AI Just Broke Marketing (And What You Need to Do Now)

Summary

Podcast Summary: Marketing Against The Grain

Episode: AI Just Broke Marketing (And What You Need to Do Now)
Date: November 18, 2025
Hosts: Kipp Bodnar (HubSpot CMO) & Kieran Flanagan (HubSpot SVP of Marketing)

---

Episode Overview

Kipp and Kieran deliver an unfiltered, real-time analysis from London on how AI has fundamentally “broken” traditional marketing—and what innovative marketers must do to adapt. Drawing on their own transformations at HubSpot and industry-wide shifts, they present hard data, tactical playbooks, and actionable trends for AI-driven marketing, B2B brand building, personalization, and content creation. The tone is candid, humorous, occasionally nerdy, and empowering, with an emphasis on embracing change and experimentation.

---

Key Discussion Points & Insights

1. AI’s Seismic Disruption: From Volume to Value

• AI is the biggest shift in marketing in years: Traditional measurable demand from search and paid ads is harder, while AI provides better ways to monetize demand.
• "We were in the era of volume... Today I think maybe we're in the era of value where we can acquire less measurable demand, but we can have way more ways to extract more customers from that demand." – Kieran [03:34]

---

2. AI Content has Redefined the Internet

• 74% of new web pages are AI-generated ([03:46]), prompting a shift from brand trust to individual/creator trust.
• "The Internet is massively becoming an AI content Internet versus a human-created content Internet." – Kipp [03:46]
• Consumers now prefer content from trusted individuals/creators over brands due to AI spam and overload, accelerating a trend from B2C into B2B.
• "Creators are going to become a core part of your content strategy, but also a core part of your distribution engine through paid." – Kieran [08:09]

---

3. Search is Broken: Rise of ‘No Click’ and AI Search Engines

• Search results are being cannibalized: “8 or 9 out of every 10 Google searches in AI mode now result in no click.”
• "The most fundamental disruptive thing... my clicks are disappearing. What do I even do?" – Kieran [13:42]
• AI assistants like ChatGPT, Gemini, and Perplexity are overtaking Google’s 'blue links' and will soon surpass organic search volume ([15:00]).

Tactics: Answer Engine Optimization (AEO)

• HubSpot increased share of voice in AI assistants by 800% in under a year ([19:59]).
• AEO is more like brand marketing than performance: Less than 2% of HubSpot’s measurable traffic now comes from AI assistants, but this presence is critical for long-term relevance.
• Strategies:
  • Leverage tools like Xfunnel for AEO at scale
  • Generate hundreds of content variations for niche product questions
  • Focus on where LLMs (AI models) source their content and secure strategic citations ([22:45])

---

4. The New Data-Powered Personalization

• Proprietary data is now a core marketing differentiator.
• Google AI Max and similar tools automate targeting, ad creation, and campaign optimization better than most human teams ([24:47]).
• Marketer’s role is shifting: “Our role is to make sure they have the best data set to build the best targeting.”
• The path forward: More campaigns, more granularity, more niche segmentation ([27:04]).
• Email is revitalized through AI-based micro-personalization; click and conversion rates soar with these new tools ([29:04]).

---

5. AI Prospecting: Key to Scaling Sales

• Inbound AI prospecting (triggered by user actions) vastly outperforms outbound with generic data.
• HubSpot achieved 35–40% increases in monthly booked meetings from AI-powered email prospecting ([29:45]).
• Success is tied to strong data and tailored prompt engineering—not generic messages. It took HubSpot six months of testing to achieve breakthroughs ([31:45]).

Letting Go of Control

• Marketers need to accept "good enough" AI-generated emails for scale:
  • "If you want every word of an email to be perfect... you're going to really be at a disadvantage in the next era." – Kipp [34:03]
• The focus should be on optimizing prompts, not copyediting every output.

---

6. Multimodal is The Next Frontier (2026 Trend)

• "Your website is now part human, part robot."
• Multimodal agents (AI that can use text, speech, and visual demos) are drastically increasing engagement:
  • 5–9 minute average sessions with avatar chat agents on HubSpot’s site ([36:18]).
  • Multimodal avatars conduct demos and discovery, pass summary to human sales, turning reps into closers.
• "It's the first time we're starting to market to agents, not consumers." – Kieran [34:57]

---

7. Short Form Video and Creator-Driven Paid Distribution

• Short form video is now the primary way B2B buyers prefer to learn about products.
• The success playbook:
  • Invest 5% of your paid ad budget in creators to generate and distribute short-form videos.
  • "If you can master short-form video, you can master every single type of content." – Kieran [39:52]
• Modern short-form marketing mirrors the infomercial—but in 30-second bursts.

---

8. Actionable Takeaways for 2026 Marketing

• Build and scale an Answer Engine Optimization (AEO) playbook.
• Invest in data infrastructure—differentiate with proprietary data and context.
• Shift your content strategy from brand-led to creator/individual-driven.
• Embrace prompt engineering for personalization at scale.
• Test and invest in multimodal AI agents for qualifying leads and demos.
• Integrate short-form video and collaborate with niche creators.
• Let go of perfectionism—move fast, iterate, and let AI optimize outputs.

---

Notable Quotes & Memorable Moments

• "[AI] has completely killed the marketing that we all knew. The old rules are dead." – Kieran [10:42]
• "You can't shape consumer habits. They will be shaped by the platforms and the tools they use." – Kieran [17:39]
• "Brand marketing is coming back into the circle." – Kipp [20:21]
• "Prompt is your product." – Kieran [31:45]
• "If you want every word of an email to be perfect...that's the fact of the future..." – Kipp [34:03]
• "It's the first time we're starting to market to agents, not consumers." – Kieran [34:57]
• "Short-form video is an incredible tactic to not just acquire attention, but market your products and services. The new virality today for AI startups is the product demo." – Kieran [39:52]
• "Taste means, have I seen tens and hundreds of thousands of this thing, can I then arbitrate what good looks like and what bad looks like, and then can I work with AI to build the good version over and over again?" – Kipp [43:39]

---

Important Timestamps

• 03:46 – 74% of new internet content is now AI generated
• 13:38–15:52 – 80% "no click" searches; Google’s AI mode kills traffic; ChatGPT forecast to overtake organic search
• 19:59 – HubSpot boosts AEO share of voice by 800%
• 29:45 – 35%+ increase in booked meetings from AI-powered prospecting
• 34:57 – Multimodal agent and chatbot engagement breakthroughs
• 39:52 – Case for B2B short form video and creator-led paid strategies

---

Tone & Closing Vibe

The show is equal parts pragmatic, visionary, and wryly optimistic. Both hosts emphasize that while change is daunting, those who experiment—and let go of old playbooks—will thrive in the new AI marketing era.

"I'm an optimist, and I've never had as much fun doing marketing as I have had in the last year...I can't imagine doing anything other than marketing right now." – Kipp [45:33]

---

Final Action Steps

1. Prioritize AEO: Get your content into LLMs, optimize for answer engines, and track share-of-voice in AI responses.
2. Double Down on Data: Unique, integrated data is key to targeting, personalization, and protecting your moat.
3. Go All-In on Creators and Short-Form Video: Partner with niche creators and experiment fast—B2B is catching up to B2C.
4. Invest in Multimodal and AI Agents: Start testing website chat avatars for qualification and demos; let AI qualify, let humans close.
5. Let AI Personalize, and Let Go of Perfection: Empower AI-driven messaging—perfectionism kills velocity.

---

This episode is a must-listen (or must-read) for any marketer feeling the ground shift beneath them, and offers both reassurance and a blueprint for what to do next.

Transcript

[A]

Hey, everyone. Welcome to today's show. We have a special show for you today. We're here live in London and we are going to break down all of the things we've learned in using AI in our marketing in the last year. And I'm going to share them all with you today. Kieran, what are we going to tell them?

[B]

We're going to tell them how search has been transformed over the past two years and how we've managed to increase our visibility in AI assistance by over 800%. But that's not all. We're going to talk to you about how you do real AI prospect, and we've managed to increase our book meetings by over 30, 40% month on month. And we're going to give you the exact playbook. But that's not all. We're also going to tell you why multimodal is our biggest trend in B2B for 2026 and some of the experiments that we are running to show you how it can transform how you sell your products.

[A]

So we're really excited to be here, share all our learnings with you. Let's get into today's show. Hey, everyone. Hey. Thanks. Thanks for having us, everybody. This is a very unique treat. We normally make a podcast where I'm in a little dark room in my basement and you're in a little dark room in your house. It's a little different.

[B]

We are doing it in front of real people, not AI bots. I'm still not sure which one I.

[A]

Prefer, but we know you prefer the ipods. It's okay. Well, thanks, everybody for being here. We host a podcast called Marketing against the Grain. We're going to do a live episode for you all today. And it's really about everything we've learned in the last year of using AI in our marketing and the things we've learned and we're kind of framing it in a way of the web and marketing have changed so drastically that we've got a handful of crazy stats and action items for everybody here to go and do. And so, Kier, maybe you can set us up with what the heck has changed? Like, how did AI mess up everybody's marketing over the last year or two?

[B]

Yeah, like, how do you even do marketing anymore? I think this has been one of the most seismic shifts that we've ever seen. I fell into marketing. I was a software engineer. I was not a very good software engineer. I clearly would have been a lot better if I had Claude Code and some of these AI assistants. And so I went into marketing. And I was in marketing kind of the era of like SaaS, right? Like when inbound was really taken off and we had all of these new channels who acquired demand and we could convert that into customers. And it was all very measurable. And what I see happening this year is really a kind of bifurcation in terms of how we do marketing, in that AI is disrupting how we acquire demand. Who would have thought that Google's blue links would have been disrupted within the space of three years? That's pretty huge. And what's happening is because search is being disrupted, most businesses are putting money back into paid advertising. And so we see the cost to acquire customers from paid advertising continue to go up. So it's getting much harder to acquire measurable demand. But AI has given us way more tools to monetize that demand. And so the way I think of that is we were kind of in the era of volume. We were able to acquire a lot of volume and convert some of that into customers. And that volume was quite measurable through search, through paid advertising, through some of these other channels. And today I think maybe we're in the era of value where we can acquire less measurable demand. We're going to talk about that on this episode. But we can have way more ways to extract more customers from that demand. And I think that's where AI really shines today, is the ability to convert more of what you get.

[A]

AI has really changed how we're thinking about things, but it's not just like what tactics marketers should do. What we want you to understand is what are the inputs that's driven all this change. And the first one we want to cover here today is Kieran. That AI is drastically just changing what the Internet looks like and what's wild to me and the stat that's probably blown my mind the most recently is 74% of new web pages contain some type of AI generated content. The Internet is massively becoming an AI content Internet versus a human created content Internet. And that has a whole host of implications, maybe break down why this is so important and what marketers should think about for us as we're thinking about this changing landscape of the Internet.

[B]

One of the most mind blowing charts I saw last week, I think it was put out by a friend of the show, Ethan Smith from Graphite, where they did a lot of research on how content is created today on the Internet. And we have now in the last month started to reach a point where more content is created by AI versus humans. And I Think that has really big repercussions. Now I will say one of the worst uses of AI, and I say this every time I talk. It blows my mind. When I was a small child, I've always been a nerd. I was into sci fi. And when I started to use AI tools, I couldn't believe it. We have AI technology. I'm living long enough to see this come to fruition. And we give those tools to marketers. And some marketers say, you know what I'll do with this incredible sci Fi technology is I'll automate my LinkedIn comments. And I'm like, wow, what a great use of AI. But there's a reason I'm using that as an example, because there's two things that I think will happen. The more content is generated by AI, the more. The more I think people will switch towards trusting content from someone they trust, like an individual versus a brand, and the more slop we see. And there's a great example from people who do not work in AI that they are starting to see this in their feeds. I have multiple different WhatsApp groups and one of them is my friends who have no interest in technology. I can't talk to technology about them, I can't talk to AI. But them, they don't care. But in the last week when Sora 2 came out, they kept texting each other saying, why is Stephen Hawkins all over my feed in Instagram? Every second video is a Stephen Hawkins video. And the reason was because people were creating them with Sora too. And they started to say, I'm not going to even use Instagram anymore. It's becoming full of spam. And I think that's going to cause people to want to consume content from individuals. And we have always thought about B2B looking like B2C over time, where you actually can see the person shifts away from brands and shifts towards creators and individuals. And I think that's one of the biggest trends we're going to continue to see in B2B.

[A]

Yeah, so we have this shift in AI generated content and it happened with text very fast. And now with Sora 2 and these new AI image models, you're just going to see AI generated content everywhere. Right now, it's a big deal. If somebody makes like a video ad with AI, that's just going to be commonplace over the next year. So I think you outlined some of the big implications here, which is people have gone from caring about and trusting brands to caring about and trusting individuals. And we even see that, like on the HubSpot, like our LinkedIn strategy, like, nobody cares about our HubSpot LinkedIn profile anymore. They care what Yamini has to say, what you have to say, what our head of product has to say. All of those individual profiles are now what the brand actually is. And that is fundamentally very different than how most companies in Europe or anywhere in the world have been thinking about telling their stories and influencing their audience.

[B]

Right. We got kind of lucky with this. Where you and I had seen the trend in B2C and the way we've always thought about building marketing engines is if you looked at B2C, the same trends will happen in B2B just three to five years later. And so we acquired a company in 2021 called the Hustle because we wanted to be much more in the creator space and much more in the individual content creator space. It has a really popular podcast called My First Million. Sampar was the founder of the Hustle. And one of the incredible trends, if you look at, and we are known as the inbound company, we generated a lot of demand from these kind of brand led channels. Today we generate more content from human LED channels, which are individuals creating content on YouTube newsletter podcasts, than we do from traditional brand led channels. And I think this is a trend for everyone in the audience to pay attention to because I think creators are going to become a core part of your content strategy, but also a core part of your distribution engine through paid. And we're going to talk about that as well.

[A]

So I'm going to actually say something nice about you and to you. Wow, okay.

[B]

Can I delete my phone?

[A]

Yeah, I was going to say if you want to record this from your perspective, please. When we acquired that Sol, you had a slide. And I imagine there are a lot of people watching on YouTube or in the audience here who are like, oh, I have this good SEO strategy. I have a lot of content on my website, but I don't have a lot of content other places. And I don't have this diversified content strategy yet. And Kieran had this brilliant slide which was like, hey, before we did all of this, it was basically like, here's our audience and how they spend an average day and we show up in bits and pieces of that day. We when they search for stuff on Google and then you're like, well, after we do this, we're going to show up when they listen to a podcast on their way to work in their morning, when they open an email newsletter on their lunch break. And it was this really powerful before and after. Oh, we are much more integrated into the touch points of how people consume and interact with information now that we've kind of diversified this strategy than we were before. And so for anybody who's thinking about trying to diversify that, that's just like a great simple heuristic to communicate it with inside your company or your team. And in the YouTube comments, we'll try to get a, get an image of that slide and drop a link to that in the, in the YouTube comments.

[B]

Yeah, we did that as part of the pitch to the board to acquire the company. And what we really showed was when you are a B2B company and you're optimized for search, which we had historically been, you were a part of your customer's life when they asked a question, when they searched for an answer.

[A]

Right.

[B]

So you showed up when they needed an answer. And what we wanted to be is we wanted to show up as an ever present part of their lives. And to do that you have to be part of the channels that they not just when they're asking something. On Google. Now, we did get somewhat lucky in that we saw the trend B2C to B2B. And AI is 100% accelerating this. It is 100% accelerated the shift for people to consume content from individuals. I would say if you just think about your own content consumption habits, I wonder how much you consume from individuals versus how much you consume from brands. I bet you it's much more on the individual side.

[A]

I think that's really true.

[B]

Look, AI has completely killed the marketing that we all knew. The old rules are dead. That's why we're giving you the AI Marketing Rebuild Toolkit. It includes 10 prompts to fix your traffic, ads, email and video strategy for how people actually search and buy. Now, if your marketing feels broken, this is the reset you need. Scan the QR code or click the link in the description.

[A]

Click.

[B]

Now, let's get back to the show.

[A]

Speaking of individuals, one of the things that I love about humans is the legacy of their online profile names. Like what are the crazy profile names that they created when they were earlier in their life or earlier in their career. And for those who do not follow you on X and Twitter, I still don't want to call it X. I still want to call it Twitter. So people who follow you on Twitter, your Twitter handle is search brat.

[B]

Yeah.

[A]

And because you started in the SEO world was the you failed at programming.

[B]

I will say in the gray hat SEO world, I think you're being nice.

[A]

To Yourself calling it gray and not black. I mean, you had some different websites where you were doing a bunch of borderline SEO practices after you were a programmer. And the second thing we want to talk about today is how drastic the change in search has been. And you know, you as the search brat and SEO interested in this world have long been a student of search. But man, the data around the amount of searches that are not ending in a click and what is shifting out there is very dramatic.

[B]

It's incredible. I think this is like just such an incredible change. Now I have a really soft place in my heart for SEO, because when I was a software engineer and I wasn't very good at it, I was pretty lost in terms of my career. I was like expected to be an entrepreneur, I expected to build things, I expected to have a very successful career. And then I realized I was an average programmer and that's like not a nice realization. And search gave me something to actually excel at because it took some technicalities and some creativity. And when we first used ChatGPT, I remember we did the show in November 2022 and we had used it and we talked on that very show that this was going to be hugely disruptive for Google. And actually a lot of people disagreed that Google's blue links would be steadfast, they would never get disrupted. But why did we see that? Because if something is easier and faster, humans will always gravitate towards it. And everyone kind of thought, well, it's not as accurate. I don't think people care. It's easier, faster, wins each and every time. So if you look at the data, traditional kind of Google blue links have already moved towards no click. Like we had feature snippets, we had knowledge boxes, it was always moving in that direction. I think six out of every ten searches would result in a no click. The data with AI mode added, which is Google's, hey, front and center. We want you to use AI before you even look at the blue links. In terms of their beta test, they are seeing anywhere between eight or nine searches result in no click out of.

[A]

10 out of 10. It's like an 80% of searches, just no clicks back to the website.

[B]

Because AI does a great job of answering that question. And so we are starting to see clicks disappear. And when people say, well, hey, search volume is increasing, it will increase. Because AI is a different experience. You don't search in keywords, you have a conversation. So search volume is 100% going to increase. And that is not the thing to look at the thing to look at is the disappearance of clicks. And that is the most fundamental disruptive thing I think happening probably to a lot of people in the room as well. Like hey, my clicks are disappearing. What do I even do?

[A]

Well, first of all, the fact that ChatGPT is the top five most visited website on the Internet in three years is just like mind boggling that, that, that is where we have gotten to in such a short period of time. And you have AI mode getting rolled out very aggressively on Google Gemini. If you're AI dorks like you and I, you're using Perplexity Comet or OpenAI Atlas as kind of like this AI first browser and you have forgotten what the 10 blue links even look like because we're just having conversations with our AI all the time, which is both good and bad. But that is, I think the experience that we're having is emblematic of what the mainstream experience is probably going to be over the next couple of years.

[B]

Well, AI mode puts it into ChatGPT is still semi not mainstream with the average person who's not at the tech. It's still somewhat adopted by people in tech in certain sectors. I think AI mode, because Google is the front door for the Internet will help to onboard huge amounts of people who are non tech related onto AI as a search assistant. This here chart is also pretty incredible. In 2028, ChatGPT is forecast to overtake organic search. And actually if you add in AI mode in Google, there's a good data point out there that says over 75% of all organic search will be consumed by AI first. And again, that is just such a seismic change in the way that we market and the way that we actually grow our businesses. And maybe what we should do is give everyone some tips on what we've been doing about it and what we would do about it if we were you.

[A]

Yeah, so obviously we've established that you have to play here. One piece of interesting information is that 89% of B2B buyers have already cited that they're using AI search for, you know, buying decisions and research. So it's already here. It's not just a consumer quick, quick response research tool, it is a real business research tool. And so when we think about what we've been doing here, we've been really, I'd say in the last like 15 months, heads down in this search engine problem. And we were in denial for a little bit and it was, it was a rough time. We tried to do a lot of things in the short term to try to get more traffic from Google and traditional search. And that worked a little bit, but didn't work great. And we decided instead to really lean heavily into answer engine optimization and how we were going to show up in these modern search engines. And it really completely transformed our content strategy and the just way we approached the problem. And maybe you could outline for us what are the big things that shifted in how we did that.

[B]

Yes, you have two types of search, where you have informational search, where you can create a lot of educational content, how to content teach people to do something, and then monetize a small percentage of that into customers. And then you have transactional search. And transactional search is people searching for things around your product or services, like the best X product, the best 10 products that do something. And so there are two fundamentally different types of search. And so we saw educational search get disrupted much more rapidly. Which makes sense because what happens with these AI assistants is they turn information into action. And so the user used to have to say, what are 10 ways that I can optimize my Facebook Legion ads? And now they can say to AI assistant, optimize my Facebook Legion ads. And the AI assistant is using your content to be able to do that. And all of that has been consumed by the AI assistants. And so that has been cannibalized pretty, pretty rapidly. And so actually what we decided was that is going to happen. That is a trend that you cannot fight against. There's nothing that we can do. We can make our content better. We can integrate something into our unique data points, unique customer case studies. You need unique examples, things that aren't in the LLMs into that content. But ultimately that's going to happen because AI is just a better experience for the user. Transactional search, we were actually much more bullish, would not get consumed by the AI assistance. So it would not become the kind of platform of choice when people are trying to decide what products they want. And we have started to see that also start to be cannibalized by the AI assistants in the last six months. So a lot of product recommendations, a lot of transactional search has started to be cannibalized by the AI system. So you can't fight against the trend. Go to markets are built based upon where consumer habits are going, not where you want them to go. And I think that's a really important thing to understand as a marketer. You can't shape consumer habits. They would be shaped by the platforms and the tools that they use. And so you have to actually integrate your marketing plans into these AI assistants. And that's what we've been doing. We've actually had a AEO POD AI Engine Optimization podcast stood up for the last year and we've seen huge results in terms of our ability to increase our share of voice. Or share of voice is if I implement and add a bunch of questions to AI in terms of how I think my consumer is going to ask questions about my products and services, how many times do I appear in the answers? And that's kind of your share of voice. And we'll go into some of the tips on how we did that. But the stats for us, there's two core stats I'll share and then we can maybe get into the tips. We've increased our share of Voice by about 800% in the AI assistance over the last six to eight months, which means that we're much more visible whenever we have questions asked about our product or services. And when you look at our vertical, we're the most visible out of any other software in terms of how many times we appear in those answers. And we've increased our referral traffic from AI assistance by around 1,400%, which is like measurable traffic being clicked through, through and coming to the website. And we can see it comes from ChatGPT, we can see it comes from Gemini. Not so much Claude. I don't think Claude are actually trying to play in that space, to be honest. But. And this is the big but, this.

[A]

Is a big but.

[B]

That traffic is still less than 2% of what we would have seen through organic. It's still tiny in terms of what is measurable that's coming from these platforms. So I would say that aeo, the AI optimization looks much more like brand than it does performance marketing.

[A]

Brand marketing is coming back into the circle. That's one thing that you need to understand. And a lot of the traditional brand metrics are part of this AEO mix. The other thing we haven't mentioned yet, but we should say is one of the main tools we used to get those results was this tool called Xfunnel, which helps you scale your AEO. And HubSpot just agreed to acquire Xfunnel about a week ago. So we're gonna bring that to all the HubSpot customers in the HubSpot community, because we found that to be incredibly valuable in how we're scaling our AO efforts. So I personally am super excited about that. The team is amazing, huge.

[B]

A big part of it is that tool. We were a customer and we were so blown away by the tool in terms of the results we got working with the team that we were fortunate enough that we could acquire them and make them part of the kind of HubSpot platform. And there's a couple of things that I can share with you that will be pretty useful in terms of you really want to like, you have to play in this space. And so it's kind of a must have. And I think one of the interesting things you can do is you can use AI to build an ICP ideal customer profile. We've shown how to do that on the podcast if you want to go get that tutorial. But once you have that ICP and AI can kind of build a real great representation of your customer, you can then ask AI, hey, what are all the ways that this customer would ask questions about my product or services? Because remember, it's much harder to understand how people are conversating about your product than searching via keywords. Searching via keywords in B2B, usually 3 to 5 keywords per product. Pretty easy to like, find those keywords. Having a conversation about a product or service. Well, how does my customer even talk? How do they ask questions about my product or service? So then you can get going and you can say, Well, I get 100 different ways that they can ask those questions. I upload them into X Funnel and then I get my baseline. What is my current share of voice. And the things that we have found that really help increase that share of voice are niche content, which is like an old school SEO thing. But the reason makes a ton of sense because in the old B2B world you have a product page optimized for three keywords because that's the only ways that people would search around your product or services. Now you need like a hundred variations of that product page because of the way people ask questions about that product or service. You need to have a small piece of content that the LLM can integrate into its answer every time that question is asked about your product or service. Niche content, very boring, very impactful. It's worked really well. Frequently asked question pages, all of these things have worked really well. Citations huge for unlocking increases in share of voice. And what's interesting about citations for AI is in the Google world it was much more democratized, right? Like you just get links from high authority sites and all of these sites kind of were agnostic of Google. What's more interesting about LLMs is they all have these kind of partnerships and depending upon who they're partnered with and what license agreements they've actually signed. Those sites are going to appear much more frequently in terms of the ones they're using to determine what their answers are. So when you think about your citation strategy, which isn't links, it's just that you are mentioned on these websites, you really want to understand what LLM you're optimizing for and then who their license agreements are with. Because doing it for ChatGPT is very different from doing it for Google Gemini because they actually have different license agreements and they're looking at different websites to form their answers or questions.

[A]

Yeah, so like, for example, we have a bunch of time that we now spend on Reddit because Reddit is a key source for ChatGPT, because ChatGPT and Reddit have a licensing deal for Reddit data. And so that means we now have to play much more in the Reddit community and it has impact. Okay. We got a lot more ground to cover though. I know that you are obsessed with search, and I think we gave AEO its proper due. But one other topic that I think is super important is data. And data is basically the foundation for how all personalization is going to work in, in the future. And we've seen some transformational results through data and personalization. And what we've actually really found is you need to figure out what data matters to your buyer and to that marketing tactic, and then you need to use that data to actually better target and better personalize the message. Right, right.

[B]

I think paid advertising is like one of the kind of areas that doesn't get enough coverage in terms of how transformative AI is being actually maybe more than search. We are an early user of Google's new AI Max tool.

[A]

Do you want to explain what that is and everything? Because it's kind of wild.

[B]

So AI Max is basically Google's new tool that if you give it the data, it does the targeting, the ad creation and everything for you. You basically have the data and you have the money. And so if you think about where we're going with AI, it's actually one of the more interesting examples because. Because they have basically taken a lot of what we used to do in the paid advertising role and used AI to do it better than we can do it. And I was an early tester of pmax actually about two years ago, which was their first push into this AI within the platform. And they consistently failed to outperform our best paid marketers at that time. And two years later, their AI Max product is actually drastically increased conversions on paid Advertising based upon like a B testing against us. They've been better at targeting, they've been better at creating the ads and converting that into customers. And that is like pretty interesting. I'm not sure if it's scary or.

[A]

Not, but you think it's a little scary.

[B]

And so what's our role? Our role is to make sure that they have the best data set to build the best targeting and in paid advertising ads on to convert more customers and try to optimize our budget. And all of the platforms are going in this direction. Google is ahead, but meta will go in the same direction. All of these platforms will go in the same direction where they will handle the targeting, they will handle the ad copy, and they will also at some point, I suspect, handle things like landing pages. And so what you will have to have is a great data set. And what we are really focused on is like better data, better data, better quality of data. And we kind of push that into the platforms.

[A]

I think data is one component, I think story and message and what you're trying to run that campaign against is going to be a critical marketing skill, more so than it's ever been. The other thing that I think is really important to understand is that I believe we as humans have a complete endless capacity to consume when something works really well or is really effective. And so when I look at the better AI personalization and targeting through ads, for example, right now we're limited to that manual work. So we're limited to the number of campaigns we can run. We're just going to run ten times more campaigns.

[B]

Exactly right.

[A]

And there's going to be a ton of human work involved in figuring out the basic segmentation and messaging. And then the AI is actually going to go test and iterate and manage the targeting and everything for us. But one of my big learnings from the last year is that the world of AI makes everything much more granular. Like you just said in the where you need more niche granular content in advertising, it's going to be you're going to run much more niche granular campaigns. In brand marketing, for example, like the model with brand marketing, you spend hundreds of thousands of dollars to create this big ad campaign. You get a couple of 60 second, 30 seconds and 15 second spots and you place them in an AI video created world. You're just going to have 100 different assets, 200 different assets, and you're going to place them in much more finely tuned ways. And so we're going to continue to get more granular and create much more marketing assets and campaigns as the scale happens is my belief. Do you agree with that?

[B]

I think we're moving from the kind of segment level era of marketing, like most B2B is done on these kind of segments and these hypothetical versions of your segments. And all segments in B2B marketing are basically firmographic tem demographic. You're like company of this size in this kind of geo. Whereas now AI allows us to move to like these micro audiences and do much more marketing at scale for smaller groups. And I think the perfect example of that is email. I remember when Kip and I were talking when I was still at Zapier and we were kind of talking about coming back to HubSpot to do AI and he was talking about some of the AI work and I was like, yeah, email is like everyone's integrating AI into email. It's probably like not that high of a ceiling. You're like, well we're seeing like 500% increases in conversions. And that email is like one of the best examples of a channel that AI has breathed new life into. Because since then we've actually seen phenomenal results with integrating AI across all of our email workflows and tailoring the content to that individual. And I think that has been fascinating to watch where you have this historical channel and AI has kind of like reimagined it.

[A]

It was like a 40 some year old channel. And we were I think resigned to the fact that like our click through rates were always going to be like 2 to 3% and it was just going to be what it was going to be. Open rates are going to be 30 to 40% if we were doing well. And now with AI you can get the engagement click through of like a human sent message which is orders of magnitude higher at the scale of traditional email marketing. And you've done a lot of work when it comes to prospecting and helping book sales meetings using AI. And this is one of the prime uses of email that I think any growing business is going to work on getting better at and scaling. So tell us what you've learned.

[B]

Yeah, so I would say the majority of companies I talk to are failing with AI for prospecting. And the reason they're failing with AI for prospecting is because the output is generic. So they add in some, you know, base level data and then AI personalizes it to the individual and it all sounds the same as every other message and when you look at prospect. And so I would divide prospecting into two parts. AI's ability to prospect into your inbound demands or something gets triggered. They sign up for a form, they do some action and the AI reaches out to see if they want to book time with a sales rep. That's inbound. Outbound is like pure cold. Not in your database, has shown no internal signal. And you kind of reach out to see if they want to book time with a sales rep. Now I'm going to cover inbound. Outbound can work and it does work if you have great data sources. It's much more saturated right now. I think inbound is a huge opportunity. Let's break prospecting into two parts. I think this is also interesting. If your company is marketing to small companies, then prospecting via email is actually a viable solution via AI. You can actually book all of the meetings via email only. And so you can automate that entirely because you don't need the human in the loop. When you're marketing to like larger companies really mid market, anywhere above, I think 25 employees, you start to get into the need to do email, call and social outreach to book a meet in. And we see about 6% of our meetings in mid market and corp are only booked via email only. Right. So you need these multi channel sequences. So the human has to be in the loop. And what you do there is AI is able to assist the person to have much more productivity, cover much more space. And it does that by creating the email for the person. And the person could decide that they want to send that or not, or edit it or not. It creates a call transcript and then it creates all the social outreach. And we've seen big results on both of these email only. We've seen about an increase in 35% of month on month meetings. But we've also started to see big, big increases in our sales team's ability to book meetings where AI is integrated into their prospecting workflows. So why do most companies fail at prospecting? It's because they are using really generic data and generic prospect prompts. I could just collapse your whole thing into the ability to get better data sets. We've seen 2, 3, 4, 500% increases in the conversion rate to meet in the more data we're actually integrating into that flow. And then prompt is your product. And I will give this one quick story. And so I have this thing where you have to demonstrate value in six weeks. So you do six week sprints and if you can't demonstrate the value in that, we move on. Because I don't want this kind of long period of time. Where you're trying to do something, it's not proven successful. If I had enforced that on the team doing this, we would have been unsuccessful. It took us six months to actually start to see increases in the AI workflows versus our traditional workflows. Six months. But then when we saw it and we really got the prompts working because we sat with the sales team and we iterated and we iterated and we iterated. We've just seen huge increases over the past eight to 10 months. And so there is a real learning experience so you don't give up too quickly. But the data and the prompts, I think, are two of the most important things. And customizing those prompts to sound like.

[A]

Your sales reps, I think there's one other really important thing. And Kieran, we've often, when we talk to people who listen to the show, they say that we're their kind of virtual co workers, virtual marketing therapists in ways. And we are each other's marketing therapists in a lot of ways. And so we need to have a little bit of a therapy intervention for just one minute here. And that intervention is. You're going to have to let go. I'm going to say it again. You're going to have to let go. If you want every word of an email to be perfect and you want to review every single word of an email, that email is not going to do as well. That is the fact of the future. And I know that's very controversial, but I believe it in my bones. It doesn't mean that you don't have to spend a lot of times making sure that you're constantly improving and annotating the quality of these emails that AI is writing and sending on your behalf. But if you want to approve everything and you can't afford to have any word out of place, you are going to really be at a disadvantage in the next era of email marketing.

[B]

Exactly. I think marketers are in email moving from, you know, copywriters to prompt engineers. And the prompt on the email, the other thing I would say is you break it into modular parts. So you have a prompt per subject line, you have a prompt for your intro prompt for the body copy prompt for the call to action, break it into multiple prompts. But that's the skill set. And the people who are really excellent here are not trying to like check every email, they're trying to check every prompt. And they are engineering. They are back engineering the best sales reps. So they take the best sales reps emails and they back out from there. And I think that is a big part of it. And the actual follow along here, which if anyone in HubSpot spends a couple of minutes with me, it's only like within that first two minutes I've talked about multimodal. And multimodal agents I think are one of the biggest investments to make in 2026.

[A]

Can you explain what a multimodal agent actually is for everybody?

[B]

So I think the future we live in is you have a dual purpose website. I think part of the purpose is to serve your AI agents with lots of niche content because that content is not needed by the user. Right. It's like one of the more interesting things. It's the first time we're starting to market the agents, not consumers. So we have all of this niche content on our website. It's for the agents. It's specifically for the agents. Which means that I would say when someone is kind of using AI, but 90% of the research is already done. So what's the purpose of the website? Historically we do the website to actually help you research our products and services. But now you know everything about them. I think it turns into your first kind of sales interaction. And so the historical text based chat I think converts into multimodal chat. And multimodal chat means that you have a real sales rep, an avatar, not a real one, actually an avatar that can do multimodal. So multimodal means that they can do visual, they can actually show you something within that chat. It does audio, so it can talk to you and it can do text based chat if you want. Now we've experimented with these. We've had AI avatars on our website to conversate and actually do a lightweight demo. It can actually bring you through a demo of HubSpot there and Dan and we've seen the engagement go through the roof.

[A]

How long? This is the thing that blows my mind. If somebody's coming and they're chatting with this avatar, they're having a conversation or they're typing in and getting some questions back. How long does a session like that last?

[B]

So the average one on the website is like 5 minutes. This is wild, unqualified traffic. The average one when they've actually shown some form of intent. So we have an avatar that helps when you ask to do a demo down market, the avatar will actually do that demo for you immediately if you want. The average engagement rate there is eight to nine minutes. Someone talked to the avatar for 10, two hours. Now, I did not want to look at that. Call transcript because I did not want to know what they were talking about. It was too long. I don't know what was going on there. But the engagement rate was through the roof. And that was actually quite surprising because I didn't know if consumers were ready to like get their information from an avatar, do the demo through an avatar, and when you look through all of the transcripts, they're asking real questions. This is not like someone just interested in talking to the avatar. They're asking about questions about the products, services and HubSpot. So you can imagine the kind of prospect and someone talks at the avatar and then we actually decipher intent from that conversation. And then the prospecting agent immediately follows up with an email asking, do you want to book time with a rep? That's what we are doing. And so multimodal also drastically changes sales because instead of the sales rep having to spend their time doing a lot of discovery and qualification, which is like, hey, here's a product, what's your budget? What do you want HubSpot for? What are the use cases we can solve? So in that first call, most sellers are actually doing discovery and qualification. It turns every seller into a closer because what we do is we capture that conversation, we run it through an LLM and then we provide the rep with the context of that conversation. So when they're doing their first call, they turn that call into a closing call because they've already done all of the discovery and qualification through the agent and I think it collapses the sales. You can get like increases in deal velocity and close deals much, much quicker.

[A]

I think in addition to like these agent experiences, which may seem a little science fiction to some, there are a couple of really important lessons that you went over. That part of your website is going to be for robots, not humans. That's a real big shift. And that the experience that somebody needs to have on your website is getting deeper. It used to be really wide because they were going to hunt and peck and do a bunch of research. Now it's getting deeper. Whether it's with an agent or with ChatGPT, whatever may have you like, the depth of information they need is much, much deeper. And as a marketer, we're going to have to have a lot more of information that might be private inside. Our company now is going to need to be public as part of the public facing experience.

[B]

So I would kind of describe that use case for next year. If you're kind of thinking through is this something we want to invest in is like your AI sdr. The ability to have an agent on your website, conversate with the person, show the product and then the ability for you to connect that into your email and prospect and book time with a rep. If you don't mind me asking one quick question for some participation because I'm kind of really interested in this. Who in the audience is thinking about that use case, investing in that use case or that's something that they actually, you know, want to actually do?

[A]

Who wants to use AI for qualifying marketing and sales next year? Raise your hand if you do. Okay. I would say it's about maybe 20% of the audience.

[B]

Okay.

[A]

Okay.

[B]

Okay, cool.

[A]

That's super helpful. Okay. Something else that I know is top of mind on folks mind Kieran is short form video and content. And one of the things that we've seen is AI video and image models have gotten very good and is leading to a lot of prevalence. You've seen, you talked about Sora 2 earlier. Short for video is becoming more important in B2B, becoming a higher influence channel in B2B. If everybody out here has been doing more traditional web content and less short form content, how should they be thinking about scaling short form content?

[B]

There was a great stat I read recently, where in B2B the primary way people want to learn about products and services is short form content. And I was even surprised by that because again, that's a B2C trend that we're starting to see. In B2B it makes sense because AI, I think accelerates our attention collapse into small incremental parts where we can't like watchful videos. We want to watch short form video. I think short form video is an incredible tactic to also not just acquire attention, but market your products and services. The new virality today for AI startups is the product demo, right? When you actually see how fast they're growing, it's because they're able to show these magic moments in short form video and those things are really taken off. The other thing we talked about backstage when we kind of went loosely through this is if you can master short form video, you can master every single type of content. It's actually one of the harder things to do is make really interesting content, engage with someone in a short period of time. But I do think again, if you want to be ahead of the curve, what happens in B2C happens in B2B. TikTok shops or E Comm store is built off selling through short form content. Humans are humans, whether they're buying consumer products for themselves or buying products for their company. And so this one here, I think is a real investment to make that you can be ahead of the curve. Because I don't think many B2B companies have mastered this and for good reason. It's very, very hard. But I do think this is a similar trend that we talked about earlier, where you have creators becoming a core part of your product or service. The one tip I would give is we talked about this. Take 5% of your paid advertising budget and give it to creators to create content for you. At Zapier, our biggest channel through paid was paid advertising through short form video that creators created for us. It was actually the way we acquired most of our demand on the paid advertising side. And so starting to integrate this into your paid is probably the fastest way to get started.

[A]

Yeah, find a couple of people in your industry who have YouTube channel, TikTok channel, big Instagram account and go partner with them and do some sponsor content and see how that compares to some of your programmatic paid is what we're saying. The other way to think about this as retrospectively my life, maybe I was destined to be a marketer when I was like 7 years old and just would watch infomercials all the time. And TikTok and short form video is like a modern day infomercial but in like 20, 30, 40 seconds. And the great thing is if you can tell a great story in 30 seconds, then you can tell it in whatever length you need to. And that's why it's becoming a really foundational format for storytelling.

[B]

And I think these tools are democratizing. I'm actually working on a short form documentary like five minute style using VO3 one, their new model. And they actually have released tools now that allow you to stitch together 8 second clips much more cleverly. And again, what do you need to be good at? You need to be good at prompting, right? If I can prompt correctly, I can create pretty good video assets. I'll give you a quick tip there. One of the ways that you can get really good at VO3 prompting very, very fast is actually have Perplexity or one of these assistants create an onboarding doc. So just say, hey, I want an onboarding doc for someone that I'm going to teach how to use VO3 and make them really great at prompting. It will create you an onboarding doc and then I give that to a custom GPT as the document. And now you have an assistant that can actually prompt for VO3 for you and so you can shortcut a lot of these things. But I think these video tools are a real must have tool for most marketers to learn. One of the things we'll end with or the way that I see marketing really drastically changing is marketing has always been art or science, right? Like there's like a scientific part to it, there's an art part to it. And I think that what AI does is it pushes the skill sets to each end of the spectrum where you have to be either deeply engineer led to understand really how to integrate AI across all your marketing or have incredible taste. And I think you're big on taste being a central part of like what separates good marketing.

[A]

Taste is not this arbitrary thing that people have or have not taste means. Have I seen tens and hundreds of thousands of this thing and can I then arbitrate what good looks like and what bad looks like? And then can I work with AI to build the good version over and over again? And I think that's where there's a bunch of opportunity. The fact that you're building a five minute AI documentary, it's going to makes you the least cool person I know and that's okay. Okay. Wouldn't be a show if we weren't giving you some actionable things coming to the end of 2025 and we want to give you a few things that you should 100% put in your 2026 marketing strategy, things that you should be doing. Kieran, on this list we talked a lot about a year. You obviously need to build an AEO playbook. We've got that in here. What are the other like one or two that are must do on this five for you?

[B]

I think this is a technical one. I know it's hard. I spent hours and hours with customers going through this one as well. I think having a great data layer and being able to like easily integrate data across all of these AI functions and features and applications you're using is a must have. The better the data, the better the context, the better your output. And that differentiated input also means like how can you differentiate your data and context from everyone else if you have some unique data sources, internal, external, that's going to really separate you from the pack. I think every SaaS brand's content strategy should look more like creators than it does brands. I think that's a hard shift to make, but it's the only way to stay relevant in terms of what's happening.

[A]

And then you have to give up control, let the AI personalization get into your messaging and email strategy to improve those conversion rates.

[B]

Can I finish with one question?

[A]

Of course.

[B]

Who is bullish about marketing next year because of AI?

[A]

Who's excited?

[B]

All right. Who's bearish?

[A]

Who's scared?

[B]

Who's.

[A]

Who's concerned? All right.

[B]

I would say I'm both.

[A]

That's classic. Look, it's very reasonable to be scared, but I'm an optimist, and I've never had as much fun doing marketing as I have had in the last year. And I mean that very genuinely. I am having the best time ever, and I can't imagine doing anything other than marketing right now. If you have not subscribed to Marketing Instagram, please go to YouTube and hit subscribe on that channel. We would really appreciate you, and thank you all so much for having us today.

[B]

Appreciate you.