In this episode
Kevin White, Head of Marketing at Scrunch — acquired by Sitecore — breaks down why “AI search” is at least three separate things happening at once: crawlers extracting content, models synthesizing answers, and agents acting on a user’s behalf. Collapsing them into one bucket sends teams after the wrong work, starting with the wrong scoreboard. Kevin argues that referral traffic from ChatGPT or Claude is the visible tip of a much larger iceberg, that consideration and citation are the two metrics to start with, that AI search is now the front door to the brand rather than the website, and that the practical work splits into two tracks: serving a specific human buyer with content no competitor has written, and delivering that content to agents in a form they can actually consume.
Key takeaways
- AI search is probabilistic by design, not broken. The same person asking the same question in the same session can get a different answer. No brand reaches 100% share of voice; the best brands Scrunch sees land in the 60–70% range.
- Referral traffic from ChatGPT or Claude is the tip of the iceberg. The bot visits the site, gathers the information, and returns it inside the AI feed — the user never arrives, so traffic systematically understates the value being created.
- The two metrics to start with are consideration and citation. Is the brand showing up for the prompts that matter, and is it being referenced as a source of truth by the bots crawling the site on behalf of the person behind the prompt.
- Handing AI search to the SEO team is the typical approach, not the best one. SEO understands crawlers, rankings, and reporting, but brand and communications own whether the brand is represented accurately, and demand gen sees AI platforms as a new growth channel — three different stakeholders, one initiative that has to come from the top.
- AI search is the front door to the brand now, not the website. People ask a bot to explain a product or a pain point instead of visiting, so the information the bot retrieves has to be accurate and on-brand before a buyer ever makes contact.
- Citations concentrate in the long tail. Wikipedia, YouTube, Reddit and niche blogs turn up constantly; identifying which sources are actually cited lets a team reverse-engineer the content and markup the models prefer.
- Agent-ready means rich HTML, minimal JavaScript, and clean structure. Identifying agent traffic at the edge and serving it an appropriate experience creates a reinforcing loop — the machine gets what it wants in the form it wants, and comes back.
- The arbitrage is the “long long tail,” not high-volume keywords. Merge the brand’s true differentiators with a sharp view of the ideal customer and write for that one specific person: less traffic, higher conversion, and the default choice because no competitor has covered it.
- Chunking, bullets and schema markup come after the customer question, not before it. There’s legitimacy to the formatting advice, but it’s what you add once the content genuinely serves the buyer — and it adds credibility for the human reader too.
Chapters
- 0:49 — What changes when machines are the first readers of your brand
- 2:43 — Kevin’s path to Scrunch and the Sitecore acquisition
- 3:52 — What Scrunch builds: visibility, prioritization, action
- 6:28 — What teams get wrong when “AI search” is one bucket
- 7:21 — Probabilistic by design, and why 60–70% is the ceiling
- 8:52 — Referral traffic is the tip of the iceberg
- 9:21 — Should the SEO team just add AI search to its scope?
- 10:48 — Why brand, comms, and demand gen are stakeholders too
- 12:05 — Earning the trust of a machine
- 13:10 — Reverse-engineering what actually gets cited
- 14:10 — Agent experience at the edge: rich HTML, less JavaScript
- 18:07 — What good measurement looks like beyond referral traffic
- 18:54 — The “long long tail” and where teams can absolutely win
- 21:20 — Do brands dissolve into a list of attributes?
- 22:09 — Where chunking, schema and FAQs actually belong
- 23:40 — One URL, two readers: humans and agents
- 26:56 — Marketing directly to agents, one year out
- 28:05 — Staying agile: half the role as hands-on IC work
Why “AI search” is at least three different problems
Crawlers extracting content, models synthesizing an answer at the moment of the prompt, and agents acting on a user’s behalf are distinct activities with distinct fixes. Kevin’s starting point is understanding what the bots are actually doing on the site and what objective sits behind them — there’s a human behind every one of those objectives. Teams that collapse all of it into a single “AI search” program end up optimizing for outcomes that don’t map to any of the three.
Referral traffic is the wrong scoreboard
The traditional loop was rank, get clicked, measure the visit. In AI search the bot goes to the site, gathers the information, and delivers it back inside the AI feed. The user stays there. Watching ChatGPT or Claude referral traffic captures only what surfaced above the waterline; the rest of the iceberg is whether the brand is being mentioned, whether it’s being cited, whether agents can reach the site at all, and whether JavaScript or robots.txt is quietly blocking them.
Why the SEO team can’t own this alone
An SEO practitioner understands crawlers, rankings and reporting, which makes them an obvious contributor to implementation and measurement. But accuracy of brand representation isn’t what that role optimizes for, and it matters more now that the bot — not the homepage — is the front door. Brand and communications have a direct stake in what the model says about the company, and growth or demand gen sees AI platforms as a new acquisition channel entirely. Leadership has to come from the top; the implementation lands across the marketing org and beyond it.
How you earn a machine’s trust
Partly by speaking a different language, but mostly by removing friction. Look at what’s being cited today — Wikipedia, YouTube, Reddit, and a long tail of niche blogs — and reverse-engineer the content types and delivery formats the models favor. Then look at where agents get stuck and strip the cruft away. Scrunch’s own differentiator is identifying agent traffic at the edge and serving it rich HTML with little JavaScript, organized the way a machine wants to consume it. Nothing is published about how the ranking works, but the pattern Kevin sees is a reinforcing loop: give the machine what it wants, in the form it wants, and it comes back.
The long long tail is the arbitrage
SEO rewards showing up for relevant keywords with volume. AI prompts don’t look like that — they’re dictated, specific, and personal. The opening Kevin points to is deliberately building content for that shape of query: take the brand’s genuine differentiators, take a real understanding of the ideal customer, and write for that one person asking that one specific thing. Traffic goes down, conversion goes up, and the brand becomes the default answer because nobody else wrote about the thing that surfaced on a customer call or in a forum.
One URL, two readers
A video or an animation may explain a concept beautifully to a person and give an agent almost nothing but metadata. That’s an argument for adding readable context, not for stripping the human experience out. Kevin’s view is that both can coexist: same URL, traffic routed to different experiences, the agent receiving a digestible version of the same intent while the human gets the designed one. And plenty of humans — as Greg notes — would rather read the bullets than sit through ten minutes of video anyway.
FAQ
Is optimizing for AI search the same as SEO? No. Doing SEO well helps, and the skills overlap on crawlers and measurement, but AI search engines are probabilistic — the same question can return different answers in the same session — and the metrics, stakeholders, and content strategy all diverge from classic organic search.
Why isn’t referral traffic from ChatGPT a good measure of AI search performance? Because the AI bot visits the site, retrieves the information, and hands the answer back inside its own interface. The user typically never clicks through, so referral traffic captures only a fraction of the value — Kevin calls it the tip of the iceberg.
What should brands measure instead? Two things to start: whether the brand is in consideration for the prompts that matter, and whether it’s being cited as a source of truth. Beyond that, whether AI agents can actually crawl the site or are being blocked by JavaScript or robots.txt.
What share of AI answers can a brand realistically win? Not 100% — the systems are probabilistic by design. The strongest brands Scrunch observes sit in roughly the 60–70% share-of-voice range.
How do you make a website consumable by AI agents? Serve agent traffic rich HTML with minimal JavaScript, structured cleanly, and make sure content locked inside video or animation has a readable equivalent. Author information, FAQs and structured markup help — but they come after the content genuinely serves a specific customer, not before.
Where should a team focus content effort for AI search? The “long long tail”: highly specific, differentiated content written for one clearly defined customer and the very particular questions they ask. Bots are good at finding that content, competitors rarely have it, and it converts better than broad, volume-driven pages.
About Kevin White
Kevin White is a B2B tech marketing leader who has helped shape go-to-market at Segment, Retool, and Common Room, and now leads marketing at Scrunch AI, an AI-first customer-journey platform. He got his start in SaaS at Gigya — where he was handed Marketo and told to ‘figure it out,’ the spark for a learn-by-doing growth mindset that merges analytical rigor (attribution, lifecycle infrastructure, reporting) with creative offer- and channel-building. That growth foundation carried him to leading whole marketing orgs, though he’s candid that the climb pulled him away from the hands-on craft he loves, and at Common Room he deliberately stepped back into a senior IC role. At Scrunch he focuses on one of the most overlooked shifts in modern marketing: websites are now consumed as much by AI bots as by humans, and brands that make their sites legible to LLMs — through structure, markdown and JSON, schema, and a parallel AI-optimized experience — win the emerging AI retrieval channel.
Resources
This episode is brought to you by Scrunch. Scrunch gets your site AI-ready so you show up in answers, get cited, and grow revenue. Learn more at Scrunch.com
Reach your customers with Reddit. Spend $500 in ad spend, get $500 back in ad credit! Learn more: https://advertalize.com/r/491818c79fb1873f
Enjoyed the show? Tell us more at and give us a rating so others can find the show at: https://aglbrnd.co/r/faaed112fc9887f3
Connect with Greg on LinkedIn: https://www.linkedin.com/in/gregkihlstrom
Don’t miss a thing: get the latest episodes, sign up for our newsletter and more: https://aglbrnd.co/r/35ded3ccfb6716ba
Check out The Agile Brand Guide website with articles, insights, and Martechipedia, the wiki for marketing technology: https://www.agilebrandguide.com
The Agile Brand is produced by Missing Link—a Latina-owned strategy-driven, creatively fueled production co-op. From ideation to creation, they craft human connections through intelligent, engaging and informative content. https://www.missinglink.company
Transcript
[00:00:49] Greg Kihlström:Hi, I’m Greg Kihlström, host of “The Agile Brand,” and here’s a question for you. What happens when the most important visitor to your website isn’t a person at all, and it decides what your brand is before a buyer ever hears from you? For years, the job was to be found. Increasingly, the job is to be quoted accurately by a system you don’t control to an audience you never see. Today, we’re getting into what actually changes when machines become the first readers of your brand, why AI search is really several different things happening at once, and why treating them as one problem sends teams after the wrong work, what a website has to become when it’s serving both a human reader and a machine that summarizes you, and how to measure and be accountable for visibilitythat’s probabilistic rather than deterministic. I’d like to welcome Kevin White, Head of Marketing at Scrunch. Kevin, welcome to the show.
[00:02:25] Kevin White: Hey, Greg. I’m so excited to be here. Thanks for having me. I’m looking forward to getting into things.
[00:02:30] Greg Kihlström: Yeah, definitely looking forward to diving in. I know this topic is top of mind for so many right now, so lot- lots of, lots of things to discuss here. Before we do that, though, why don’t you, give a little background on yourself and your role at Scrunch?
[00:02:43] Kevin White: So, I head, the marketing team at Scrunch. We were recently acquired by Sitecore, so, I’m no longer, just, like, the, the head of marketing. I’m now, part of the, the Sitecore team, and my team’s part of that team, but, exciting times and just more fuel for the fire with that acquisition. and then prior to Scrunch, and how I got in introduced to Scrunch is, leading marketing teams and, and, especially with, like, a growth, and a growth lens or focusing on, like, what’ll create demand and saw that the space was, like, a new viable channel and started seeing a lot of, what Scrunch was doing myself and getting curious there. And so that’s what brought me from a head of marketing at a previous company called Common Room into Scrunch, just kind of, scratching the, the curiosity itch
[00:03:28] Kevin White: and, seeing, like, a lot of potential in the market. And, I think a lot of that’s come to fruition in the last, year that I’ve been at Scrunch. It’s only been, it’s only been a year, but it’s, it’s been quite a journey.
[00:03:39] Greg Kihlström: Great. And so to– I know you’ve mentioned a little bit about, about what Scrunch does, but, tell us a little bit more. What, what does Scrunch do? What’s the company’s core mission? Who, who’s the primary customers you serve?
[00:03:52] Kevin White: So, we are, building for a future where there is, if you think about where the market’s headed, there’s a lot of AI agents, people prompting agents to do things on their behalf. we’re building, a product and platform for that future. that started off with AI visibility, just, “Hey, how’s my brand and my website showing up in the different LLMs?” But if you think about, that as a starting point, then there’s things like, okay, what should I do about it? Should I change content? Should I, increase citations? Should I do, different things to show up better in the search results? And then also, like, how can I act on it? Can I, ship that content? Can I provide a better experience for agents? And so, that’s the, that’s kind of the, the world we’re building for is that full, full, full loop, that full flywheel, if you wanna call it, where not only do you get that visibility as the starting point, but then also, know what to prioritize and then, be able to act on it and then, repeat the loop over and over again. That’s kind of like the key to growth and where marketing teams especially are trying to figure out right now. so, so that is, kind of like the vision and how– what we’re building the product, of Scrunch. and then, I would say, like, the, the types of companies that we work with are much more on, like, the larger enterprise side of things, both on the agency, so, like, big agencies
[00:05:22] Kevin White: working with, big, like, CPG and large brands, and, and providing services for them and then also, like, on the direct side, lots of considered purchases, like financial services, travel and accommodations, lots of software, enterprise software companies, that we work with.
[00:05:42] Greg Kihlström: Great. Great. Yeah. And, just a, a note for the listeners, we had somebody on to talk– to show us, s- the Scrunch platform a bit on our other show, One Amazing Thing, so definitely recommend you check that out as well. So, let’s, let’s dive into what’s going on right now, and as you mentioned, it’s been a busy year in, [chuckles] in this space, and there’s, the umbrella term AI search is g- become kind of a catch-all, but there’s a lot of things going on here i-in that. So can- Crawlers are pulling content, models are retrieving and synthesizing things at the moment someone asks a question, agents are acting on somebody’s behalf. When teams collapse all of that into a
[00:06:28] Greg Kihlström: one or even two buckets, what, what work do they end up doing that, doesn’t, doesn’t matter as much or doesn’t have as much of an impact?
[00:06:36] Kevin White: I would say, the first place to, start to navigate to this new world is just, like, understanding what the bots are doing and, what the bots are doing on your, your website, what are their objectives. There’s a human behind those objectives, so a lot of that stuff just starts with, for these prompts that I care about or for these, these things that people are asking that I really care about, is my brand showing up or not? And then also another corollary to that is, for the, for the set of, is, is my website in consideration or my brand in consideration for this thing that a user’s asking, but then also, is it being, cited? Is it being referenced as a source of truth for those AI bots that are coming and crawling your site and trying to do
[00:07:21] Kevin White: work on behalf of the person behind the prompt? Those are the two kind of things to start with measuring of, like, to tell you what’s going on and maybe how to change your strategy or how to adapt your strategies. It’s definitely not just the SEO world where, yes, if you’re doing SEO well, if you’re doing search well, you’re probably doing AI search well, but, there’s a lot of nuance and a lot of difference to that, one, one being that, by a feature, not a desi- an, and not a, not a bug, is that the AI search engines are probabilistic. So the same person asking the same question the same time, even in the same session, will get potentially, like, a different answer. And so you need to, like, think about how your brand is, is showing up and,
[00:08:07] Kevin White: like, you’re never gonna hit, like, that 100%, 100%, like, sh- share of voice rate. the, the best brands we see are in, like, the 60 to 70% range there. And if you think about, like, what that changes or, like, what to maybe cut or, adapt to, I think the f- one of the first things is just, like, the metrics that you’re tracking because traditionally we’re looking at, okay, are we showing up in search, and then are people visiting and, and going to our site? Are we getting referral traffic? And if you think about how that l- that looks on the AI search side of things, people are… The, the AI bots are going to your site and, and visiting, and get- gathering information, and pulling it back to the user. so the user’s all in the, the AI feed. and so, if you’re just looking at
[00:08:52] Kevin White: referral traffic from ChatGPT or Claude, you’re really just looking at the tip of the iceberg. And so exposing that whole rest of the, the iceberg, or what’s be- beneath the surface of is my brand being mentioned, is my brand being cited, are, are AI agents visiting my site, and, like, are they getting blocked by… Do we, do we need to fix something about our JavaScript, our robots.txt? Like, those are the things that are… I wouldn’t say, like, you need to neglect the other things, but those are new things in this world where you need to be paying attention to.
[00:09:21] Greg Kihlström: Yeah, yeah. Well, and, and is it a false assumption that SEO and GEO or AEO, what- what- whatever the, the org is referring to, are, are more similar than they actually are? it’s, in, in other words, like, what I’ve seen is the first step to doing anything with AI search from a optimization standpoint is, “Hey, we have this team that’s already doing SEO, why don’t they just add to their capabilities?” Or, like, is that, is that the best approach? Is that the… what, what, what should companies be thinking about there?
[00:09:56] Kevin White: I would say, I would say that is the, the typical approach. I don’t know if I would say that that is the, the best approach because the, the, the SEO person, does understand how crawlers work, does understand, like, how they’re, they’re ranking in, like, the different organic search results and, reporting on metrics and stuff like that. And so there’s a lot of corollary to, from that role into, like, who would own some of the implementation and metrics and stuff like that around the AI search world. There’s also, a bit of a difference to the different teams that are stakeholders and that care about this stuff. and one really crisp example of that would be more of, like, the brand or, brand or communications team. And so not only… Because, because AI search is kinda like the front door to your website, um,
[00:10:48] Kevin White: or, or your brand. And nowadays when people are asking AI bots versus, like… Or asking AI to tell me about this product, tell me about, like, this pain point that I have, they’re not going to your website anymore. And so the information that the AI bots need to go and retrieve needs to be accurate and needs to represent your brand in the right way, and that’s not always something that an SEO person is looking for. And so, I would say, yes, that person is definitely one of the contributors and stakeholders, but there’s a lot more to it, too, especially on the accuracy and brand representation on the PR and comms and brand side of things. And then there’s also, like, a whole nother vector of growth, like a growth demand gen, type of person who might see, AI platforms as a new channel for
[00:11:33] Kevin White: growth, and that’s a little bit different than what the SEO person, would see as well. It’s more similar than the brand one, but still a different, different, pillar within the org. So-
[00:11:43] Greg Kihlström: Yeah
[00:11:43] Kevin White: … we’re seeing, like, a lot more diversity and I think, that leadership needs to come from the top and the initiative needs to come from the top, but then the teams that implement it all are all impacted by this, completely new way of users finding things, discovering things, evaluating things. It’s like, it’s, it applies to many, many different parts of the marketing org and even beyond the marketing org, I would say.
[00:12:05] Greg Kihlström: Yeah. Yeah. Well, and I think, some of, some of this also goes down to, a lot of the, a lot of the ways that we’ve traditionally defined a brand or unique selling proposition, all those kinds of things- a, a machine is persuaded or, or trusts, if I can even use th-those terms, in very different ways than a, than a human. even, even thinking through what, what does luxury mean to a machine, right? you have to quantify those things i-in perhaps a different way than you might for, for a human. So how do we, again, to use the terms, like how do we earn the trust of, of these machines so that they can be cited and as a good source? I mean, again, there’s, there’s a…
[00:12:50] Greg Kihlström: There’s precedent here with SEO, but, but to your other point, it’s, it’s, it’s different, right, when we’re, when we’re talking about AI citations and, and things. So, pricing specs, policy information, all, all of those kinds of things, like how does, how does a, how does a brand think about it from, from that, machine trust perspective?
[00:13:10] Kevin White: It is very interesting to thinking about like, um… It’s an interesting perspective of like how do I, how do I earn the trust of the, a machine? Um-
[00:13:18] Greg Kihlström: [laughs]
[00:13:18] Kevin White: I do think, [laughs] I do think, like do you have to, like, speak a-
[00:13:21] Greg Kihlström: Just anthropomorphizing it a little bit, right? But yeah.
[00:13:25] Kevin White: It’s like do you have to speak a different language to the machine? And like in some regards you do, but I, I do think it is, over time especially, it’s like very early stages right now, but over time especially, it’s gonna go back to like, the mas- the machine is ultimately working on behalf of the human, and so like thinking about it from the empathetic view of what the human is trying to get, will lead you to the best outcome. but I do think, especially right now, catering to the machines, is more about, enabling that machine to consume the content in the way that it wants to consume it. and there’s lots of like signals and signs that you can see for that. One, one is just, measuring or looking at like what, what people are searching for and what’s being cited, and then you can kinda like reverse engineer,
[00:14:10] Kevin White: like oh, this is what the machines like. whether it’s a Wikipedia article or YouTube or Reddit or some niche blog. We actually find like lots of the niche citations are, the ones that have like a, the long pr- long tail of citations is where a lot of that stuff is coming from. and so if you can identify what are the sources of truth of citations, you can kinda see like the types of content and the types of, like code that’s being delivered to those machines. and then, the other things, catering to machines that you’ll, you’ll see are, are they able to, like y- Looking at like what, where do they get stuck and, what, what’s not necessary and removing a lot of the, the cruft away from, from those, machines. And so right now, it’s, it’s kind of like a, a, a differentiator, that we have with Scrunch and, and, and part of the product that we’re building is for the agent experience. So identifying that tr- agent traffic at the edge and then delivering a different experience to those machines that’s,
[00:15:12] Kevin White: rich HTML, like not a lot of JavaScript, organized in the right kind of fashion. All those things are helpful to give the machine what it wants, and then it’s kind of like a reinforcing loop where the machine goes, like searches your site, it gets what it wants, it gets it in the way it wants, and then it comes back and it, and then it will go back and back again because it now trusts your brand or your site is, kinda how we see things going, and s- we, we’ve definitely seen some evidence of that. Of course, all this stuff is not like public of how it works, or the, the machines don’t tell you they’re like, “You are the preferred source.” but, but they do, they do it in, in s- in, in some ways, and so we can kinda see like evidence of that. and I think, we’re still in the very early stages of like how all this stuff works, and over time there might be… I like to say like there’s no mat cuts of AI search right now like there was, with, with, at Google kind of like leading and telling people like what works and what doesn’t. But over time, the market is, gonna mature and we’re gonna find out a lot more.
[00:18:07] Greg Kihlström: I, I know you touched a little bit on the, the measurement component as well, but I do… I’d, I’d love to hear your perspective and, and Scrunch’s perspective on, how should… Because y- you’ve mentioned it’s probabilistic. It’s not gonna, what is it, stochasticity or whatever of LLMs. Like, it’s not… It, it’s gonna do different things different, different times. but When, when you’re a trusted source, it will be more consistent than others and, and things. So, how, how should brands be thinking about measuring beyond that referral traffic that, that you mentioned can’t- isn’t the only, it’s kinda the, the tip of the iceberg. So, like, what does good measurement look like know- knowing also that, we’re, we’re evolving as well?
[00:18:54] Kevin White: I think, where a lot of people, teams get, get kind of, hung up, and this is different than, than, the SEO space, is that, you want to show up for the most, like, any relevant keyword with a lot of volume in the typical, like, SEO world. but in, in AI search, people are searching, like, very different long tail type of things. Like, if I think about the way that I search for things, using an AI bot, like, it’s usually very specific, and oftentimes I’m, like, using a tool to dictate into the chat stream itself and so my– The way that I’m searching for things is very long tail, niche-specific, to, to myself. And, one benefit, and maybe, w- arbitrage for, for teams that are, like, looking to, to grow fast in AI search is that, you can almost build or create that long tail content. I like to call it, like, the long long tail. and if you build it, the AI bots are very good at going and finding that specific type of content. and so the way that I, u- usually recommend, brands, like, measure and start to, like, how they would diverge in their strategy is, to identify, like, the things that are, like, true differentiators for, for your brand, and then also have a really good understanding of your ideal
[00:20:17] Kevin White: customer persona. And, like, I mean, this is, sounds like very basic marketing stuff, but, like, merging those two things together and creating content specifically for that one person or, that one person who’s using AI to, like, find certain things that are, like, very specific, like product specs, that actually tends to work really well, is to focus on where you can absolutely win, versus, like, a more broader, like, general SEO type of, of strategy, which is more of, like, partially a volume game, but also, like… It’s like a, a, a game of volume plus intent. I would say, like, the AI side of things is very much more, like, long tail intent. And yeah, you might get less traffic, but you’ll get a lot higher conversions. Um-
[00:21:00] Greg Kihlström: Mm
[00:21:00] Kevin White: … and you’ll stand out from competitors ’cause you’ll be, like, the default choice because, like, no one else does this one thing that you wrote about that you found out from a customer call or someone, like, writing something in a forum. It’s like you can, you can speak now to that specific user. You can create content for that specific user, and, like, they will find you if they ask the bots for that one specific thing.
[00:21:20] Greg Kihlström: Yeah, yeah. Well, I, ’cause I, I think there is a… Fear may be a strong word, but I, I think there is a fear where, if we just start speaking the way that you would assume a machine would read, which is let’s just have the attributes out there and say, it comes in black and gray, and it’s small, medilarge, and, then all of a sudden brands start all being just a bunch of attributes, right? So I think, I think what you’re saying is that there is, there’s something beyond that that can be very brand specific and very, and very differentiated. But is it a, is it a mix of both? Like, do we, do we also need to think in sort of the quantifying the, the attributes, but also the
[00:22:05] Greg Kihlström: more, the more long tail thi- Like, how does, how does a brand think through that stuff?
[00:22:09] Kevin White: I mean, I do think there are lots of, they’re not necessarily myths, but there’s lots of things out there that you’ll, you’ll hear like, “Oh, we need to chunk our content. We need to write in bullets. We need to, like, add schema market- markup to all this stuff.” There is some, legitimacy to that, but I do think that that comes after the fact of, are we servicing the right user, with specificity that will, like, really speak to them? And then after that, it’s like, okay, if we’re gonna write something about this or create content around this, or we have existing content that we need to optimize around this, can we make it in a digestible format for, for the AI bots that are going to visit it so they can find it more easily? should we add an author or FAQ and all this kind of stuff? And then I would argue that, like, that type of stuff adds to credibility both for the human visitor and the AI bot visiting too. So, like, you really… I mean, I think the north star should be, who’s the customer, who’s the, the person who’s actually gonna be doing the buying there, and then, adding on the
[00:23:07] Kevin White: things that’ll, like, help the AI agents consume that content after the fact. It is something that is a little bit like, if you think about what we, what we can do at Scrunch, is that you can have the best of both worlds, where, like, you have that human experience and then you also, like, deliver content in the way that the, that the AI agents wanna consume it, at the edge. but I still think that, like, the both things can kinda, like, live together, and that, like, it doesn’t necessarily mean, like, that the human, content or experience needs to be diminished because, like, the AI, like, in service of the AI, bot visit.
[00:23:40] Greg Kihlström: Yeah, yeah. Well, yeah, ’cause I mean, it’s– I mean, I just, I try to think of, of analogies. It’s like, in, in-store retail is not going away just because e-commerce is growing in popularity year over year. So it’s like the, the human website. Like, pe- humans are still gonna go to websites. I think it’s already been shown that they’re showing up at websites later in the funnel, than-
[00:24:03] Kevin White: Yes
[00:24:03] Greg Kihlström: … than they used to on, in a, in an increased way. But to what you’re saying, there’s, there’s gonna be life for, for both options. So, brands, should brands be thinking about basically one website that serves both, but just in, in technical ways is different? Or, like, what, w- I guess, how, how is that solved best?
[00:24:28] Kevin White: I would say, you definitely need to, like… I feel like I’m gonna, like, reverse on what I just said a little bit. But, you definitely need to, like, create, make sure that, like, when the AI agents are visiting your site, that they can consume content. Like, one example of this could be like, okay, maybe a video, would… A, a video or, like, some animation on my page would, like, really explain this concept super well to a human. but if you put that up on your page to an AI agent, it’s gonna, like, struggle a lot because it’s gonna be a lot of JavaScript. It’s gonna be an- not content in a structured way that it wants to consume. So-
[00:25:06] Greg Kihlström: Right
[00:25:06] Kevin White: … I do think that, like, y- you need to think through that bit of it as well, where it’s like, okay, we can’t just, like, throw a video up and, and then, like, have the, if the AG- AI agent visits the site, then they’re not gonna, like, consume any information other than the metadata that’s on the video itself. It’s gonna be, like not enough context. And so, adding that additional bit of context to, for the AI agent to c- to crawl, and consume is, like, probably not a bad idea. I would also say that it’s probably not a bad idea for the human visitor to see that stuff, too. Like, maybe they don’t wanna see the video and they wanna crawl and then, like, scroll instead and read. So, I think, I think that you can have the best of both worlds. And then again, like, the, there, there could be a world, and we think there is gonna be a world where, you can, you can create two different experiences without having it to be a super complex overhead type of thing, where, you just take the, the intent of the human written site and then, like, turn it into, a digestible, piece of content for the AI agent. The URL stays the same, you just direct traffic into, like, different, different, experiences and, like, the AI gets what it wants and the human gets what they want, and every- the world is a happy, magical place. [laughs]
[00:26:19] Greg Kihlström: Yeah. Yeah. Well, I, I gotta say, I am one of those people that would much rather read bullet points than watch a video, that takes 10 minutes to explain those 10 bullet points. So maybe me and the machines have something [laughs] in common, so.
[00:26:32] Kevin White: I think, I think there’s, there’s more that humans have in common with machines than what maybe we’re thinking on, on that level, and, like, the marketer maybe gets in the way where like, “Oh, I wanna create this really cool experience,” and, like, it’s not actually how people wanna consume things there probably is something to that where, humans did design AI, so maybe the… or engineers designed it, so maybe the engineers are, the, like, it’s, it’s, it’s the preferred method for, for engineering type of mindset, I don’t know.
[00:26:56] Greg Kihlström: Yeah. Yeah. Totally. Yeah. Definitely. Well, Kevin, thanks so much for joining today. Definitely, fast moving space, so lots of, lots of, lots of things going on, so alw- always exciting to talk through this stuff. I got two last questions for you as we wrap up here.
[00:27:10] Greg Kihlström: First one, if we were having this interview one year from today, what is one thing that we would definitely be talking about?
[00:27:17] Kevin White: Oh, man. it’s so hard to say, with, with, c- ’cause things are moving so fast. I would say this is where I think the market is going, and it’s, like, gonna be more around marketing directly to agents and, that’s not just on your website, but like, things like your, your API, if you’re open sourcing your products, like, all that type of stuff. I think that is where a lot of the market is going and, how do you become, like, the default choice for agents that are not only visiting your site, but that are, doing all the work, like, within, within a harness, like within CloudCode or within, Codex or whatever, harness is popular at the time.
[00:27:57] Greg Kihlström: Yeah. Love it. And last question for you, what do you do to stay agile in your role, and how do you find a way to do it consistently?
[00:28:05] Kevin White: That’s an easy one for me. I just do the, um… I, I, I love my team. I have a team, but, I try to do, about half of my role as, like, hands-on IC work. and so that keeps me, keeps me grounded, I feel like, where I’m not talking about all this stuff, but I’m, like, actually doing the work, in there. writing emails, creating things, using, using, like, the, the, the AI tools. and my team might not like that because I come in and, like, blow things up or, have some crazy hair-brained idea and, like, hand it over to them and be like, “Hey, can you take this across the line for me?” But, I do like working, working in that way, where I feel like I’m grounded and, like, still doing the work and it’s keeping me sharp.
