, you know, whatever, likeif you and I search the same
thing, you're gonna get adifferent answer than I, you
know, unless there's like noother choices in that category
of what we're looking for.

SPEAKER_01 (34:09):
So be careful with that because there are examples
of in very defined niches withexcellent content coverage, you
will see similar things poppingback.
So, like, look, a part of searchhas always been popularity,
right?
Like, and you know, whetherengines like to admit this or

(34:32):
not, time and resources getassigned to specific verticals,
okay?
News, weather, sports, all thesebig verticals where everybody
goes every day and consumes thatcontent, that is more important
than bedazzling roller skates.
As a vertical, I think we canall agree that even if we

(34:52):
balloon this up to just rollerskates, that's a much smaller
area of the internet with manyfewer people engaged in it than
the news.
So, so if it's you as an engineor as a platform applying
resources, and it's reallyimportant, the engine applies
human resources, engineers.

(35:15):
The platforms are applying AIsystems that cost tokens, which
translates to real dollars.
Either way, the company isinvesting money, okay?
So you can't say, Oh, it's justAI, that's cheaper than a human.
Ah, not at that scale, it's notlike it definitely is not.
And when you apply those things,you don't apply them all to

(35:36):
solve the problem for rollerskates, you apply it to solving
the problem for news and forfocusing in those areas.
So it's we're gonna see that.
We will see things overlap onthose odor edges.
That's my point.

SPEAKER_03 (35:50):
So, what I what I took away from what you said was
different categories, there'sdifferent amounts of resources
that are are meant to crawl theinternet and to also figure out
what's most relevant, like newson one end of it versus maybe
roller skates on the other endof it.
So if you're in the data forroller skates, right, and in in

(36:14):
in long-term memory, and you'relaunching a new product or
whatever, and you don't havethat authority, it could take a
lot longer to show up.
And there, I mean, there's amoat there that you might have
to do a lot of work before youstart showing up because a lot
of resources weren't applied toupdate that category.

SPEAKER_01 (36:32):
Is that is that I'm gonna redirect that, Matthew.
Yeah, and I'm gonna define thisa little better.
Okay.
It's not about some categoriesare slow, some are fast.
Okay.
If you have a problem and itaffects consumers of news,
you're gonna put resources tofixing the problem or refining
something in that categorybecause it's a scale issue.
It affects many more people muchmore frequently.

(36:54):
If that problem only exists inthe roller skate niche, you're
less likely to say, everybodyshow up and go fix that problem
because that one affects asmaller number of people.
The trajectory with which yournew product is the same between
the two of them.
It's not gonna matter.
Same core algorithm, it's thesame weightings and temperatures
and whatnot, right?

(37:14):
Um, it's not like okay.

SPEAKER_03 (37:16):
So there's not, there's not like you're
categorized here, and then no,no, no, no.

SPEAKER_01 (37:21):
And like way early, early on, when um, and this is
gonna stretch a lot of peopleback, right?
But like when Google was firstlaunched and and MSN search was
first launched, right?
Like they had verticals, right?
You would click on a tab and gointo search specific for news
for autos and whatnot.
At that time, you still hadgroups of people assigned to
work in those areas, and youmight have more people on one

(37:45):
team than another team based onvolume of traffic and problems
to solve.
Conceptually, you still havethat kind of verticalization
happening, but we've long agomoved, and this, you know, I
talk about this in the book,right?
Like, search is not just a whatif lookup table, it's really
complex, right?

(38:06):
Like, I encourage people to golook up what BM25 means and how
it's utilized, look up what BERTis and how it's utilized and
when it started to be applied.
You will see that the currentiteration of what I consider to
be traditional search is reallyadvanced.
Like it's really advanced,right?

(38:26):
No one's just gonna go out andcreate a search engine today.
Like, that's we're way beyondthat.
Um, it where we are intraditional search, it can
almost see LLMs from where itis.
The LLMs are an entirelydifferent animal, though.
What they do is entirelydifferent.
And in fact, you could actuallysay, and it would be completely

(38:48):
valid, I think.
Look, the LLM is not aboutsearch, it never has been about
search, it's been aboutinformation retrieval.
And if you want to argue thatsearch and information retrieval
are the same thing, okay, thenthese are the same things.
I argue that the execution isthe difference, not the meaning
of the word, but how you do itis what creates the difference.

(39:12):
Different process, differentmath, different requirements.

SPEAKER_03 (39:16):
Yeah, we were talking in the pre-interview, is
like LOM and machine learning,there was no resources put in it
because no one thought it wasgonna work.

SPEAKER_01 (39:26):
Well, look, you know the story, right?
Like 20 years ago, everybody wasclaiming they were doing it.
10 years ago, everybody wasclaiming it was doing it, and
everyone else was looking atthem going, You're nuts.
Five years ago, someone came outand said, We're gonna do this,
and a few people went, hmm.
Four years ago, boom, yeah,TPP3, and the world changed.

(39:46):
And and and suddenly all thosepeople who were nuts are all now
billionaires, multibillionaires, they're at the
cutting edge, they're you knowat the front end of everything.
They made Google play catch up.
I mean, I don't know what totell you about a movement.

SPEAKER_03 (40:04):
Well, this is just why this is my argument on why
it's not the same.
Because you you said it.
The the process is it it itmight give you an answer at the
end, but how it computed thatanswer that technology did not
exist before now.
So there's no way to say this isthe same thing as this, and I'm

(40:27):
just gonna do the same thinghere to do that.
I'm like, I mean, I've learnedhow search engines work and I've
learned how LMs work, and theydon't work at all.
Like, I mean, search engines yousaid were black box, but like
I'll tell you LLMs are a wholedifferent category.

SPEAKER_01 (40:43):
Look, I'll I'll put it to you this way um search
engine, and you could usewhatever you know verb you want
or adjective you want in here,right?
Like it forces, it encourageshumans to do work.
Okay.
A search engine means I stillhave agency because a search
engine says, Hey, look, you gaveme diddly squat for information,

(41:06):
but here's what I think you wantbased on that, because I did a
lot of smart calculations in theback end and and am I close?
And then you, as the human, go,well, hang on, let me read
through your list of stuff, andI'll pick one that I think is
the most enticing.
Okay, that's the engine forcingthe human to complete the final
mile.

(41:26):
The human brain is wired toconserve calories.
That's why we got smart as aspecies and started with tools.
That's why we had the industrialrevolution, that's why we have
LLMs today.
We are about preserving energythat in our core, in our genetic

(41:49):
makeup, we are programmed toburn fewer calories whenever
possible.
Now, might not seem like a bigdeal, but first off, your brain
burns the most calories per anyorgan in the human body.
So there you go, right?
It's big, it's thirsty, it sucksdown a lot of resources, and it
kind of knows if I can offloadstuff, I'll do that, right?

(42:11):
Like, you know, that's easierfor me.
It's a survival thing.
So search telling us to go lookthrough one or two pages of
links and determine in this messwhich one is the best.
I'm burning energy, I'm havingto think.
All I wanted was an answer.
Zero click comes along.

(42:32):
Now, all of a sudden, all ofthat stuff gets summarized, chop
right at the top.
Here's the answer to yourquestion.
Marketers panic, and here weare.
Guess what?
It's only expanded itsfootprint, it's continuing to
expand its footprint.
And LLMs are the next logicalstep in that.
I have a personal digital butlerthat I get to yell my request

(42:53):
at, and it comes back and goes,I got you, bro.
Here you go, here's your answer.
And it says it with suchauthority that the lazy part of
our brain kicks in and goes,I'll trust that.
That sounds confident.
Okay, first off, be carefulbecause what you've done as a
human being is you've handedyour agency on decision making

(43:14):
to something that does not evenunderstand your interest in the
moment, can't possibly, becauseyou didn't communicate it.
Because again, humans are lazyand we suck at communications.
So, like there's a lot going onthere, right?
And so it's really difficultthen as a search marketer to sit
down and say, Yeah, I trust theoutput.

(43:36):
I can use like there's so muchhere that still needs to mature.
A billion consumers a day don'tcare because they ask a
question, they get an answer,and they move on with their
lives.
And here we are.

SPEAKER_03 (43:49):
So the risk in decision making is a whole new
thread that I would love to godown out of the because that
that is where where I have a lotof interest.
To your point, like you you veryeloquently said it is like
people are like, oh, uh, LLMsaren't that much of search,
right?
Like, so just focus on SEO.

(44:09):
And I think everybody's givinglip service to it because it's
not that big of a percentageyet.
And when I saw what washappening, and like people
crossed that Rubicon, people arenot really going back.
Maybe they're going back toGoogle to verify, right?
So just uh okay, let me let medouble check, let me see if I'm
missing anything.
But but ultimately, most peopleare lazy, right?

(44:33):
Like if you just want to puteverybody out there in an
aggregate.
And so they're gonna go to this.
And so this is gonna eat search,like you wouldn't believe.
And when Google launched, like,what are they focused on now?
They're focused on YouTube,right?
Like that's like they becausethey know the mark.
And I felt like even AIoverviews were like just like

(44:53):
holding the water back ofeverybody moving over, like we
got to keep people here.
And then it's funny when I'vetalked to people over the last
year, they're like, Yeah, I'musing AI, I'm using those AI
overviews.
They're like, I'm using AI bylooking at the overviews, and
they are, I guess.

SPEAKER_01 (45:07):
You know, here's here's a universal truth that I
think everyone is going to cometo understand.
Um, so I don't think that chatGPT necessarily like you're not
going to see Google's marketshare slide and chat GPTs go up,
because chat GPT isn't trackedin the same vertical, in the

(45:27):
same bucket as a search engine.
So you can still get 90% of yourtraffic from Google, but if the
size of that search bucket iscut in half because the volume
of searches goes down overall,and all of that volume moves
over to an AI-powered system,you will still see that 90% of

(45:48):
your traffic is coming fromGoogle.
You will still see that theyhave 90% market share.
The problem here is that now themarket share in search is less
valuable.
And it's my prediction thatGoogle being every search
engine, they know this becauseyou can't fight human behavior.

(46:08):
Okay, look, it took Google nineyears to reach what I think of
as market saturation, okay,around the billion user mark.
Um, it took three and a halfyears for ChatGPT to do that.
So, like the world is a verydifferent place, okay?
And more and more consumers, youknow, I I relate this story
sometimes.

(46:29):
I had a friend of mine, he wentto um his daughter's school one
day, and he had to stand infront of the class and tell the
class, like, this is what daddydoes, right?
And, you know, he's like, Oh, Ihelp businesses get more
exposure on Google and rank atthe top of the list on Google so
that you know they're bigger,they're more successful, and
everyone makes more money.
And one of the kids in the classraised her hand and said, Can

(46:52):
you explain to me what Googleis?
I see my parents using it and Ihear them talk about it, but I
don't know what it is.
These are 10-year-olds.
So these are kids with phones,with devices, they're internet
natives, and she's 10 years old,really doesn't have a frame of
what Google is and the value itprovides.

(47:13):
That's a very real thing thatevery tech company is facing and
fighting.
So, yeah, AI overviews, stop gapmeasure until we find a way to
monetize AI overviews because westill need traffic and search
and those ads because that'swhere all the money is from.
And you slice it for whateverreason you want, right?
Like that's one reason.

(47:34):
There are a whole lot of otherreasons.
There's infrastructure, there'stech deck or uh um tech debt,
there's like all kinds ofreasons and rationales why it
makes sense.
There's also a lot of people whostill want to click on actual
links, so you can't just flip aswitch and make a move.

(47:54):
But what if you woke up onemorning and the entire industry
that you are attached to, a$22billion a year industry that is
so deeply threaded into everysingle business on earth now?
What if they woke up one day andsaid, We're changing direction?
We think that's more important,and we're all gonna go talk

(48:14):
about that and focus on that andnot do other things.
That look, that right there, redflag, flashing warning light,
danger, danger.
Like, I get it, it's very real.

SPEAKER_03 (48:24):
That that was my fear, right?
That was my fear, and I startedusing you know chat GBT and
going, like, I don't understandhow this works, I don't
understand how this is working,but I'm using it more and more
and more.
And I just every everybody'sgonna move that direction.
And and now you're talking aboutwhich we don't have time for,
but like the the bot economy,right?

(48:44):
Like these things are gonna havetheir own agency to be able to
do different things and buy andsell.

SPEAKER_01 (48:49):
And the whole, yeah, the whole agentic thing is like
look, if you think you're gonnajump into agentic, you really
better go back and scoreyourself on SEO because in order
to get to agentic, you have togo through AI.
And if you sucked at SEO, youstill got so much to learn and
do before you can grasp and wrapyour head around AI, which will
then prepare you for agentic.

(49:10):
It's not there aren't shortcutsanymore.
This is the wild west.
Go and learn, right?
Like you have to go and learnthese things.

SPEAKER_03 (49:18):
Yeah.
I mean, you you said everythingwas moving so fast, and you you
have to jump.
I mean, I'm doing trainingsalmost every night just to just
to keep up with what's going on.
I would love, I I know we'regetting close to time here, um,
but I would love to talk aboutvector embeddings.
I think a lot of people have amisunderstanding about that.
And and you you've done a reallygood job and and of how you talk

(49:40):
about things.
You're a great teacher.
So I'd love for you to kind ofshare with how people should be
looking at that.

SPEAKER_01 (49:46):
Okay.
For vector embedding, I wantpeople to think of a compass.
Okay.
Um, in theory, and we're gonnageneralize here because as
somebody who uses compasses inone of their hobbies, I I know
the discrete difference, butgenerally speaking, a compass
will point due north or magneticnorth.
So it's pointed in a direction.
Now, that compass is pointed onwhat you can consider to be

(50:10):
zero.
Anything to the left or right ofit is close, but not perfectly
lined up.
The only things lined up are onzero.
So if I give you a 0.1 or anegative 0.1, you know you're
left or right, you know you're alittle more or a little less
aligned.
That's vector embeddings.
Every single entity and objectis cre is taken, and we're gonna

(50:34):
separate vector from embeddinghere for a moment.
The embedding is themathematical representation.
So everything, Matthew Bertram,is converted into a number, and
that number then is stored in avector database, which is
different than a knowledge graphand vastly different than a
traditional database.
Okay.
I'm gonna throw some words atyou all, and I want you to go

(50:56):
look these up because I'm notgonna explain them, but they are
companies that will help youunderstand.
Weaveate, pine cone, super base.
They sound weird.
Go look that stuff up.
You're gonna explore a wholeother world, right?
It's it's a rabbit hole, so gethydrated and dive deep.
But the embedding is amathematical representation of

(51:18):
an entity or an object or athing or a word or or of
anything.
Um, that number then exists in athree-dimensional space floating
there.
The vector is directional, thevector is your north, your
south, your west, your east,your somewhere in the middle.
And that's going througheverything, always.

(51:40):
Vector embedding refers to howclose is that number to the
number created for the query.
So, what shoes does MatthewBertram like?
That's our query.
There are very few things thatline up directly on the zero
line for that and give you anoverlapping vector embedding.
And even then, it'll be reallyhard to get it perfect because

(52:03):
the words you use, thesyllables, uh, it's slightly
variant, but you'll end up withthe same thing.
He likes hush puppies, so thereyou go.
And everything to the side ofthat, he likes Merrills, he
likes Nike's, he likes Adidas.
All of these things get fartherapart.

(52:24):
The vector changes.
You're one degree, your twodegree, you're three degrees off
zero on that compass as thatchanges.
And the reason that changes ishush puppies are comfortable
shoes for older people.
That's Matthew's choice.
Merrills are hiking shoes foractive people, still shoes, but

(52:45):
the database knows that that's adifferent thing.
Adidas is very much into sportsand specific sports.
So now you're three or fourdegrees off alignment.
So still talking about shoeshere, and maybe Matthew's got a
pair because he can't get rid ofthem because he relives his
youth, you know, once everymonth, and that he won't give up

(53:06):
his Adidas, right?
But but that's the whole pointbehind vector embeddings.
So your your job now as an SEO,how close to the target can I
get?
How close to that line can Iget?
It's not a bullseye.
You are not attempting to hit astationary spot on the wall.
You are attempting to chase ameteor hurtling through space,
and you want to come up as closeto behind it to land on it as

(53:28):
possible as you can.
And there are points given forclose.
So can you get in line?
Can you get close?
It's a three-dimensional gamenow.
I saw somebody, um, somebodysent me a 3D, it was a gif of a
vector embedding in vectorspace, and it was amazing

(53:54):
because it was a representationof as somebody wrote the query,
every character that was addedto it was another point in this
thing.
And it was this really weirdkind of started spiraling,
flattened out, then came upanother layer and turned in on
itself, and then stopped whenthe enter button was hit because
now the whole query wasrepresented.

(54:15):
But it's not a straight line, itis a weird corkscrew in three
three-dimensional space.
That's an invector embedding ofa question.

SPEAKER_03 (54:24):
Gosh, what I was going through this.
Yeah, well, that that justreminds me of math, uh in the
Oxford program that I took, uh,where it was like entropy, and
like where's the yeah, and Ithink that that's where's the
end?
Where's the start?
Yeah, like and and and dependingon what you do, where it would

(54:44):
fall.

SPEAKER_01 (54:45):
Um here's what worries me, Matthew.
I wrote an article about themath behind all of this, and I
titled it the math behind this,and I put the image was warning
math ahead, and it was one ofthe least engaged with articles
that I wrote all of last year,which is shocking because on my

(55:08):
Substack the people there arefollowing for SEO, and I'm like,
so I tell you this is aboutmath, and you you bounce.
I'm sorry, your future is allabout math.
And if you don't like it, I Isincerely think what we are
about to see as an industry iswe are gonna see a whole lot of
the industry is about to take aleft-hand turn off the highway

(55:30):
and go in a whole new direction,and a bunch of people are just
gonna keep hurtling toward thehorizon and they don't fall off,
they don't, you know, like theend is there, it's all very
polite and calm, but it's justlike a big parking lot, and
there's nothing special aboutit.
Whereas everybody else turnedleft, and then you got to
backtrack and turn right to tryto catch up.
Like, you know, just come leftwith us and learn.

(55:52):
That's the future.

SPEAKER_03 (55:54):
So, so they're they're like they're mainly
content people, right?
Like, I think uh there's contentpeople that masquerade as SEOs,
and there's a lot more math of alot of checklist SEOs, though,
right?

SPEAKER_01 (56:07):
Like people who run around create, like they they
create checklists by aggregatingchecklists from others, and then
they sell based on a chat GPToutput that I can do this audit
and it's more in-depth thananybody else's audit, and
somebody pays them for that, andthen you know, like it's there's
a whole lot of that going onthat it's not gonna happen

(56:28):
because to your point earlier inour conversation, I can't give
you a checklist, even if I giveyou a checklist.
Would like I get in my book, Igive checklists for things like
here's a 90-day plan forlearning all this stuff, right?
Here's a way to talk to yourexecutive about this stuff.
Like, I get practical checklistsaround that, but I can't give

(56:49):
you a practical checklist on howto rank in an LLM today because
first and foremost, I don't haveAI model success rate metrics,
cement semantic density scores,zero click surface president.
I don't have an understanding ofmachine validated authority from
the machine.
These did they don't share anyof this.
I can't see inside theirdatabase to understand if I'm

(57:10):
actually getting covered.
Like, am I being crawledproperly?
Like, I can't see any of that.
A retrieval confidence score.
I I want it, I know what itmeans, but but I need their data
to help me understand.

SPEAKER_03 (57:25):
Oh my gosh.

SPEAKER_01 (57:26):
Yeah, you know, like like we're not there yet, right?
Although I outline all of that,obviously, and I give you know
understanding and and what themeanings are of these things,
and I also break them out intobuckets so that people
understand look, here's what youcan use immediately, here's what
you can go start doing today.
And a lot of this is manual now.
Do you want to know how you'reperforming and how you're you're
ranking, misusing that word, inan LLM?

(57:49):
Well, you better carve out time,you know, with some Gatorade
because you're gonna need fourhours a week to sit there and
pound queries in and capturewhat the full response looks
like and create a spreadsheetmanually so that you can track
that over time.
Pro tip do it, do not worryabout how you're gonna track
what you're seeing in it.
Track it all, and then once amonth, hand the spreadsheet to

(58:11):
ChatGPT and tell it, find thetrend lines, tell me what things
are, you know, interestinginsights from this, and then
it'll bring that back to you andgo 43% of this says this, and
18% of this, but something I seethat's growing is this, and and
it'll give you a starting point.
That's the key thing becausethey'll tell you what's not
going to give you the startingpoint.

(58:32):
Traditional SEO tools, those arenot gonna give that to you right
now.

SPEAKER_03 (58:36):
Well, I think that's a good place to end.
I would love to branch off intoa number of different topics and
conversations, and you know, youstart like, you know, so I'm I'm
working on Langchain right nowand some orchestration with
agents, and this is all new.
Like it never, you know, not adeveloper, had to learn it, had

(58:56):
to start at the beginning, hadto, you know, I didn't even
really want to talk about AIbecause how do I talk about it
unless I was an expert?
And so I had to go back andrelearn everything, rego through
everything that I knew about SEOand try to connect those dots
and build an even strongerfoundation.
And and I really think the themachine layer's done that for a

(59:18):
lot of people and helps connectthose dots a lot faster.
Um, and so I would encourageeverybody to go check it out.
Um and and I may grab a copy.
I would love to get a signedcopy by it, right?

SPEAKER_01 (59:31):
And uh we'll come to use it and we'll we'll we'll
grab copy and I'll writesomething nice in it.

SPEAKER_03 (59:36):
Fantastic.
I would love to also have youback on and talk about as we
keep progressing, like whereit's going, because the thing
that we haven't talked aboutthat that I'm seeing as the
biggest issue, and we'll kind ofmaybe leave this for the future,
is I'm sick of using all thesedifferent tools and then trying
to triangulate in my head whatthe answer is.

(59:56):
Okay.
And all of these need to usedata pipe.
And APIs to come together togive the LLMs all the data they
need to give me the best answer.
And then we talked about thetrust and the judgment of like,
is this data correct?
Uh, and and what is the risk ifit's not correct?
Like, I think that that's wherethings are going and it's moving

(01:00:17):
so fast.
And you got all these agentsthat are getting spun up and are
doing all these things andhaving agency.
And like, I mean, they're stillmaking big mistakes.
Like, there's still a lot of bigmistakes still happening.
And so, from an enterprisedeployment standpoint, it's
scary of how fast we're movingand how much new technology is
out there.

(01:00:38):
And like, okay, the developersare working on their thread, and
then you got the SEOs that areworking on their thread, and
then you got a lot of peoplejust using it but not
understanding it.
There was like a uh I was I wasat an event and there was a
threat, uh cybersecurity threatum internally at a company and
they got charged a ton of moneybecause the developers were

(01:00:58):
using it inside and didn'tnotify, and it was grabbing data
and point.
So it's just it's it it's it'sthe wild west.
And there's so many differentfacets and factors to look at
it.
And I feel behind every day.
I um I'm so totally thankfulthat I was able to have this
conversation and and share it umand and shed some light on it.

(01:01:19):
And I I really feel like now'sthe time for community for not
everybody to protect what theyknow because it's all out there
and to kind of share it and tocome up with some some new
frameworks and some newfundamentals to look at stuff
that are maybe guided by thecommunity.
Because like you were talkingabout doing all those tests, a
lot of people that I know don'thave time.

(01:01:40):
They're trying to execute for aclient.
And then there are some bigthought leaders that are
publishing huge studies, andlike then I'm like trying to
save that.
I've built a little scraper tosave all that to then look at
that information because everytime I log into LinkedIn, like
my head spins, and I'm like, I'mmissing so much information,
it's moving so fast, and then Igot to figure out okay, where's
my learning plan?

(01:02:01):
What do I need to know?
How do I add additional value toit?
And I mean, it it's it's fast,it's fast and furious right now.

SPEAKER_01 (01:02:09):
So um it is crazy.
Um I will I will give you thepro tip.
Um, if you think this is movingfast, you need to really
understand that updates thathappen at the platforms like
ChatGPT and Claude and whatnot,they produce and push out

(01:02:31):
updates more than twice as fastas what we're used to as SEOs
and Google doing updates.
So Google doing updates hascreated a cadence.
ChatGPT's cadence is more thantwice that fast.
So, first off, I think don'tworry if you can't keep up with
every single thing, becausewhat's important is that you

(01:02:53):
know how to get back into thatpipeline, get back into the
hose, and you're like, okay, I'mnot entirely lost here, right?
Like you get some of it.
Um, I think what's incrediblyimportant for people though is
just don't lose faith on it.
Like, use those platforms.
I did this last night, I satdown with ChatGPT and I said,
Hey, look, I'm sick of scrollingthrough the news because there's

(01:03:15):
always a bunch of crap in therethat I'm not interested in.
And it doesn't matter how much Itrain my news feed, it still
puts crap in I'm not interestedin.
Can you act as a news scraperfor me?
And if I ask you that I'minterested in a particular topic
from a particular angle, can yougo find that information?
Right?
So every day at whatever time,and yeah, I'm interested in

(01:03:38):
what's going on in Iran, and itwas like, oh, yeah, no problem.
And it went and got all mystuff.
And I'm like, yeah, but there'sanother angle to this that I'm
curious about that this stuffisn't talking about, which is
the military positioning andposturing around what's
happening, and you know, becauseuh I am a military buff, you
don't do things withoutpositioning assets beforehand,

(01:04:00):
and it's somewhat easy to seethose positionings happening
these days.
So I want to know about that.
But it's like I want ChatGPT togo get what I want it to go get
for me and bring it back to me.
You can use the same thing ingathering information about AI,
you can just create a promptthat you keep in Word that you
keep adding people's names tothat you want to hear from, and

(01:04:22):
you go back in and you justcopy, paste, and it's like, go
get me news from these people,has to be within the last seven
calendar days.
Today is this date.
These platforms they don't havea clock, they do not have a
clock, so they don't understandwhat day and time it is.
So if you just say, get mesomething current, good luck.

(01:04:45):
Like current is highly mobile intheir mind.
But if you tell them today isJanuary 20th, it is you know
11:22 a.m.
Pacific time, find me thingsthat are within seven days of
right now.
You've given it something it canthen take out and stamp against
everything and do a comparisonon and bring you back more
useful information.

(01:05:06):
And you can also tell it, takeevery one of those that's more
than 100 words and summarize itfor me.
So, like use the system againstitself.
That's the key here.

SPEAKER_03 (01:05:16):
I love it.
Well, Dwayne, it's been apleasure to have you on.
Thank you so much.
Is there anything that you'reworking on?
Uh, you we talked about yourSubstack, you sent it to me.
I'll put it in uh the show notesfor sure.
Uh, I'll put a link to your bookuh there.
Um, just I want to give you anopportunity to kind of share,
share any last words with theaudience.

SPEAKER_01 (01:05:34):
I well, I just first off, thank you for having me on
the show.
Everyone, thank you forlistening.
If you're still here with us,this is awesome.
Um, I do have frameworks.
So if you want to hit my websiteat Dwayneforster.com, a bunch of
the frameworks from the book arethere.
They're free for anybody whowants them.
Um, most of the frameworks arestill in the book, though.
So I will encourage you to takea look at that.
If you're interested, grab acopy.

(01:05:55):
I would deeply appreciate it.
Um, and obviously, if folks wantto get in touch, feel free to
reach out.
I'm on all the socials and I'measy to track down.

SPEAKER_03 (01:06:03):
Wonderful.
Well, everyone, thank you somuch for listening.
Please leave a review.
Uh, it helps us follow, like,share, Shaiko, uh, as we we've
talked about.
Uh, until the next time, my nameis Matt Bertram.
Bye bye for now.

All Episodes

Episode Transcript

Popular Podcasts

Betrayal Weekly

Dateline NBC

Stuff You Should Know

.css-15opob5{left:0;position:absolute;top:0.8rem;} All Episodes

.css-14f5ked{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:2;overflow:hidden;}SEO’s New Frontier With Duane Forrester

Episode Transcript

Popular Podcasts

.css-r6mb8g{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:1;overflow:hidden;}Betrayal Weekly

Dateline NBC

Stuff You Should Know

All Episodes

SEO’s New Frontier With Duane Forrester

Betrayal Weekly