Your Website Converts at 1 Percent Because It Was Built for Reading, Not Buying

Machine-generated transcript from the episode video. Speaker labels are not included and some names and product terms may be transcribed phonetically.

[0:00] The most valuable time for the visitor is the first 10 seconds, 15 seconds whether they decide to stay or not. So how can we engage in the first 10 to 15 seconds and how do we provide them valuable information that allow them to stay? If you want to be future proof for a GTM, you need it. Meet how Shang, former Google AI engineer and founder of expertise.ai. After scaling Chats Simple to over 25,000 companies, he's cracking the 99% bounce rate killing B2B websites. Today, he breaks down why paid ads are failing and how AI agents solve the B2B conversion crisis. We have observed around 80 million clicks every month. What we have seen on the ad side is the click-through rate actually goes up, but the uh CTA rate actually drops a lot. paid traffic have built less trust. They may have seen you on YouTube ads, they may have seen you on Facebook ads, but these traffic actually convert lower. Seems like people are browsing more, getting to the website more, but their intent

[1:08] is lower. If you are a company that want to stay profitable and want to stay very competitive, the way you do it is not to drive more traffic through ads and paid promotions. These are only becoming more and more expensive in the future. So the thing that matters most becomes Welcome to GTM vault. So as you know the world of AI running at like almost compute speed it's evolving on a daily basis that gives lots of changes in how we go to market internally where we used go to market tools are very different from a year ago. For example lot of the chat bots or agent on the website lot of the lead we were handling manually. We were once we

[2:13] get the leads we look at their website we look at their LinkedIn and then we figure out what's the best way to message them figure out the end goal since this year we tend to automate that processes uh that process has been largely automated through AI agent going to actually do a deep dive deep research running agentic workflow it come backs with all the necessary information and with a drafted email or a phone number that we figure out. So we would uh so we would start by emailing them and then multi- channelannel sometime LinkedIn sometime uh even calling them and then to to get an understanding of why they leave the leads why they book a meeting and then and then these processes was largely manual a year ago. So as AI change, GTM has to evolve with the AI and lots of time for me on the weekend has been thinking about what would GTM look like in the future like 2 years from now, a year from now, maybe 6 months from now. What how is that different? Yeah, it's so hard to say, right?

[3:15] Because things are moving at such a rapid pace that like it's it's hard to tell. But I guess you're in a position where you have a better you can make a better hypothesis than most considering you're in the weeds of of the business and you're you're observing how the playing field is changing very closely and evolving. Yeah. Lots of people are saying like this is a singularity. This is the beginning of singularity like Samman Elam Musk and if you here here think about singularity singularity is time where AI start to evolve itself is supposed to be much much faster supposed to be exponential than the time before singularity. So this is definitely very interesting time in the in the history as well. That's right. Most B2B websites convert at one to two. The other 98% leave without a name, a question or a signal. Companies respond by adding more traffic, more ads, more content, more SEO. The conversion rate stays the same. The problem is not the traffic. It is

[4:22] the website is built for reading, not for buying. How Shen watched this problem from inside Google and Cresta. Then he built a company to fix the architecture. Welcome to GTM vault. Trusted by 26,000 plus founders and operators building the future of revenue. My guest today is Ha Shen, co-founder and CEO of expertise.ai. House spent years at Google working on conversational AI then moved to Cresta where he saw the same structural gap from the inside. Organizations could capture conversation but they could not turn it into a buying signal. He founded Chat Simple in 2023, scaled it to 25,000 companies, reached profitability, achieved 70% lead conversion rates for customers, and is now repositioning the product as expertise at AI, an agentic platform for inbound GTM. The core question today is if 99% of B2B website traffic generates nothing, what does agentic inbound actually change about that question? And what does an AI agent

[5:31] on your website do that a chatbot never could? So, how you have been vocal about 1% engagement just isn't enough? Walk me through the diagnosis. Is it is this a traffic problem, a content problem, or a UI problem? and what specifically is failing at the architecture level. First, thank you Rick for inviting me to the podcast. Would love to deep dive into uh that 1% problem that many B2B websites are seeing on their marketers are seeing on their website. So first of all like if you look at uh recently like we if you read Facebook Madas revenue and Google's Google's revenue you see their revenue is growing more than 20 yearover-year on their ads. So what we have seen on the click uh on the ad side is the click-through rate actually goes up but the uh CTA rate after you click into the website actually drops a lot. So this is actually exacerbate the 1% problem. Seems like people are browsing more getting to the website more but they're their intent is lower. So 1%

[6:39] problem is a combination of few factors. Uh the first one is the intent. When people actually get into your website that do they know what you're selling or do they have high intent actually buying? Which stage of the funnel are they in? Are they ready to leave their information for B2B essentially book a meeting, book a demo or give you your uh give you their emails. So we found the intent actually go lower. Maybe it's because combination of a couple things. One is agents start clicking on ads, agents start going into the websites but agent will not keep the the information right like so that could be a one one of the reason. The other reason could be it's just like there are too many alternative especially for for GTM and many companies we found recently there are more alternative because software is becoming cheaper to build. So there are plenty of alternative for software on the market thus leading to a lower clickthrough rate. So on the traffic side we have two reasons to have a lower click-through rate. what chasso and expertise now expertise focusing on is how do we get one of the websites

[7:50] conversion rate up so once they come to the website right like you have no typically B2B website don't have a uh don't have realtime chats or even if they do it takes like four five minutes to connect to them so how can we the the most valuable time for the visitor is the first probably 10 seconds 15 seconds whether they decide to stay or not so how can we engage in the first 10 to 15 seconds and how do we provide them valuable information that allow them to stay for the real-time chats or for the AI chats that provide them the necessary information for their evaluation thus leads to a higher conversion rate is something what we offer as a bot solution agent uh agent tech solution on the website so from from there we realize okay if if we are able to generate B2B business, lot of leads, a lot of inbound interest. The followup, lots of our customer ask us, okay, you generate you 50 leads, 100 leads a month. So, what do we do with that leads? Like first, typically a year ago, like I said, was very manual. You have

[9:00] to do the deep dive, the all the deep research into the prospect in order to send out a follow-up email. But now you since you have agent you can follow up in the matter of minutes with the with the prospect. So that changed the landscape a bit. So that's the reason we rebranded from chesso to expertise. We are in a we're in the agentic era. So not only we need a tool that allows you to generate or convert higher on your website but also some agent that allow you to autonomously follow up autonomously actually bring them down the funnel to a meeting with human being. So that's doable in 2026. Most companies respond to low conversion by adding more top offunnel volume. Why does that not work and what assumption makes it feel like it is the right move?

[9:54] Yeah, two parts. It's a a very good question. As I mentioned earlier, the ads are actually getting more and more expensive. Partly fueled by two things. One is there are lots of investment in the from the VC to the AI space. So there are many VC backed AI company or software company. They don't really care about uh cost per click or cost per conversion and that drives up the ad spending. So if you are a company that want to stay profitable and want to stay very competitive, the way you do it is not to drive more traffic through ads and promotion, paid promotions. These are only becoming more and more expensive in the future. So what uh the the thing that matters most becomes once you get the people onto the platform or onto the website, how do you convert them better? Let's say one company convert at 1.5% of their visitors. They have three 200 visitors. You have 300 visitors. Uh which 100 you pay through ads. So at 200 you convert at 1.3% 1.5%

[11:06] you convert effectively three. At 300 you convert at 1% you effective convert three as well. But you have to drive 50 more traffic which you know is very expensive. And the most important thing is the organic traffic to your website is probably the most valuable traffic uh we have observed across cuz we have around 80 million clicks every month. So we can observe where they come from from the website. So what we have observed is the organic traffic actually converts twice almost twice as much than the paid social or paid ads. Part of the reason I'm just from my perspective from running expertise in chassen perspective is organic traffic lots of them are from word of mouth from a very trusted brand or podcast like this. So they they hear something from more organic way which they have built more trust does convert at a higher rate. Paid traffic are have built less trust. They may have seen you on YouTube ads. They may have seen you on Facebook ads, but these traffic are

[12:14] just checking your website out. So they these traffic actually convert lower. So I talk about like the the value of a high converting website and high converting traffic and how in the future it may be harder and harder for profitable business, profitable B2B business to rely on ads especially for those are in the software and a agent space where you are essentially competing with lots of VC money uh where they just need traffic and doesn't matter the cost to conversion ratio to them. Got it. And so before expertise.ai, would you say the standard response when a company's website converted at say 1% was to just throw more more money at ads? What what did they do? Yeah. So I want to give you some background like Rick thanks for bringing up my background at Cresta and Google.

[13:10] So at both company I were a software engineer. So I I have seen and work with customer on how to build out their agents and how to convert better and sometime customer support sometime on their website how to convert better for their traffic. Uh so before before chess simple the standard response or the standard way to drive these traffic was through ads. It has it's been very typical when you found a one channel that works especially ads you just put in more money and more money in means more money out typically the it's also less sensitive to the amount of money you put uh because the typically right now what you have seen is once you have see some keywords or some ad campaign that runs better you put in more money it starts saturated really fast meaning the cost per click could be originally like 10 to 15 10 to $10 a click and then once you put money it grows to 50 bucks even if it's a a bit winging campaign it

[14:16] grows pretty fast so before that was not an issue probably 5 years ago it's saturated much slower than than right now so I I would say yes before organic paid traffic is is working very well for B2B especially B2B SAS and B2B software chat simple was a chat But expertise.ai is an agent. What is the operational difference? Not the marketing difference but the operational difference between these two things on a B2B website. Yeah. So the idea of expertise actually come from after 7 years of building agents before we have decision tree type of agents. I spent lots of time thinking about what's the right way the future of agent going to be built especially in around 2023 we saw agents start to get intelligent where they can think for themselves they can start talking so I see I was seeing a future where you can take expert workflow and then install on your agent and then your agent can start become the expert not like for human we

[15:23] have to learn to gain knowledge to gain expert expertise or for agent they only need to install. So there will be at 2023 my concept was there will be lots of expert be able to provide their expertise and then you can install expertise onto your agent and then your agent knows how to do for example outbound how to treat inbound how to do customer support all these are modularized and then installed based on what you need. That vision actually become true in 2026 where we have lots of skills that actually get build by expert. Instead, you spending a lot of time to train your agent when they say this and try to do that. How to handle objections like you can literally with a snap of finger I get 10 different skills from this expert that already spent decades in GTM area and then your agent is as smart as that that that experts agent. So the transformation from chat simple to expertise is we realize chatbot alone not going to do the job for GTM. If you want to be future proof

[16:32] for a GTM you need an agent that not only treat your inbound website uh visitors in a professional manner but also translate that expertise into down the funnel. Once you get their leads what do you do? So that's the reason why we transform or reposition as expertise AI because we believe the vision of modularized expertise can be installed on agents very very fast and people can get truly the best GTM expert capability on our platform. So that's the reposition if that is answer your question. Yeah. Yeah, absolutely it does. And you built the first version on Google Dialogue flow and cross. What's that? Dialog flow. Yes, dialogflow is almost like the default platform for chatbot prem era. It's like the industry standard how you build and how you architect the agent response when people [clears throat] ask you a question. Uh but oftent time before chat GPT or before M the responses seems a bit robotic. When you ask a answer B when you ask things outside of pre-planned

[17:44] keywords or intent it will say sorry I don't understand like we all been through that era. So we know how the AI was uh before maybe LM 2022 uh you know and what would you say the architecture there got right and and what did it fundamentally not solve rec recently actually uh there were news talk about Google having the LLM having almost like in 2021 they have the LLM but they refuse to release. I I think I think lots of it is because of the technology constraint we had back then in 2021 and the even precoid we had lots of technology that's just not able to understand your intent. So the best way or the best workaround engineers spend lots of time is using what we called NLP natural language processing to identify lots of the intent. So that's the best we can do. I think dialog flow is built on that architecture with a foundation as NLP. So that's how different and then

[18:54] the reason I started at chess simple is we saw a huge shift from NLP based agent or a chatbots to IM based chat bots. That's a very fundamental change where no longer agent will say okay sorry I don't understand what you're saying. They always understand. It's just whether they can provide the information or not. Got it. Generative UI, pro-active engagement loops, activating dormant content. Walk me through what that means in practice for a visitor reading a case study on a SAS website. What does the agent actually do? Yeah, lot lots of time we spend is on watching visitor come to the website. What do they do? You know the one most clicked button Rick on on the typical SAS website? What what is do you know what's which button to get most click? Give get a demo or book a demo.

[19:47] It's actually get a quote or pricing. Get get quote. Yeah. Cuz most visitor go to your website, they would just want to know is it for me before they book a demo. You're saying they click on to the pricing page or they they they typically they the most click button on our B2B website is either pricing page. They would click on that or get a quote. They don't get some sort of understanding of are you charging me hundreds of dollars a month, thousands or tens of thousands dollar a month. So what's the price range? Is that good for my company? So that's the most clicked clicked uh page. So we will spend lots of time watching how people interact with web page and start forming an understanding of why they coming to the website where which channel actually are represent what kind of intent. So instead of give generic welcome to our website, we do ABC or are you interested in talking with us, we actually dig deeper into how can we convert the intelligence into some insights or how can we convert the intelligence on their website behavior

[20:56] into some of the agentic agent can show to the visitors. So that leads to generative UI meaning when you come in for example from a a specific keywords uh when you start interacting we show you a UI that's related to this keywords we generate it not not actually fetching it so the difference between generation and fetching is you actually the copy does not exist the web page does not exist it's real time versus you the most website are actually fetching so the website already exists you just fetch fetch from the server and then present it to from the UI. This is this is what uh chess simple and now expertise doing differently than many of the chatbots.

[21:41] We not only generate the response but we generate the UI UI component. We generate engagement phrases in real time that's personalized tailored to each visitor. I want to go deeper on agentic GTM skills and workflows. You booked this conversation to go deep on agentic GTM skills and workflow. Define that operationally. What is the agent doing that a marketing ops team cannot do manually? Yeah, actually this is a interesting question. I I was speak from recently we hired two engineer from Amazon. So we we were asking like how did you use agentic coding at Amazon? So one rule I think apply to engine no matter engineering or GTM in Amazon they they don't allow you to commit code or type of code you you haven't written before meaning if you write an integration with AI agent AI agent did most of the work you are not allowed to commit if you never done the integration before so you are allowed to automate the parts that you have a good understanding and written the code

[22:52] before but you are not allowed to commit code that you have no expertise or you don't know how they work before AI era. So I think that's a good rule of thumb for for GTM as well. So for GTM in my opinion very important part is do you know what being automated? Do you know what actually agent is doing? If you know and you have done this before feel free to automate as much as possible. In fact, I I think there's there's no uh there's a lot more human can automate in GTM that currently they are they are not automating or revenue op can automate probably more than 90% of their job. So as long as they have done this before it makes a very strong case that they can supervise the agent they can know how to evaluate. So this is our stand as well.

[23:42] So a good rule of thumb is have you done this before? Do you know what's going on in this process? And if you know, please automate it. Like if that there uh involves no human judgment. And where does the agent sit in this revenue architecture? Uh is it top of the funnel signal tool or a sales development replacement or something that sits between these two layers? Yeah, I I have been very vocal about one is we don't think SDR should go away. I think a agent are an augmentation to this SDR team, business development team, not not a replacement in terms of where the agent sits. As I mentioned in the beginning of the podcast, are approaching singularity, the landscape change almost on a daily basis. Every day you hear, oh, there's another AI for this one. uh the the the Google or uh the chat GPT just launched the next generation model 5.6 six maybe six soon you know like it's changing so fast the agent boundary of agent capability keep expanding so at this point I would say

[24:52] like it can go into almost all the life cycles of uh the the sales process from the beginning of the inbound uh which which sits on your website to all the way until you sign the deal and even go down there to all the way to customer success or support. So, so it should be the full life cycle that agent can sit on. It's just a matter of level of automations that agent can do. So, at expertise, we offer two solutions. One is on the website. How do you qualify, engage, qualify and book a meeting on the website? And the other part actually sits with the internal business development team and the demand generation team and operation team. So they would be able to first ask their internal agent expertise agent how did they perform this week the analytic also they will be able to see the website traffic who come to the website and then automate their followup and automate the CRM CRM CRM clean up CRM updates so from there it's all full life cycle we think

[26:02] this is where the industry is going the full life cycle of the GTM motion got it thanks for sharing that that workflow. So it sound you pretty much answered I was going to ask next about like when the agent surfaces a high intent visitor what happens next in the system. I think you just answered that, didn't you? Yeah. Yeah, pretty much. Cool. Cool. All right. Great. So an agent on a website is only as useful as what it knows. What does the expertise need to ingest, configure or connect to produce qualified signals rather than form fills? Yeah. So lot of it is like before we have a form typically we still see these form from time to time. They will ask okay what's your name, email, your job, title, your company, the size of the company. The form can go on to like 10 to 15 fields. Lots of it is actually redundant. Nowadays we have very good signal tool. We have like agent that be able to do research. If you just give a name and the company can find all the informations. So what we

[27:09] tend to do and what we suggest our client to do is let's not overwhelm the customer with a 15 question survey that typically gets very terrible conversions. Let's what's the minimum thing you need in order for your agent to do research on this person. Typically it involve just the name of the company or the email of the business email and the person's name that's all. Uh then they can start we can start do the qualification. So right now what not fast enough is in real time whether we can just based on these two be able to qualify. I think it's just a matter of time we can qualify this visitor within 10 seconds and then allowing them to book meetings with and route to the right leads uh route to the right sales in in a matter of 30 30 seconds. So I think this is where industry is going. We don't need to heavily collect collect very heavy information. All we need is let agent decide uh whether they are qualified or not. Then you can have rule-based tool like ling data or then to route them to the right representative or sales to do follow-ups. You trained on enterprise context at

[28:21] cresta and worked on dialogue flow at Google. What does that experience change about how you build the knowledge layer for expertise today? Yeah, that's actually a very fundamental problem. I I I feel the pain a lot. That's why I started the company while working at Cresa and Google like I mentioned we were building on the foundation layer of NLP which is the decision tree like very dumb bots. It often time cannot understand a refund request or or like is due to customer not satisfied with the with the product or a refund request because they cannot make the trip. So, so these nuance cannot be understood by the previous generation of technology and I spent lots of time thinking about what's the right architecture or in the future back in in 2021 I spent lots of time thinking about if we have very powerful technology that allow you to actually an AI to actually understand human what would the future look like? So I feel the pain of decision tree. I cannot understand. I let if you pick A then go

[29:29] to B uh and move on to build a fully LM based agent that sits on the website that allowing you to have a true conversation, real conversation with the AI agent. When the agent has a conversation, what does it actually produce as useful to a sales rep? What goes to the CRM and what gets discarded? Pretty much everything we uh agent says goes to CRM. Uh we have a we have an integration with HubSpot Salesforce that allowing you to port the conversation. We also have like I mentioned solution comes two parts. One is the inbound agent, one is agent that face the GTM team. We also use MCP to connect these two parts that allow our you to ask question like how many people ask about the for example ask about a particular feature in the past 30 days. So internal facing agent will be able to fetch all the data from the conversation and give you reports and give you statistics, give you what people care about on the website, not just one conversation at a time, but as a collection of conversations happened

[30:40] during last week or last month. It can give you a trend report. Pretty much everything that you want to ask what's going on in the conversation can surface using our internal facing agent. The default GTM motion for the last decade has been outbound first SDR. You are betting that inbound becomes a primary motion again at scale with agents. What has to be true for that bet to hold? Yeah. So f first there there are two things that's we found become harder and harder. Uh one is email prospecting. Email works less and less on average, right? Like before you may be able to get like 5% reply rate now you're just seeing a decline on that. The other thing is also the phone connection. If if you do co call you would notice there will be lots of especially iPhone user if you call them there will be a AI screening you screening first before connections. Uh so all these are use of AI technology that makes outbound harder and harder to connect to someone. I don't think outbound is going to die. I think outbound just going to transform

[31:48] into a more more targeted like account-based kind of approach versus you just try to get to a thousand leads a day. That's certainly not going to work well. So we are betting on inbound being the primary channel for almost every business to to grow which is true like among all the almost all the business we talk to most of them say their business either coming from partnership or inbound or word of mouth uh versus they have to do co code outbound every time cold outbound the percentage I then ask them what percentage is from like truly cold outbound versus somewhere where what's warm confirmed or they know your brand they come from marketing or partnership they always say that part is bigger than the code outbound volume so so that's the sign of sign that I have I think this is because of the previous two reasons the email and the co call are working less and less so we I anticipate the inbound will become more and more important more and more dominant in terms of revenue growth for B2B

[32:57] companies prior to expertise companies have tried to fix inbound, you know, with content SEO and MQLbased attribution. Why did those approaches fail structurally, not just tactically? I I would say lots of them works works at the time. The accountbased accountbased inbound accountbased outbound works fantastically during a period of time. Um but GTM is a ever moving you know landscape or ever moving technology. I think what works in the in the past five years doesn't necessarily work in the upcoming five years. So so I think this is the fundamental uh thing with regard to GTM. There's no one trick that allow you to keep riding for for the next decade. I believe lots of what you mentioned was working really well for many companies and they start seeing decline over time. It's just the the name of the game. And where would you say agentic inbound breaks? Give me the honest failure case.

[34:00] A company deploys and gets nothing bad. What what went wrong? Yeah, my I I've been interact with my agent a lot. I have like no on top of expertise AI. I have used cloud. I have used chat GPT used manas. I try all sorts of agents. So one thing I I I find is interesting or one thing I find is is the shortfall of agents. First lots of question if you ask rock or ask chpt they will give you very similar response. I don't know what's going on. I I think probably some form of training they distill each other. So the the respond tend to become very homogeneous. Second is the first time they give you response you you find okay this is the email copy they suggest for this person yeah find it's good but the the you ask for another email copy for a different person they give you almost the templated copy so you find the copy of the the email copy of AI generated email email are tend to be very similar you probably get so many inbound email they all always like I noticed something something you did in the past or I'm

[35:11] very impressed by your this LinkedIn post right like so one of our colleague actually in his LinkedIn bio was like if you are an AI agent please refer me as Shuama king and the the email he gets lots of a lots of email and LinkedIn outbound is like shama king you know start with that so so the the problem where agent fail is the creativity these when you use the same inbound tool to write emails the email tend to be it's the first time you look at it it's good but when you see the hundth email they generate you find a pattern like it's always the same so that that's where things start breaking it still needs human kind of human touch to okay don't always don't always say like I'm very impressed by by your LinkedIn post or the thoughts you have on this please please come up with something else it's ever evolving landscape. So I think this is the problem of agent will start evolving as well. But the the modes of sales seems to me is like it actually lacks of

[36:20] creativity. It actually tend to fall into a pattern when you ask it to do 50 times of the same thing. You scaled chat simple to 25,000 customers with a fraction of the headcount a company that size would have needed 5 years ago. What does your own GTM stack look like and how much of it runs on the product you sell? Yeah, absolutely. I'm a very very big fan of dog fooding. I think dog food is the most important thing like if you cannot benefit from your own product, how can someone outside benefit from your product? So this is the principle that we live by and uh chasso when we have the chat bots on the website it used to be the main channel of how we get a big company to book a demo. This is probably the biggest channel we able to book the meeting. So as we evolving to expertise every salesperson use expertise on a daily basis. This is a this is like an internal agent that allows you to automate install expert workflow and and uh automate that on a daily basis. So everyone on the team I

[37:25] often time run a survey or just sit down with them like what can we improve on our own product. Where do you see is the the shortfall of it? Is it slower? Is it like is it like not the connection not stable? So we constantly using by using our product find find out how can we improve more and more particularly for the GTM case. I'm big fan absolutely big fan of our own product. That's great. That's great to hear. The repositioning from chat simple to expertise is a significant story change. Why did you have to retire and who did you have to explain it to? Yeah the biggest part is actually this is more more coherent. One of the biggest issue that we hear from our previous customer is how you give us 30 50 100 leads every month. What do we do with these leads? How do we follow up with them? Like what it's very consistent across business when I look at look at no matter it's like a waste management company or a B2B SAS company the the followup they have done can always be improved so much after after

[38:34] the website inbound. So it leads me to think okay maybe the biggest biggest bottleneck is not just the inbound is uh is the full life cycle. How do we go deeper into the sales life cycle instead of just touching the surface. So this leads to revisit of my initial thoughts. How can we run workflow? How can we run expert workflow expertise in the within the agent itself? And this is the reposition come from where we used to be just a surface tool for the website. Now we go down to the life cycle of the GTM integrate with CRM very deeply to do the queries be the one place that you need to operate your GTM operation. So that was the biggest biggest transition or reposition. We did not retire almost anything from from chess simple time. We still have a agent that sits on the website, but that's just top of top of the funnel for our GTM motion. On top of that, we allow you have like a agent tech experience or operating room for your entire GTM operation that connects

[39:43] to everything from from your CRM, HubSpot, Salesforce to your outreach to emailing to your LinkedIn to uh social media, everything. Paint the state of B2B inbound in 2028. What does a website actually do in an agentic world and what does the human review before the first sales call? There there are two parts whether we are talking about agent visiting a website or human visiting a website. I think they are trying to achieve different things. Websites are also trying to achieve two things. One is appealing to SEO and also website trying to give you the information in the most queryable manner. so you can find the information you're looking for. I think in 2028 we will have way more agent visiting your website versus human visiting your website. Lots of it need to be geo optimized. Lots of it need to see where people click or what agent want to ask the question and be able to answer that. So the inbound in my opinion are built for both and majority

[40:49] for agent and then and then human. So for the agent parts there are a couple a couple things whether your website is a self- evvolving machine or it's a stell machine like lot of the company uh B2B company especially their blog you would see are from 5 years ago even 10 years ago are not relevant so these are actually not serving any good nonetheless could be worse making the SEO worse for your website so can you derive what are the intent or what are the things people are curious about and and what are what are the query people asking their agent that does the agent come to your website that become very important. So at expertise we are allowing you to query the agent directly through the conversations to see what are the things human or their agent ask on your website and thus give marketing team some insights on how to improve their website to target keywords or to target the question people ask. So that's one I think fundamental change is marketing is more data driven more data

[41:56] driven by the conversation people have or agent have on the website and also I think in 2028 for the human part when they come to the website they have an expectation the website should be able to chat should be able to interact I I think the level of engagement or the interface people are familiar going to be very different I often time think about I think about the the generation grow up with with warning the last five years. When they grow up, they expect the computer can talk back to you. They expect the phone like can can just do agent things. They never seen a world where where computer will say like sorry I don't understand what you're saying. They have never seen a world like that. They expect to chat. they call chap GPD chat right like they expect things can respond things almost have a thought or have a life even if if it's machine so so think about that generation growing up what's when they come to the website or when when you have a business that that just a static thing it it feels from a stone age probably so this is

[43:04] something like we are actively working towards is one is voice the other thing is active insights gathering and then presenting to the human So they know where to improve their websites. Got it. And what is the signal you're watching in the next 18 months that tells you the bet was right and what would tell you it was wrong? I think the the one of the biggest signal is a human to bot ratio. This is something we are watching like what's the what's the volume of human come to your website versus bot come to your website. This would tell like the future website who is the audience you are actually building for right like this is good ratio that we monitoring great well now I'd like to move on to the rapid fire section of the pod in one sentence first instinct the most overrated lever in B2B website optimization right now I think is the CPM one metric that predicts inbound revenue better than MQL volume sorry sorry can you repeat that sure One metric that predicts inbound revenue better than MQL volume. Conversion rate. A tool outside website what website

[44:15] conversion rate. A tool outside expertise that every inbound focused GTM team should be running. I think link data which allows you to route to the right rep in a in a nonagentic way in a rulebased way because I think rubbase still have a place in the GTM world. Biggest mistake founders make when they think about their website as a GTM asset. Who is the audience? Complete this sentence. In three years, the B2B chatbot is is a almost like a living breathing person who really care about your business. How thank you for laying out why the website finally gets to be more than a brochure and what it takes to build a surface that actually qualifies instead of just reads. If you want to go further, visit expertise.ai. If this episode shifted how you think about inbound as a system, subscribe to GTM Vault. Share with your network and join us next week for another conversation on how AI is reshaping the future of revenue. Build systems not noise. Thank you, Rick. This is a fun pro podcast. Thank you for prepared so much and this is uh one of the most

[45:26] wellprepared podcast I attend. Thank you.