How HubSpot literally buried traffic | Jennifer Lapp, Head of Growth Marketing LATAM & DACH @ HubSpot

HubSpot buried traffic as a metric last year. Not because they stopped caring about reach, but because in a world where AI systems deliver answers directly, the old playbook stopped working. My guest is Jennifer Lapp, Head of Growth Marketing for LATAM and DACH at HubSpot, where she's been rebuilding the growth engine from the ground up. We dig into the real experiments she's been running, what's actually working and what isn't, and how she's steering her team through one of the biggest shifts in the marketing space.

Check out the full episode

Check out the episode on YouTube, Spotify, or Apple Podcasts.

Key learnings from this episode

The funeral for traffic

• HubSpot held a literal ceremony — the team wrote down what traffic made them feel, put the cards on a board, and gave it a funeral speech
• At its peak the German blog hit 700,000+ page views per month; letting go of that number was emotionally hard for an SEO-trained team
• New North Star: brand presence, AI mentions, citations, and behavioral influence — not sessions or rankings

Operation Everest

• HubSpot killed 50% of its blog content using a Marie Kondo-style scorecard: does this serve our ICP? Does it bring in the right customers, not just MQL volume?
• Scoring was tied to revenue quality — leads from a given article were traced to actual customer outcomes, not just signup counts
• The cleanup happened quietly, well before the LinkedIn discourse about HubSpot's traffic drop

The pricing accuracy experiment

• HubSpot's pricing pages were built in JavaScript, making them hard for LLM crawlers to parse — AI answers were below 20% accurate on pricing
• Fix: plain-text pricing articles per region and language, with structured chunks and schema markup walking the LLM through every tier and package
• Result: pricing accuracy across all regions rose to approximately 80%; the same structural approach was applied to persona-based product pages

AEO is not SEO in new packaging

• Structural similarities exist, but AEO requires leadership-level buy-in and cross-functional collaboration that SEO never did
• The roles that are merging most visibly: digital PR, classical PR, and SEO — link building and earned media are no longer separate disciplines
• "Influencing training data" is the 2026 equivalent of black-hat SEO from 2008 — spend the budget elsewhere

From channel ownership to product ownership

• HubSpot moved its international growth team from owning channels (blog, YouTube, website, micro apps) to owning individual products: Marketing Hub, Sales Hub, Service Hub, Content Hub, AO product
• Channel ownership scattered effort and made it impossible to steer narrative or AI visibility cohesively
• Each person now owns the full go-to-market narrative for their product, not a slice of a channel

Prompt tracking at scale

• 25 well-chosen prompts beat 100 generic ones — quality of prompt construction matters more than volume
• HubSpot tracks roughly 1,000 prompts per language across five languages, built around product suite, personas, company sizes, and industries
• Best practice: start from internal data (sales conversations, voice of customer), layer keyword volume on top — don't trust tool-suggested default prompts

The LinkedIn Pulse citation experiment

• HubSpot noticed many AI citations in CRM and marketing topics came from old, well-structured LinkedIn Pulse articles (not their own)
• Test: created Claude-drafted, editor-reviewed Pulse articles in three markets (LATAM, France, DACH)
• Result: worked in LATAM, inconclusive in DACH, did not perform in France — confirms market-by-market testing is essential

AI-assisted vs. AI-generated content

• AI-generated content that serves humans is problematic — it is not fun to read and the quality bar is hard to maintain
• AI-assisted is fine: research, brief, QA — but a human must read, edit, and verify for factual accuracy before publishing
• The copyright question most marketers are not asking: if Claude wrote it, you likely do not hold copyright — the models are trained on others' work, and the output may not be legally yours

The revival of personas

• As AI search personalizes responses based on memory and user context, generic content performs worse against persona-specific content
• The same information needs different packaging for a CTO (infrastructure, security, data usage) versus a head of growth (experiment outcomes, attribution)
• Search volume was never the truth — behind every query is a person with a role, a budget, and a buying stage

Auto-generated transcript

Niklas Buschner (00:02.008) HubSpot literally buried traffic as a metric last year. Not because they stopped caring about reach, but because in a world where AI systems deliver answers directly, the old playbook stopped working. My guest today is Jennifer Lapp, head of growth marketing for LatAm and DAH at HubSpot, where she's been rebuilding the growth engine from the ground up. We'll dig into the real experiment she's been running, what's actually working and what isn't, and how she's steering her team

through one of the biggest shifts in the marketing space ever. So, welcome to the podcast, Jenny.

Jenny (00:37.13) Hi, thank you so much for having me, Niklas.

Niklas Buschner (00:39.64) Thanks so much for taking the time. hope with the introduction I didn't set the bar too high, but I think everybody that has tuned in, this will be a super insightful episode already from our pre-talk. know this. So I think this will be a very well spent hour for everybody listening. So I am 100 % sure. So can you take us with you?

Jenny (01:00.46) I hope so, I hope so.

Niklas Buschner (01:07.102) back to that moment that I just described when HubSpot buried traffic.

Jenny (01:13.58) So we realized that traffic wasn't the best number or the best KPI to look at a while ago, actually. I think you probably saw all of the LinkedIn posts a while ago, like maybe two, three years ago, where those big core algorithm updates hit, especially HubSpot and all blogs that were focusing on educational content. And we were...

Maybe you've realized as well in the audience as well, we did not really engage in that conversation on LinkedIn that much. were, for one thing, we didn't want to personally to engage in those conversations, but for another, we had a already before that shift happened, we already made the very conscious decision to not focus on that KPI.

that much because all of the shift that was happening back then was already going into Google answering questions already on their platform. Us not wanting to focus on content that was very high level. was the big part of our content strategy, right? Like everyone who knows HubSpot, the inbound marketing machine. Our content was...

very tofu, was trying to be the one go-to source that all marketers, sales reps and service people, whenever they have something that is remotely related to what our products can offer, we were the one go-to traffic source. But the shift that was happening at that time was that those queries were, Google wanted to...

have people stay on their side and not go into the blog articles. So what we did already before all of this conversation happened on LinkedIn, we killed 50 % of the content on the blogs. So we went through it and we did Marie Kondo on all of the international and English blogs and said, if it doesn't bring us joy, we will kill it. And we defined this

Jenny (03:39.71) What brings us joy very clearly, we said, okay, this is this serving our ICP, is this serving our audience? And more importantly for us from a business perspective, is this bringing in not just MRR leads or signups, but is this bringing in the right people? So we created scorecards that, or scoring system that said, okay, from this,

this lead that was generated through this blog article, they fall into a very good or not so good customer in the end from a company perspective. So that was the one side of what was happening in the background that no one really knew outside on LinkedIn. The other thing that was happening and that was more the thing that you were going for with burying traffic.

sounds really metaphorical, but that is actually literally what we did. We had a big offsite for all international blog teams and also on the English side. We had a ceremony with a big letting go of traffic where we had a minute of silence to commemorate. So basically everyone wrote down

how traffic made us feel. We were all hands-on SEO specialists. So a lot of our identity as marketers was circling around rankings, CTR traffic, trying to let go of a number that defined us for so many years, especially thinking about the Spanish blog at some point driving more than 2 million visits monthly.

Or the German blog when I joined HubSpot, I think when I joined HubSpot, was at like 50,000 page views per month. And then at the peak time, it was more than 700,000. For a business blog, we were so proud of that. And it was so, so hard to, to let go. So we really had everyone write down on a card. Traffic, you gave us the feeling of

Jenny (06:01.258) accomplishment. If we rank, we will grow. Our sessions were the impact. SEO is predictable. Those were the things that we wrote on those cards. We put them all on a board. We sat there for a moment and then we said, we had a real speech, Justin on our team back then did this, the funeral for traffic. That's what it was called.

Um, where we really went down to reflect what was traffic actually measuring for us? What impact did it reflect and where did it lie to us in a way? So that was the, the main things that we wanted to touch on with the whole team. And, um, where have we seen that other channels?

right now, newsletters, YouTube, other channels that we build up within that AI driven visibility.

may might have not resulted in a click, but clearly drove influence that we had impact on. And from that, we redesigned that whiteboard that everyone was putting their letting go messages on into the clear signals of what we want our team impact to be defined by brand present visibility.

AI mentions, citations, and this whole behavioral change that we had through the time. That was, I think, one of the hardest moments, but it was also very necessary for the team and also for me as a manager to understand and lead the way into letting go of this big North Star that we were circling our strategies around.

Niklas Buschner (08:05.591) I hope the Pope doesn't listen to this podcast because I could see him saying that the ceremony for traffic is little blasphemous, but I totally get it. So no judgment from my side. Let's quickly tap into the whole, like LinkedIn hype, et cetera, before we go into the traffic gone and then what comes next discussion, because how easy was it for you to not engage into the conversation?

Jenny (08:08.002) Hahaha

Jenny (08:14.348) You

Niklas Buschner (08:33.699) because I also remember it very vividly that were a lot of posts about this Shrek emoji blog posts and all the kind of stuff. So you're now obviously saying it from a moment in time where this is like, you remember back then, but I could imagine like in your fingertips, it itches that you want to say people what's actually happening, but obviously you can't.

Jenny (08:42.231) Ha ha.

Jenny (09:01.762) So in a way we did reply to that. We had several master classes, webinars about our, back then what we called Operation Everest, where we basically killed all of the content on our blogs. So that was, it was more collecting our thoughts in more thought leadership driven content than just engaging in the conversations.

What helped a lot back then, and I'm super grateful for that, was that we had a lot of content strategists that were either HubSpot users or HubSpot fans, or even ex HubSpotters that were taking on the conversation for us. And they were like, you know, I have worked with HubSpot for such a long time. I am sure they're, if they're like not worried about it, that is for good reason. And it was for good reason because like way before...

this traffic drop, already moved the needle around how we're generating demand. We created newsletters like the Hustle, we built up the YouTube engine, we built a media network to make sure that we just don't depend on only one source of income, if you will, like one source of traffic. And that was really nice to see.

Niklas Buschner (10:26.157) I think it's always good to have built up a community of people that are in a way supportive of a product or supportive of a company, especially during these times when you might not be able to share everything that is going on behind the scenes. Although, I mean, the, the, the tendency is more and more to build in public, et cetera. And you are obviously also sharing love stuff transparently, but sometimes you just cannot share everything. Right. But people always want to have like, I mean, everybody's is looking for reach and like.

They're seeing the story and they want to somehow engage and like if they, if you, if you hit HubSpot or if you, if you name HubSpot, then probably a certain level of reach is guaranteed. So probably we cannot blame people. Yeah. Cool. Around the ceremony and the whole shift that was going on with traffic, what would you say? How difficult was it to steer your team through that? Because I mean, now it's obviously gone and you reflect on it.

Jenny (11:07.212) That is true, Yeah, yeah, that's very true.

Niklas Buschner (11:24.675) And it made a lot of sense, but I could imagine that if you tell the team, hey, we will do a ceremony where we basically do a funeral for traffic, that not everybody is like, hey, Jenny, this sounds like a great idea. Let's do it.

Jenny (11:38.538) I think that might surprise you, but everyone was really engaged in that and was on board with it because they have obviously been thinking about that shift for a long time as well. I have the honor of working with very talented marketers every single day. I'm always amazed at how smart everyone is at Houseparties. It's insane.

Niklas Buschner (11:43.587) Mm-hmm.

Jenny (12:03.554) So they have been trying to move away from traffic on their own already. So for us, was more of a, now it's official. Now it's official. That's really not focused on it as much. We still monitor it to be completely transparent. Obviously we still have a look at rankings. We still have a look at traffic, but we do not use that as our, as in any way as a main KPI.

Niklas Buschner (12:33.103) Even if the team is super smart, which I 100 % believe you as like a person in the leadership capability, you obviously have a certain responsibility to also give people direction and to also give a certain level of clarity. So even if the conversations were already going on, maybe like during a coffee chat, I don't feel I don't feel so well about traffic anymore, et cetera. How did you approach it? Like when was the moment in time when you said, okay, we have to

Like we have to be official and we have to be honest about that there has to be a change, like an official change.

Jenny (13:10.306) I think that moment happened when we... So to guide you back historically, we had the whole traffic dip, I think it was in 20, starting after COVID. I think that was the, when all of the blogs on a lot of SaaS companies in general, or other companies as well, were losing organic traffic from Google.

we sat down in the winter of that year to define a strategy, not necessarily for recovery, but for cleanup. So we said, we are like, we are not 20, 30 people in the team. have to make sure that now that we're not seeing as much outcome from organic traffic, the effort that we

put into it is really well invested. So that's when we made the decision to do the Mary Condor style block cleaning. And after that, because we, within that cleanup, we already defined the new KPIs. said, okay, MRR coming from the blogs or from other properties, leads and signups for a certain bucket of company type that we're targeting.

Those were already defined in that period. So then setting up the strategy for long-term growth in those areas was fairly easy in a way. there wasn't, we didn't have to do too much enablement for the team because they have been mapping out that strategy before on their own, like with our help, obviously. And then we had a

a week of team offsite, on site basically in Berlin, where everyone was there. had the most amount of workshops in our history of the team, think. Spent like two, three hours in the peer groups, talking through different scenarios and also building each other up with that. And then also making clear that we're not just focusing on this one channel.

Jenny (15:36.258) for the future. our team was built up as a clear SEO team for the international blogs. And that was seven years ago. Now we've moved on for the past four or five years to owning the YouTube channels, to owning micro apps, like make my persona generators or the individual product pages, anything that is happening on the website.

So just having this ownership of just one channel kind of limited everyone in making sure that the other channels grew. So that was one of the changes that we did. And now just recently we had another change on our teams where we took that even one step further and we divided into the individual hubs that we serve. So right now every person on the team owns marketing hubs.

Content Hub, the new AO product, Sales Hub, Service Hub, all of the individual products, but owns all of the narrative and all of the go-to-market around it. So it's not just before that we were focused on, okay, you own the blog, you own the YouTube channel, you own the product pages, you own the website and you own micro apps, which kind of scattered all of our efforts a lot.

made it impossible to really have an impact on visibility, to have an impact on what is coming from AI or organic. So now having this clear focus on each of the individual products makes sure that the team understands and steers the narrative much better.

Niklas Buschner (17:25.902) And would you agree that there are still a lot of teams out there that struggle with this whole transition that is going on? And if you would agree, why do you think that is?

Jenny (17:37.794) I think for sure. Also, HAPSWAT is not... I'm pretty sure that we struggle with those things as well. I wouldn't say we have the solution for everything. Quite the contrary, I guess. But we have the opportunity or the better enablement to really go for testing. I think that is one of the things that we need to focus on, especially in AI search at the moment, because no one has the one right answer.

No one, like even though there are a lot of experts out there that claim that this is how you should do AI search, answer engine optimization or generative engine optimization. It's very early stages and the algorithms, the different engines, they change almost every day. And we see that in citation rates or in the different mention rates. We see.

what impact it has that now ads are enabled in JTBT. So making sure that the teams are equipped to boldly go where no one has been before. I think that is one of the things that is the biggest challenge that teams have at the moment.

Niklas Buschner (18:54.532) Hmm. But from what I heard, there are a lot of things that work pretty well in how you handle it. And what made me think is especially that you said, the team basically comes from an SEO, from a content background. And now you're talking about testing, about experimentation. You have, named a lot of other properties that were not considered part of like the classic content and SEO work maybe before that are now somehow in the responsibility. And I.

I see the talks about the ever-changing job market and certain roles that might become redundant or AI basically killed this role, now they killed this role, etc. So what do you think can other teams learn from how you handle it if they also have people in their team that are highly skilled but that have a background that is

not necessarily or a skill set that is not necessarily what you need maybe for the next 12 to 24 months. So how can they also make their teams future-proof?

Jenny (20:03.242) I think one thing that every manager or within leadership position should focus on is, or maybe two things. One thing is leading by example, which is extremely tough to be fair. Right now, it's so hard to keep up with all of the changes. So deliberately making time in your calendar as a manager.

to grow yourself, to make sure that you read up on the latest updates, that you are in communities, that you exchange ideas, that you do what we are doing right now, exchanging ideas and strategies and things what worked, what didn't. Because that is what you demand of your team, right? You want them to skill up, but how could you...

live or like ask your team to do that if you don't live it yourself. So that is one of the main, main things I know. And the second thing that goes with that is what managers and leadership often forgets in various setups is that with the higher responsibility of upskilling, there is still the normal amount of work that you have to do executive wise.

the operations are still ongoing, but at the same time, you have to scale up and you just have 24 hours a day. You ideally want to spend eight of them at work and not 20. So making sure that the team has the space to sit down with you, deprioritize things that are not the, that don't have the highest impact at the moment to make sure that you as a,

as a team member can focus on, focus your efforts on upscaling in a certain area. I think that is the main challenge because as you said, the skills that you had for SEO and content are different to what you now need in area where everything is undefined. That means

Jenny (22:18.994) You have to create the playbooks within your company, even though there are obviously playbooks from companies out there already. HubSpot is sharing, SEMrush is sharing, PKI, everyone is sharing what they're working on, what worked for them or didn't. But that doesn't mean that for one thing, those are future proof strategies because as we know, everything is changing all the time. What worked yesterday might not work tomorrow. What works for German market might not work for a Spanish one.

One of the things that I learned in the, in the past, all of the different markets are different stages. So making sure that your team has for one thing, the time to skill up in strategies and playbooks for other companies that worked, but also making sure that they have time to skill up on experimentation for the things that they own and then really live by this own and optimize credo. So making sure that you.

can run experiments by yourself. For me, what I have in my team is like in a good experiment that failed is worth way more than a strategy that you mapped out that you've seen somewhere, but you did not execute on. And that is something, it sounds very obvious right now, but I see that happening in a lot of companies still that people,

are afraid of taking the risk to have an experiment fail.

Niklas Buschner (23:50.886) Before we go into your actual AI search experiments, because there's a lot to uncover there. I would like to get your thoughts on one more thing, because I just listened to a podcast with Kat Wu, who is head of product for Claude Coat and cowork at Anthropic. And she talked about the merging roles between product managers, designers, and engineers. That you basically have very blurred lines already.

Jenny (24:20.332) Mm-hmm.

Niklas Buschner (24:20.785) between the different responsibilities. So that basically an engineer is more and more tapping into the product manager role, that a product manager is tapping more and more to the engineer role. Do you see the same happening for marketing roles right now?

Jenny (24:34.924) I think the best example there is the blurred lines between SEO, AO and PR. I think that is what we're seeing right now. One of the most apparent ones. Then you obviously have the different fields of SEO. I think one of the risks that I see happening in the industry right now is that everyone says, AO or G, GEO is just SEO in a nice packaging.

Which is not true, I think, from my perspective. I think there are a lot of similarities. There is a lot of things if you do SEO well, you will also rank in AI research. like, I think a lot of the things are similar. The approach is similar, but what changes with AEO is that for me, it's not a content or SEO task or priority. It's something that we have to build up from

leadership on, and you have this blending between digital PR and even classical PR that have to work much closer together than what SEO had to do before. And then maybe having this other shift between affiliate link building and PR. think that is one of the...

That's probably the trajectory where the roles merge the most at the moment, from my perspective.

Niklas Buschner (26:07.825) Do you have an opinion on these new concepts of the content engineer or the marketing engineer?

Jenny (26:17.302) You engineer as in, can you explain that role a little bit more?

Niklas Buschner (26:21.999) Yeah, as like a, course, of course. so there are certain companies out there, US based companies. think, clay was the first one to introduce the concept of the GTM engineer. I think it worked very well. So if you trace it back, they were basically the first ones to, to, to, to call people like that. And now it's basically an established category. And now there are other companies that are trying to find a word for people that have,

Jenny (26:34.05) Mm-hmm.

Niklas Buschner (26:51.811) like the skillset of someone that's coming from a content from an SEO background, but that is also able to build workflows. Like for example, with an end and with Zapier with other tools. So they can also create content or create research, et cetera, with an agentic skill set in a way. Like maybe they also have a skill in, in cowork that gives them certain briefing or that allows them to take webinars and extract interesting quotes that you can then use in blog posts.

Jenny (27:06.914) Mm-hmm.

Niklas Buschner (27:21.637) that kind of things. And they are like very much pushing these roles and this category and this idea of someone that is able to build these workflows and build these tools, et cetera, and not only do the work in the way that we used to do it.

Jenny (27:41.474) I think it's a very, you basically just described the role of a senior strategist or senior AEO marketer in my team, because that is exactly what they're doing. We have never thought about calling that engineer. I don't know. It sounds nice though. Yeah. But that is exactly what the team is, is working on right now. Because one of the things, for example, that we notice within our inbound machine, a lot of this

Niklas Buschner (28:01.734) You

Jenny (28:11.308) The things that people still focus on is creating PDF files for people to download, right? Like the classical white paper. What my team is experimenting with is taking different approaches. So creating GPTs that people can sign up for or creating, like building their own HTML tools, little apps that people sign up for as the lead generating tool instead of a PDF file that would give you the same.

the end because I think people or the buyer right now, they don't want to like, no one wants to read a PDF anymore. If you can just ask tech BT stuff or Claude, I don't know if people in Germany or like in other regions are still using tech BT that much to be honest right now, but yeah.

Niklas Buschner (28:58.877) Yeah, now it's GPT 5.5 coming out. OpenAI will probably have a head start for two weeks and then Anthropic will drop Claude Opus 5.0 and then the cycle goes on. And I think in the end it's good for all of us because the race to making better models benefits us with more capabilities for our day-to-day work. yeah, let's see.

Jenny (29:01.76) Yeah.

Jenny (29:13.526) Yep.

Jenny (29:22.73) Yeah. Yeah, that's true.

Niklas Buschner (29:26.224) Let's tap into some of the experiments you've been running because this is obviously a big part of the HubSpot growth culture, running experiments, forming hypotheses, et cetera. And to kick that section off, I would like to start with a real example. So can you please walk us through an experiment that maybe worked better than expected? So something that surprised you.

Jenny (29:50.434) That's a very boring and simple one, to be honest. The one that worked the best. So we had a big problem on our HubSpot site for all regions. And that is that we noticed early on in our research and like tracking our prompt answers that

Niklas Buschner (29:55.504) I still wanna hear it.

Jenny (30:20.18) A lot of the pricing information that was shared about HubSpot was wrong. So, and then in the worst case scenarios, we had that happening to sales reps and conversations. So we had prospects that were in conversations with sales saying, but H2PT says this and that for your pricing. So, that is something that we do not want to have happen obviously. And the main issue with that was that,

Our pricing pages back then were built in JavaScript, which is not ideal for crawlers to be accessed from, LLMs. So what, what we did there was very simple, simple in quotation marks for every region, every language, we created individual, pricing articles on our blog. Very boring content. also said Housework Editorial.

no author on it, and it walked the LLM step by step through the exact pricing of HubSpot, which is different for each region. It's different for each package. You have different tiers, you have different hubs, you have different combinations, you have different new pricing systems coming in every now and then. So it's hard for the AI to really keep track of that. Before we launched those articles, we were measuring

How well is the AI answering the question? So we had a set of what does marketing hub cost in Germany in euros? And then we developed the scoring system with Claude saying that, if the answer is XYZ, then give a score of one to 10, making sure that we have correct currency, correct pricing tiers.

differentiation between different subscriptions. So we developed that scoring system first, had a run of like two weeks before we published the new articles. And the two weeks before, I think our pricing accuracy score was way below 20%. We launched the new articles in all languages at the same, around about the same time. And our pricing accuracy increased from that to, I think now it's around 80%.

Jenny (32:43.646) of pricing accuracy for all of the different questions. And that was just by creating content that is easily retrievable. It was very well optimized. we had different, very enclosed chunks of each of the content pieces within the articles. had different schema types added to make it easy for the LLM to trail the whole article and to...

just take the information that it needs. And that was so effective. It was scary. Yeah.

Niklas Buschner (33:18.994) That sounds like a classic one for like, basically every enterprise business that has in some way a little bit more complex pricing, different stages, different like product suites for example. So would you say, I mean, with experiments, it's always one thing to have an experiment that went well and like it looks good. Would you say that this is also like replicatable?

Jenny (33:19.97) you

Niklas Buschner (33:46.718) So do you think if you would take the same approach and like go to another enterprise company, not from the CRM space, but maybe from a different space, like let's talk about ESG solutions, like maintenance management software, something like that, like very complex solutions. Do you think it would work in any context basically?

Jenny (34:06.732) For sure. And I think it also goes beyond the scope of pricing. It also goes just around the explaining what the company does, if you will. Because another experiment that we launched was making sure that our products surface around the different personas when they chit-chitty stuff, when they Google on AI.

And that followed the same principle, except that it was just a different sphere that we were targeting with it. So we had persona-based optimizations. applied the same structural updates to all of those articles. So it works. should even work for explaining different products in an e-commerce space or in a little bit more explanatory e-commerce spheres. you, like, I don't know, take...

companies like Ferment, for example, that do kombucha and stuff. think if, I don't think they have a problem with AI visibility, but like if they were to optimize articles more around the explaining and chunking, I think that is the, that is an easy win that you could do.

Niklas Buschner (35:30.577) And two questions in one. First, how do you come up with ideas for these experiments? And then second, how do you validate what's worth pursuing versus what's just noise in brackets? Maybe that came from a LinkedIn post from some GEO guru.

Jenny (35:50.498) Yeah, that's a very good question. So a lot of our experimentation is happening or like the brainstorming is happening within meetings. So we have dedicated sessions within our AI grow hours. That's what we call them, where people just come in, present or talk about stuff that they have read somewhere or that they would think would be interesting to try. I think the main

The main advantage I want to say with that is that you see that people are intrinsically motivated to experiment. So they come up with their own ideas very, I don't want to say easily because I think it's, still a lot of mental effort to come up with experiments that no one has tried before. So a lot of

teamwork, lot of exchanging ideas. Everyone is within their region or within their city in certain communities where they go to network and exchange ideas with other people. And just like try around. Then when you think about, so what do we actually do? Everyone prepares a little project brief. basically similar to how you.

do CRO testing, you come up with an hypothesis. You think this is, I think this might happen if I do XYZ. You present that you say, okay, this might be the impact that we're, that we could see from it. And then you show that to the team or to me, if it has a impact on budget or impact on timeline, then we need to discuss that with me first. If it doesn't, then everyone has with.

along the lines of own and optimize has the possibility to try things out on their own. Just letting me know basically that this is what they're focusing on right now. A good example for that was a simple test that we launched on the Latam site was we were seeing within our tracking in the AEO product that a lot of citations

Jenny (38:11.51) throughout a certain content type or like topic within the marketing and CRM sphere was coming from LinkedIn Pulse articles that were really old actually sometimes, but they were structured really well. So we looked through all of the LinkedIn Pulse articles that were generating citations within the prompts that we tracked.

Those were not our LinkedIn post articles. That was experts that were talking about Household CRM or experts that were talking about Household marketing or just like completely unrelated. So we decided to run an experiment in three markets in Latam, France and Dach trying to see if we can influence if we generate more LinkedIn post articles. That was a fairly easy lift because

We had the content already there. A lot of the content was produced on the blog. So we created a cloud project that would create the perfectly optimized LinkedIn Pulse article. We had them published and revised by editors. And we saw that this did work for the latter market. It did not so much for the French. For German market, it's still inconclusive.

So we're still trying it out, but that was, yeah, that was just an idea that came out of nowhere. Just going through all of the citation data that we had.

Niklas Buschner (39:48.487) It feels to me like this skill of being able to engage in these communities, like forming a hypothesis, forming an opinion also, and like then being able to sketch out an experiment. feel like this is from what you just said, I feel like this is probably one of the most important skills now in the roles. Would you agree?

Jenny (40:16.8) Yes, yes, definitely. But it's also one of the most challenging I'd say, because if you, if you think about the way that everyone has this pressure right now on being AI driven, like you have to do a lot of stuff in AI. What do you do in AI? You do the basic things, right? Like, or like you automate the things that you...

Niklas Buschner (40:21.096) Mm-hmm.

Jenny (40:39.648) that you don't want to spend your time on your automate writing emails, your automate reporting, your automate copying stuff from X to Z, your automate research, which then leaves you for the not automate or not AI-ifying stuff with the tasks that have the higher mental workload, the higher, all of the higher mental capacity that you need to engage within. So that means that a lot of your

daily pressure is on being, still being creative, even though a lot of your job is already just taking on the bigger mental load because you did automate in a IEFI a lot of the basic simple tasks that you did before. While before that maybe 80 % of your job was mental, high mental work, 20 % was the basic admin stuff. So, and now that,

20 % where your brain was relaxing. That was actually the time where you thought your brain is relaxing, but that is the time where your brain can focus down and come up with creative ideas. That's where they are forming. So now we're moving that space and we're still demanding people to be even more creative than they've been before. So I think freeing up a little bit more time for the tasks that you might not want AI to do.

to make sure that your brain can still perform in the creativity level that you need your team to succeed on, I think is super important. One advice that I always give for that is to really reserve time on the calendar for everyone on the team. This is one hour where you do focus on your growth. You do research yourself. You do not only use Gemini research to do...

find out like what's new in the industry, you go through the articles yourself, you check around LinkedIn, you read up on what is new because you need that special time for that, for your brain to function.

Niklas Buschner (42:49.712) I always had this idea throughout basically my whole career from being a freelancer to then turning it into the agency to then growing the agency, which I called random exploration. And I, the idea was I just have an hour or maybe one and a half where I do something like randomly clicking through the website of our clients. Now, obviously from an agency perspective, randomly clicking through.

Jenny (43:01.698) Mm-hmm.

Niklas Buschner (43:17.278) the GSC, digging through a HubSpot report, which led to me being able to create a lot of dashboards and reports where I couldn't find tutorials for. And I always felt like these random exploration hours where I didn't have a clear goal other than, I want to dig into this client, this tool, this thing, without really a plan, just with like this intrinsic motivation, it really changed a lot of things for me. Would you say that this is a concept that

I maybe should get copyright on.

Jenny (43:50.914) You can, I think actually, that's no joke. I think that's something that, at some point, you will be able to run workshops on, because I think that is one of the skills that people, unlearn right now, because we are focusing so much on, trying to make everything work with AI. So a lot of what you just described is happening through AI. Like what's I think.

It's so easy to do all of that, what you just described and get an AI summary and then read through that. that, but then you're not processing it. You are not uncovering it yourself. So that means that it doesn't stick in your brain. It's just like stays in the short-term memory for five minutes and then you forget about it. But the actual uncovering all of those secrets that you just described yourself stay within your long-term memory and then.

and bake into what you do afterwards.

Niklas Buschner (44:54.044) Okay, so everybody listening, you heard the idea here first. If you register a company or a domain or whatever sensor, if you try to do it, don't try. I will either sue you or I will have done it before. My next company will be called the random exploration company. And it's, it's, it's thanks to, your very convincing arguments that I should pursue this Jenny. So thanks so much. Thanks so much.

Jenny (45:04.175) Hahaha

Jenny (45:10.912) Hahaha.

Niklas Buschner (45:21.856) Now you're obviously a very transparent person, which is why I also ask you to talk about an experiment that maybe didn't work out as expected or maybe where you had high hopes and in the end the results weren't quite what you had hoped for. Can you also share one of those?

Jenny (45:37.474) I think so one of them was having different outcomes for the LinkedIn post articles. That was definitely a surprise to me. What I, an experiment that did not work out and I actually did not like that much either personally in full transparency was

Niklas Buschner (45:45.16) Mm-hmm.

Jenny (46:04.786) that we, I noticed that within our experiment, we tend to like within our AO experiment, we tend to go back to all of the things that we tried in SEO in 2014. So there were, a lot of experiments that seemed to me very, repetitive of the strategies that we, that we did back then. so.

For example, summaries on our articles on things that we decided within our pruning that we do not want to produce anymore. So having articles just very, very TOEFL oriented did not work for us to gain AI citations. And we quickly dropped those again. Thankfully for me, because I personally thought I don't want to...

bury all of the content and then bring it back to life just because we're now focusing on AI search or answer engine optimization. That's why I was very glad that we could, like that we, that this experiment did not yet the results that we, that we were hoping to Yeah.

Niklas Buschner (47:24.434) You gave me the keyword tofu content, which is obviously, mean, shout out on that note to our friend, Marvin Miller from all my reviews who also had you on his German podcast or my tech check where when you would have used the word tofu, he would have hit a buzzer and then there was like an alarm and you would have to explain it. Yeah. I think our audience is totally capable of translating tofu to, a great Chinese dish. No, just kidding.

Jenny (47:43.681) true,

Niklas Buschner (47:52.992) Top of the funnel content. had Saskia Sarishin, I hope I pronounced it right. I do my best. from Meldwater on the podcast and she made an argument for, I repeat for a top of the funnel content. And I'd like to quickly get your thoughts on it because she said, yeah, for sure. Tofu content does not drive meaningful results in the terms of traffic.

Jenny (48:06.508) Mm-hmm.

Niklas Buschner (48:20.987) as this traffic does not really convert or conversions. But I still see value, so I'm quoting her, I still see value in it for the sake of topical authority and internal linking. What's your take on that?

Jenny (48:28.546) Mm-hmm.

Jenny (48:37.302) Yeah, I 100 % agree. think circling back to the experiment that I, or like not experiment, the strategy that we were trying, what you have to keep in mind here is that we do have a lot of content already on all of our platforms for Halfspot. So it's not like we...

It doesn't make a difference if we add more because we were already having the topical authority for a lot of the content areas that we wanted to be covered in. But for anything that we're doing new right now, for example, now with the launch of the AEO product, perhaps by way you can track like all of your citations and your visibility, we need to make sure that we have topical authority for those areas as well.

But we want to, I don't want to say we don't want to compete with PKI or SEMrush on any of the big visibility things because our AO product is mainly serving if you have a CRM integration, because you can inform all of your tracking and your prompts on the ICP that you have and within your CRM data.

But that said, that still means that we need to make sure that people perceive HubSpot as a thought leader and topic authority within the sphere of answer engine optimization. So for those areas for us, definitely TOEFL content is very important, but I'm not thinking about content, like we're not doing any content on what is SEO right now. Like we're not...

publishing new Wiki pages that are explaining stuff that has been out there and that a GPT or Google AI overview can answer in a second. What we do want to be is we want to be taken into consideration for the answer from the AI, right? That is what we're going at. So anything, whenever the conversation goes deeper from that TOEFL question.

Jenny (50:52.49) If we haven't been considered in the first couple of questions in the conversation that is happening in the AI, we will not be coming in to the later conversation either. So the AI, as we see through a lot of the prompting is trying to like, is serving from almost always the same patterns and the same sources throughout a conversation. It doesn't necessarily change too much in between. So that's why we want to make sure that.

We are there through all of the steps that are happening in that conversation, through all of the funnel, basically, like the whole conversation film.

Niklas Buschner (51:31.028) Hmm quick question around prompt tracking because it's something that is super it's a like highly discussed topic obviously and a company of your size so of the HubSpot size can you share maybe if not the actual number the range of Prompts your tracking because I can imagine that you are not Happy with the 25 prompts that most people track

Jenny (51:38.466) Hmm.

Niklas Buschner (52:00.657) as a standard given the size of the business and the different product ranges, etc.

Jenny (52:06.178) I think to be fair, tracking 25 prompts in whatever tool is better than tracking nothing. But I can't disclose the actual number, but I can tell you that we made it somewhat of a science to come up with the prompts and the queries that we're tracking. And that is because, as I mentioned earlier, we have a product suite. It is very...

Niklas Buschner (52:13.151) Mm-hmm.

Jenny (52:34.74) in intense sales service, CRM, Data Hub, all of the different products, they all have different personas. They all have different industries that they're serving. They have different company sizes that they're serving. So we tried to come up with a holistic strategy to track the most relevant queries as efficiently as possible.

Niklas Buschner (53:03.637) So my guess would be somewhere in the range of 500 to 1000.

Jenny (53:08.93) for... Well, I'm not gonna... So I can give you... So we have a big product suite. And we have five languages as well. So we're also doing that in all of our languages.

Niklas Buschner (53:12.157) For which product range? I don't want you to get in trouble.

Niklas Buschner (53:24.175) Let's just talk about one language. in one language I would say per language a thousand prompts.

Jenny (53:31.094) I think you're close.

Niklas Buschner (53:35.241) Nice, okay, yeah.

Jenny (53:36.482) But I think the more important question is what do I advise to or what would we advise the two of us obviously, because you also are invested in that, to any company. think having a starting point of 25 to 100 queries is completely fine depending on the product. I love giving SEO answers. It depends.

It obviously depends on what product, service, or what you're offering, how many personas or ICPs you're targeting, and what your main priority in AI visibility is.

Niklas Buschner (54:18.164) Yeah. Yeah. I would obviously a hundred percent agree. So who am I to say, but I would a hundred percent agree. It's better to have 25 meaningful prompts where you feel like that's what actual customers or actual, hopefully future customers are asking. And also things where you feel like this is what really drives the business forward. So certain features, certain aspects of the product, certain use cases, certain jobs to be done clusters. It's better to have 25 in that sense.

Jenny (54:34.519) Mm-hmm.

Jenny (54:42.806) Yes.

Niklas Buschner (54:48.36) In comparison to, example, tracking a hundred that are somewhere like top of the funnel where maybe not even every prompt that triggers a web search in the end, where you basically don't have any chance to then show up. mean, there are people now talking about influencing training data. I find this highly suspicious. like

Jenny (55:07.842) That does sound a lot like Black Hat SEO from 2008. But well.

Niklas Buschner (55:10.974) Yeah, yeah, yeah. It sounds like taking cash and just burning it if you spend money on trying to influence the training data. But okay, I take the answer like in the range of a thousand maybe per language. I think that already gives people a sense. You obviously didn't confirm, but you gave us a little bit to think on.

Jenny (55:38.99) I think the, as you said, to sum this up, think the, need to track the, what is this, like as much as possible with the least effort. think that is the, the thing you should be following there. And then always also think about, it's not, it doesn't make sense to just track words. You need to try and find the,

best semantic match to the main question that people are really asking.

Niklas Buschner (56:16.18) Yeah, a hundred percent agree. are also, I don't know exactly when people listen to this, but we are also currently working on, some research around how much the response of the AI changes based on a persona. So given that search is becoming more personalized and chat GPT takes into account and also other chatbots, the memory, there is, so it will probably be.

Jenny (56:33.282) Mm-hmm.

Niklas Buschner (56:46.09) different and there will be a different attachment to your actual prompt based on if you're a CTO, if you're a head of growth or if you're like a junior content manager. And we are trying to assess that and understand better how much the responses change, which is also, think super interesting. And you mentioned persona quite a lot already during the talk. And it's something where I feel like if we bury the traffic, the persona has like it's revival.

Jenny (57:13.12) Yes.

Niklas Buschner (57:13.142) where a lot of people were talking about how they don't need personas anymore, etc. So do you also feel like personas have become way more important again?

Jenny (57:20.962) Yes, and I'm very glad about that because before it was very impersonal, to be honest. And I remember you were saying in a panel at the, I think it was the Berlin SEO and Content Meetup. I don't know what you said, but it was along the lines of if you create content that is

has very low search volume, but it is targeted to the persona. The search volume that we get from Google is also very, it's an estimation. It's the same for all of the prompts. But if we have one person accessing that content and converting on that content, it's worth way more than any generic content that could be written for anybody.

where no one converts on because they don't feel engaged with content.

Niklas Buschner (58:21.962) Yep. Yep. A hundred percent. I mean, we can also, we can also not only draw this from marketing, even from sales, like if you work, for example, in sales and you have complex deals with buying committees, you have to practice stakeholder aware communication. So different people need different forms of packaging of the same content, so to say. So a CTO does not care about maybe the actual implications of

Jenny (58:37.378) Mm-hmm.

Niklas Buschner (58:48.8) how you do the the AEO work, but he rather cares about if I throw in my sales transcripts and all that kind of stuff over there, is it safe, is it secure, is it used to train the models, blah, blah, all of that. So I think this is also something that some people, but I would say not enough people think about and care about.

Jenny (58:58.529) Yes.

Jenny (59:10.4) Yeah, because also before it was easy to ignore the persona. Because you could like, it's easier to create content that is generic. That's just a fact. It is way more research intent to really think about the persona much more.

Niklas Buschner (59:15.019) Yep.

Niklas Buschner (59:28.0) Yeah, but search volume has never been the truth. We perceived it as a certain truth, but it has never been the truth. And I think we always are like not you and me necessarily, but I also don't want to like take myself out there. I think there was this tendency to look at search volume as a super reliable metric that we basically can pray towards, but we forgot that behind

Jenny (59:31.394) Yeah.

Jenny (59:49.772) Mm.

Niklas Buschner (59:55.928) the searches and this number in whatever SEO tool you use. There are actual people with actual roles and actual backgrounds that are searching for this kind of thing. And I always felt this when we were auditing a client's SEO performances and they told us, yeah, we have all this cool content and it doesn't convert. I would always ask them, can you describe a person that would search for that? And like,

Jenny (01:00:22.06) Yeah.

Niklas Buschner (01:00:23.131) And what other person could it be? And oftentimes for top of the funnel content, was, this could be this decision maker, but it could also be a student, a student that is just like researching for that kind of stuff.

Jenny (01:00:31.19) Yes. Yeah, that is, I mean, like that is one of the challenges that we have on HubSpot content as well, especially with the Academy, because the Academy is very, it has all of the buying stages as well, right? You have the people that want to get certified in content marketing, but then you have also people that want to get certified in HubSpot sales. So there is a certain overlap of having the actual people that you want to.

have and then students. But to be fair, think I always thought, like I always had this vision of that one student that now goes into HubSpot Academy in a few years will be a marketing manager somewhere in a company, has heard of HubSpot before and is pushing for HubSpot. That was one of the strategies that we also had in mind for.

especially for the Household Academy.

Niklas Buschner (01:01:30.988) I think it makes total sense because in the end you have, you just have to show up where people like engage with these topics, where they research, but, and you basically prepared the transition in a way that we wouldn't have even been able to script so perfectly. So I still see these two buckets of people. So I see people having conviction about

Jenny (01:01:48.705) Yeah

Niklas Buschner (01:01:57.75) we cannot track every single touch point and we are not able to connect every single measure to direct ROI in this or the next quarter. This is a little bit also taken from your Academy analogy. And then on the other hand, I have people that think about media mix modeling and super complicated attribution systems and the new attribution software. And we have to be able to measure everything. And if we cannot measure it, we cannot allocate budget.

What would you say, what would be your advice from your experience also to those people? What's the right way forward given the whole AI visibility ecosystem changes?

Jenny (01:02:41.282) I think that has been a challenge for me personally as well. We have a multi-touch attribution model on our team, which makes it hard to justify some efforts always, and it will always be. Like, best example, we have very, very popular newsletters in our regions.

Newsetters subscribers do not really count as leads, but we do have quite an overlap of subscribers that will become leads at some point. So how do we attribute the impact of the Newsetter? Still like open question. don't, it's not like I can, yeah, I can't like 100 % answer that, my, so we have different reportings where we can see this is, this is

Niklas Buschner (01:03:28.791) Cliffhanger.

Jenny (01:03:39.468) the impact that this new setter within the different region has on a net new leads conversion or net new sign up conversion.

That being said, even if that number was low, our main, our main North star for that newsletter, for example, is not just the way or like the amount of leads and MLR it brings in. It's also for one thing, we are an email marketing software as well. So we have to have a good newsletter. That is just one.

That's like, can't just not do it or invest less resources into it. We try to optimize for all of the channels where we can't 100 % attribute real positive ROI. We are trying to automate as much of the manual effort as possible. So that is probably not the ideal stage, but that is what I always advise to my team. If you spend...

If you don't see what comes out of it and you spend way more time on this program than you should spend on the other one that brings in more ROI, then this is probably not the best investment.

Niklas Buschner (01:05:03.223) Hmm. There's a final topic I want to get your opinion on because we somehow already touched it. we talked about AI, we talked about content. So let's talk about AI content. So, and there's a lot of discussion and also some controversial talk, which I generally appreciate because I think it's good that opinions are exchanged and real examples are brought to the table. So what's your take on AI generated content?

Jenny (01:05:16.321) O-A-I-Con.

Jenny (01:05:33.602) I'm not completely against AI-generated content. I think that AI-generated content should not be serving humans though, or like at least not 100 % AI-generated content. Because it's just not fun to read. I don't know. I think content should always be AI, or like not should always be, but content can 100 % be AI-assisted.

Um, so meaning research, brief, um, QA, but it has to go through human hands and eyes, um, and work, um, at all times. That is my, uh, the goal that I have with content. Um, fun fact, I am, I have my own website where I run a lot of experiments like that on, my own as well. And the, um, I have two articles on that website that are 100 % AI generated.

Um, and they are amongst the ones that generate the most citations for my prompts, where I think it's a very dangerous field that we're operating in right now to have AI quote AI is probably not the safest thing. That's why I think if you generate content with AI, you always need to read it, edit it, um, and make sure it's factual.

Niklas Buschner (01:06:48.984) Mm-hmm.

Jenny (01:07:02.786) factually correct because that's you like any AI model is hallucinating especially in the in recent months I think hallucinations have gone up by quite a lot so not pushing anything out there that you haven't read or seen or edited plus don't generate content just for generating content

I think a lot of the things that marketers are very easy to do is just generating a ton of content. I've done that myself, especially on my own website, just to generate content. always create content. Like that's why I say create content, not generate. Create content that you would want to read as well. It's the same for never send an email to your database if you don't want to read that email yourself.

Niklas Buschner (01:08:00.409) Playing a little devil's advocate now, if we look at where software engineering has moved to with all the model advancements like cloud code, Opus, 4.6, 4.7, a million context window, et cetera, there are more and more people that are transitioning from like a hands-on engineer role to an architect role. So they are conscious about certain decisions.

Jenny (01:08:01.921) Yes.

Niklas Buschner (01:08:30.072) have to like come up themselves. They have to think deeply through a problem. They have to, not only care about how something looks like, but what are the implications for infrastructure, blah, blah, blah. But then in the end, the actual code is written mostly more and more. And I think even Anthropic is sometimes sharing numbers that 90 % of the code for certain products is written with Claude code then, is written by AI. So.

Why shouldn't it be possible to move towards a content creation process that keeps in mind all the quality assurance and the necessary context and the human oversight, but where the actual word is then not written by us anymore, at least to the extent of 90%, for example, to draw from the code example, but written by AI?

Jenny (01:09:25.218) Well, did you know, you probably know because otherwise you wouldn't have brought it up, but if you write code with Claude, you don't have the copyright on it. It's open. It's an open sphere. So that applies to content then as well. If it applies to any thought that you've put into, if you didn't create it yourself, you don't have the copyright to it. Easy as that. So, um.

Niklas Buschner (01:09:29.442) Hahaha.

Niklas Buschner (01:09:40.034) Mm-hmm.

Niklas Buschner (01:09:44.76) Mm-hmm.

Niklas Buschner (01:09:49.868) Mm. Yep.

Jenny (01:09:55.532) Does that answer your question? Yeah.

Niklas Buschner (01:09:57.481) And not, not, not, not really, but I think it's an important point. was already thinking about getting someone from the legal space on the podcast to think about all this copyright implications, because my sense would be that the idea of copyright, how we have it codified, for example, in German law is somehow coming to a limit. And I'm not saying that copyright is not important. I'm saying it is important, but as always.

Jenny (01:10:06.165) Yes.

Niklas Buschner (01:10:25.258) Regulation is way behind technological innovation. probably there has to be. So I would rather, who am I? I'm not a lawyer. I have a lot of lawyers in my family, but I could see a necessary regulatory change that gives me copyright on certain aspects of code or written words that I have produced with the help of AI, but that are based on my thoughts. It's obviously hard to distinguish.

Jenny (01:10:29.495) Yes.

Niklas Buschner (01:10:54.348) But I don't feel like it's right that if people have, for example, if you would have put a lot of ideas and certain tools you built for HubSpot that you have no right whatsoever taken aside all the, if the company owns it, if you owns it, but that there is no protection whatsoever on the thing you produce because it obviously comes from your brain power.

Jenny (01:11:07.66) Hmm.

Jenny (01:11:15.33) But I want to argue against that, especially for content creation or any creative work. The models are trained on existing things. So it's not like they come up with anything new and the same applies to code. So if you don't sit down yourself and you don't write it yourself, can you compare with someone who has done that? Who has done the...

own work in it? Do you have the same right to that copy while half of the work basically is stolen from and because you didn't do the training yourself, Claude did it for you. Like that is like the whole ethic conversation that we should be having in the law sphere when you think about that because it's not comparable. I sit down, so best example, my dad's a programmer. If I sit down,

and program something that he could have programmed, can I compare myself to him?

Niklas Buschner (01:12:20.481) It's a very important question. don't have the answer to that, but I agree. I agree that it's something that we have to, we have to talk more about these implications. think we often only talk about AI from either like marketing or whatever background would we also have to talk about what changes. And also, I think, especially the way you framed it, we as humans have to decide.

Jenny (01:12:22.484) Yeah, because it's like...

Jenny (01:12:30.348) Yes.

Niklas Buschner (01:12:47.097) how we want this to be handled. do we think it's the same? Does someone deserve the same amount of protection or should it be handled differently? I think it's very important questions. This is why I also started thinking about getting some run on this podcast from this field. But yeah, I agree with you that it's not a trivial thing and

Jenny (01:13:06.145) Yes.

Niklas Buschner (01:13:15.797) I just always try to challenge our way of thinking and concepts we have from the past that are maybe not a perfect fit anymore for where technology is heading.

Jenny (01:13:33.088) Yeah, I love that. That was very deep in the end.

Niklas Buschner (01:13:35.576) Wow. love that. I love that. Okay. Crazy. who would have thought that we go from, traffic burying ceremony, traffic's a funeral to leadership principles around how to steer teams through change to AI search improvements to that, to copyright. Love that. yeah, yeah. let's.

Jenny (01:13:59.776) the ethical dilemma of AI.

Niklas Buschner (01:14:04.925) So looking at how long we've already been talking, I hope people enjoy this as much as I'm doing right now, but I still want to wrap it up a little bit. And something that I would like to ask you is, what's, I would say, the biggest misconception or maybe the biggest confusion about AI search that you keep running into?

Jenny (01:14:32.098) I think the biggest one that I see happening every day, and I am probably gonna get a lot of comments about this, for me, having this misconception of AEO or DO is just SEO, that is I think the biggest one. Because it's not, I think we spoke about this a little bit earlier.

True that certain aspects are similar, you have a lot of, but it's a different skillset. It's a different, it's different teams working together. Yeah, full stop.

Niklas Buschner (01:15:15.597) Yeah, I think you have a strong argument and I also think you have very smart people in your corner on that. Not only from Germany and the Dach space, also from the US, for example, people that really lead the charge in this field. So thanks for sharing that. Now I have a final question, which I... Giving copyright and giving credits...

stole or got inspired from Lenny's podcast, but I always say it like that. So shout out to Lenny, who is probably a fan of this podcast. What didn't we talk about that we should have talked about?

Jenny (01:15:56.77) Ha

Jenny (01:16:03.575) I actually, maybe I did not read the questionnaire to the end. I'm sorry. But no, think, no, sorry. I think we did cover a lot of the things. I'm thinking we could have done more Devil's Advocate. I really enjoyed that in the end. I think there's a lot of things that there have been discussions, especially around AEO at the moment, that people

Niklas Buschner (01:16:11.193) You're exposing the whole behind the scenes process. No, just kidding.

Niklas Buschner (01:16:31.225) Hmm.

Jenny (01:16:37.266) are not talking about or that are talking about. Maybe one thing that we should touch upon in some shape or form, or you can within your podcast is the dangerous behavior that is happening right now in different spheres, where you have a lot of people that say they know all the truth about AEO. And then there's a lot of people that just don't engage in those conversations.

because they feel scared that they might get a negative comment because they have a question about it. I think that is one of the things that are dangerous right now. Having answer engine optimization being this one big topic that all SEOs and content marketers are focusing on at the moment. There is a lot of timid behavior in

in the communities where people are just afraid to ask. I think that's something that we should have touched on a little bit as well. But I don't, I should not open that now because then we will be here for another hour.

Niklas Buschner (01:17:45.879) Yeah, that's a great point.

Niklas Buschner (01:17:51.877) Probably, probably, but we can at least share a quote from the podcast I did with Lily Ray, where she said that, so she talked about the humility and she said, quote unquote, nobody fully understands how these systems work. Not me, not the people at the large language model companies be skeptical of anyone claiming otherwise.

Jenny (01:18:00.407) Hmm.

Niklas Buschner (01:18:19.628) Invest in the things search engines and LLMs can't take away from you. Original research, proprietary data, expertise, a brand people actually talk about. Everything else is rented land.

Jenny (01:18:24.375) Yes.

Jenny (01:18:29.302) Yes, I 100 % agree. And I think we should push that message out because a lot of people are just...

scared about making doing something wrong. But as Lily said, there's nothing you can do wrong with your own data, with your own research, with your own content. The rest is green fields.

Niklas Buschner (01:18:54.436) Yep. Word. Let's conclude the episode. It has been insane. I really enjoyed it. I hope everybody can feel it if they listen. We have two final notes. The first one is, if people want to follow you around, what's the best place for that? We'll put the link in the description, obviously. Okay.

Jenny (01:19:17.418) LinkedIn.

Niklas Buschner (01:19:19.51) and your website where people can follow your experiments that you're doing that you just mentioned.

Jenny (01:19:25.482) the website is Märchen Brause, which is a fairy tale podcast. that's where I run the experiments on. so there's like no summaries on it, but if you're just generally interested in old fairy tales, in the original ones, I'm doing a lot of, research on it. So little behind the scenes, I have a, I have a degree in literature. so that is my guilty pleasure is going through.

Niklas Buschner (01:19:28.346) yeah, that's this one.

Awesome.

Niklas Buschner (01:19:37.924) Mm-hmm.

Jenny (01:19:53.352) old fairy tales from finding out the stigmas and all of the stereotypes that have been baked in there. And it so happens that it's nice to optimize those for SEO or AEO. Yep.

Niklas Buschner (01:20:06.212) So cool. mean, probably for people, they, so what I would do, for example, is I would go to that site and if I see that there's some sort of update or whatever, I would say, okay, what is Jenny testing there? It could be something that she's tinkering on for testing at HubSpot. And then I can, then I can also draw my own conclusions from that.

Jenny (01:20:18.594) You

Jenny (01:20:27.04) You can, yeah, sure. Try it out. also gave a, so I'm using HubSpot to host that website. So I had the very, very early, early, early beta of the AEO product on it already. So there's a lot of things that I'm testing through that. And there's a summary of it on the HubSpot community. I posted that yesterday where you can see the actual like...

Niklas Buschner (01:20:35.011) Okay.

Niklas Buschner (01:20:41.944) Mmm.

Jenny (01:20:55.93) no performance of the website in the air search at the moment. But that's also because my queries are a little bit stupid. I have to revise all of my prompts that I'm tracking for Märchen Brause. Yeah.

Niklas Buschner (01:21:03.866) Okay, so cool. The second announcement we have is, and we will probably do another episode before we actually do that, but that we will do our first conference on October 22nd in Berlin. Location still TBD, but we have two awesome locations in the final round. And you, Jenny, will be on stage.

Jenny (01:21:32.002) Yay!

Niklas Buschner (01:21:33.39) giving a keynote, sharing great stuff, talking about growth, talking about AEO, talking about what has worked, what hasn't worked, actual behind the scenes looks, this will be great. We will obviously have some speaker announcements, et cetera, coming out, but still for everybody that is a true fan and listening to that, if you want to have, like before we have landing page and everything going out, if you want to have a reserved seat,

for the tickets, which are obviously limited, so it's not a thousand person conference. We will in the end be 175 people. Then reach out on LinkedIn and I'll make sure that you get one. But this is super awesome and thanks so much for being part of that.

Jenny (01:22:11.907) nice.

Jenny (01:22:19.661) Thank you so much for having me. I'm really excited because also until then it's some time where I probably share a lot more other interesting stuff that we've tried out so far.

Niklas Buschner (01:22:25.476) Yeah.

Niklas Buschner (01:22:31.342) Yeah, very much looking forward to that. So I definitely won't miss your keynote. Jenny, thanks so much for taking the time today. It has been an awesome recording. This is why we are like almost at 90 minutes, but time flu. I appreciate that so much. And I wish you all the best and your team at HubSpot. hopefully we'll speak soon about everything that is still to be spoken about.

Jenny (01:22:53.741) Well, thank you so much. Bye.

Niklas Buschner (01:22:55.364) Thank you so much, bye bye.