The Buyer Answer Gap: Where Static Sites Fail and Autonomous Buyer Agents Step In

Julia Nimchinski:
And now, we’re joined by… yeah. We’re joined by Richard Dumas, co-founder and CMO of SlateCX, and today Richard’s gonna present the buyer answer gap. Richard, welcome, super excited for this, and how have you been?
Richard Dumas:
I’m well, how are you, Julie? Thanks for, thanks for having me.
Julia Nimchinski:
I’m excited to dive in, yeah.
Richard Dumas:
It’s so crazy.
Julia Nimchinski:
Amazing numbers.
Richard Dumas:
Lots of exciting content today.
Julia Nimchinski:
Yeah.
Richard Dumas:
So, why don’t we, why don’t we just jump in, and I’ll start by sharing a few slides with you. Start here… Okay, can you see that now?
Julia Nimchinski:
Perfect, yeah.
Richard Dumas:
Okay, so, for the audience members who don’t know me, I’m Richard Dumas, I’m the co-founder of SlateCX. I was previously the CMO of a company called Inference Solutions. Where we build AI agents for customer service.
Inference was sold to a company called Five9, a large publicly traded contact center company, and powers thousands of different customer service organizations from some of the largest contact centers in the world, like CBS and FedEx.
So last year we founded SlateCX, and with the goal of building AI Agents and Agentic marketing, instead of for customer service. So, I’m here today to talk to you a little bit about a study that we just did, and it’s called the Buyer Answer Gap.
And the Buyer Answer Gap is the gap between the kinds of questions that your buyers ask and need answers for in order to move a deal forward, and what they can actually get on most websites, or even through most chatbots.
So I think everybody in the room will… will agree with this, that B2B buyer behavior has completely changed, that they want to do their own research, they are doing their own research. And they have the tools at their disposal to do that research before ever talking to sales, and we saw that in the last session as well.
And the real problem is that the software and the tools that sellers have to use really hasn’t kept up with this change in the market.
So we know all the statistics, you know, Gartner says 80% of the B2B buying purchase now happens, without or before a vendor contact, and Forester will tell you that 4 out of 5 buyers use AI tools to research vendors before talking to sales. So we all know this is happening.
So we built a study, and what we did is we created a set of questions, and we wanted to understand, well, gee, what kind of questions do buyers really ask in order to move a deal forward? And in order to come up with these questions, we looked at some frameworks that were already in the market.
So one was the Gartner B2B Buying Jobs Framework that talks about the different areas and types of questions that most B2B buyers need to resolve in order to, to land a deal.
And then deal deciding questions, those came from the Security Alliance and other kind of due diligence frameworks They really need to understand more detail around how your product works, where your data lives, what does it take to implement this product, what kind of support do we get after we go live, all those kinds of questions.
On the left-hand side, there are a set of questions that we kind of think is more of table stakes that most websites answer today, and then the deal-deciding questions on the right-hand side, which tended to be in the purview of a sales rep or an SE.
really the model for all of us in B2B marketing has been create a brochure as a website, get someone to come there, get them to click a contact us form, and then set them up with a sales rep or an SC so they can dive into all these deal-deciding questions.
But what’s really happened is this has changed because buyers now have their tools to ask many of these questions up ahead of time, before they ever get into a sales call. So if you look at the answers that we got, we graded 108 different sites. And what we did is we crawled them to see if these answers could be found on their main web pages.
And take this with a grain of salt, because this is, was there an answer found at all? Not necessarily how good was the answer. So when you see the table stakes ones, things like who uses you, what problems do you solve, do you have any case studies, you know, those are pretty universally answered, on most sites.
But as you get into the more detailed questions, the things like, can you prove an ROI for us, or you know, why you over a specific competitor? Very few sites were actually able to answer that kind of question. So the gap… so here’s the gap, and the answers aren’t there, and we graded two things, right?
We looked at the vendor’s site themselves to see whether the answers were there, and then we went to AI, we went to ChatGPT and Claude, and asked the same set of questions to see what would be returned there. So, on the vendor’s site, you’ll see 55% of the questions didn’t have any answer at all, and overwhelmingly, there are claims.
I mean, we as marketers make lots of claims. It’s gonna cut costs, it’s gonna reduce time to value, things like that. But explanations of how you actually accomplish it were not very prevalent.
The interesting thing is, when you go to AI and you ask the same questions, generally what the AI does is search the vendor site for those same kinds of answers. So if the content isn’t on the webpage, it also fails through the AI as well.
And just as an explanation, this is a little bit different than just general AI visibility, where you say, you know, what’s the best SMB accounting software package? And it goes out and just looks at third-party sites and not necessarily the vendor site.
But if you say, tell me about Intuit’s software, then it’s generally going to crawl Intuit’s site to try to find an answer for that. So some would say that this has actually been a good thing, right?
Because what we’ve tried to do as marketers is hide some of the answers, hide pricing answers and things like that, just so someone will get on a call and talk to a sales rep. But we feel that more and more, if the buyer can’t get these kind of answers.
They’re going to look for AI, or they’re going to find another vendor who will provide those kind of answers for them. So, kind of a cute analogy is imagine being in a restaurant, and you’re ordering, and you ask, you know, what’s the seasonal special? And you get an answer back that says, our menu delivers award-winning culinary experiences.
Or, you know, does it contain peanuts? I’m allergic. And they say, we take food quality seriously. So this is the kind of analogy between the gap of the types of questions that buyers need answers to and what’s actually provided to them.
So, you know, here’s an example, it’s anonymized out of here, but, you know, we’re also looking at one of your direct competitors, you know, tell us, you know, why you over them? Honestly, where are you better, where are they better? And generally, what we saw is, you know, no answer, and what AI built, the comparison out of third-party answers.
you know, G2 or Capterra user reviews and analyst rankings. With actually no input from the vendor themselves, so you’re left out of the ability to put your best foot forward and position your product well. And then there’s a whole set of questions that static websites just aren’t designed to deliver, right?
They’re designed basically as static brochures. So if someone wants a very personalized answer for their specific company or use case, like, for a company of our size, what’s a realistic payback period, what drives it? The standard website will just make general claims, but it can’t personalize an answer like this.
So, we also ran through the same set of websites and searched to see if any of them had AI SDRs or chatbots. What we found is there were… maybe half of them or so had some type of chatbot, but most of those were customer service bots where you have to enter an email address in order to answer… ask a question.
Or they were old-fashioned, menu-driven, click here for this question, and not natural language, or they were off altogether. And… the ones that were conversational were still being used, essentially, to book a meeting’s… book a meeting and not answer the detailed question that the customer has.
So, the example on the left is your data encrypted at rest, and can you share your SOC 2 compliance? You know, I’m sorry, I’m not able to help you with that, would you like to book a demo? That kind of thing.
But we did find 7 out of the 108 that had really high-quality chatbots or AI SDRs on the website who could engage in a conversation and actually provide pretty good answers for most of the questions that we had. So what that brought out is it’s… it’s not a question of, is it possible?

It’s actually, do you… are you willing to leverage the technology to deliver this kind of experience to your customer? So, with that, why don’t I just switch over and give everybody a quick view of how they can run this on their own website. And just give me a second here, I’ll bring up my browser. So, you can go to our website.
This is the whole study that I just described to you, the buyer answer gap. You can read all of the statistics, you know, 6% answer Y over U over the competitor, for example. So in this case, you can click on a button here, and I’ll be… My alter ego, Peter Dumas. And I’ll be Peter at… HSE.com. And let’s go ahead and grade our site.
So I’m also giving you a demo of what SlateCX is, what the workspace looks like. So SlateCX provides what we call a buyer evaluation workspace, and a buyer evaluation agent on the left.
So you can see that my information was passed over to the agent once I filled in that little form, and the agent now gets complete context from the HubSpot contact record, any history, any pages I’ve visited, any things I’ve done in the past, what my title is, what industry it is, and all that is fed into the agent as context.
And then over here, you see that there are a set of different cards. So, the website grader is actually a web application card. And I’m going to pick on hard… skill exchange here. Ingrade your site. So, we could put this in. Oops, that’s true.
So just to point out, when you grade your site, what the AI will try to do is it’ll try to figure out what industry you’re in, who your top three competitors are, what your buyer’s pain points are, but, you know, it’s making guesses about that. You can put in your own information here to get a better answer. And, you know.
I want to prepare everybody ahead of time, this was designed for B2B SaaS companies, and obviously Hartsell Exchange is an education community, so clearly we wouldn’t expect them to grade very high on things like, you know, are you SOC 2 compliant, and where does my data live, and things like that. So the grader will start running here.
What it does is it checks, again, on your site, is the answer there, and is it any good? And then it goes out and asks AI, can a buyer get an answer for that? So while that runs, I’ll show you a little bit more of the Slate workspace. So again, this is a complete workspace. We have different cards that are here.
So this is a video card, for instance, that’s on the workspace. I’ll get straight to the point. So you can play a demo there. The agent and the cards, you know, are in sync, so they know about each other. So in here, let’s say I want to find out about goal-seeking behavior.
Those are some pills, some suggestions for things that you can Ask the agent in the beginning. And then you’ll get an answer here, and then as it’s answering questions, it’s also looking for content that you have in your database that it can push onto the slate.
So you see here, what it did is it actually added a card that showed you, within the Slate Admin UI, you set up a series of goals. So again, we’re in the Autonomous Agent discussion today. This is an autonomous agent.
You give it a bunch of goals, and it tries to achieve those goals with the tools that it has at its hand, and all of the content that you have provided to it. And it can, you know, share any of that content as well.
It’s also proactive, so you can see that, you know, since you’re in demand marketing, are there specific signals you want the agent to capture and act on? So, it will try to qualify the buyer and also ask questions and collect information. I’m, you know, interested in improving… Engagement.
So, again, it’ll give me answers to that, and it’s also collecting all of this information as we go through the process here. But one thing that I mentioned in the slides in the beginning is there’s kind of these deeper deal-deciding questions.
So I’ll just cut and paste this so I don’t have to type everything again, but so this question is, okay, walk me through how the product actually works, end-to-end, visitor lands on our site, then what? What happens technically at each step?
So this is the kind of more in-depth question that you probably wouldn’t be able to… or you’d have to piece it together from lots of different pages on your website.
So you get lots of specific information here, and it also, is adding you know, additional cards, and these cards can be static webpages, they can be PDFs, they can be videos, they can be calculators like walnut and those kinds of things. So a whole variety of different product, content types, I’m sorry, for the card.
So what we believe here is that this is creating what we call compounding engagement. So. People ask more questions, the bot encourages the buyer to ask more questions and explore things that they may not have thought about in the past. And it can also, you know, we can also provide it with specific skills to do things.
So, for instance, an ROI calculator can be a skill. It’s not necessarily a source of information or knowledge or a card that it pushed. But it’s a skill that helps you, do something. So in this case, it’s like, yes, So they brought up an ROI example for me as a card, but I can build my specific ROI analysis as well. You know, I can say.
Yeah, show me an ROI example, given that I have 10,000 monthly visitors, an average deal size of $50,000, and a win rate of 30%. So, again, that idea that it can create these very personalized answers that you couldn’t get anywhere else from static content. And not only that, it can create dynamic cards based on the answers for that analysis.
So this is a chart card that’s starting to show responses, completely customized, weren’t preloaded in any way, like a video card or something that I have as an asset. In just a little bit here, what I’ll do is I’ll show you, you can also save a slink.
And once you save it, that contact is actually created in your CRM, so it’s a way of collecting new leads through the conversation, and then I can also share this, so let’s see… I’m gonna share it with? I think I have your email address, Julia. Let me just copy that.
So you can share it by email, or you can cut and paste a link into Slack or Teams so that your whole team can use the slate to collaborate on. So you should receive an invite, and once you get that, we can actually talk here together, we can reference all the same content and cards, all the cards are shared with you, in the same environment.
And let me just switch over here, and I’ll show you… here’s the HubSpot contact record. Let’s see if we update that. What you’ll see is… there’s Peter Dumas, and we have a Slate CX UI card, specifically. So that gives you all of the intelligence about this specific contact.
You can see how many messages they sent, how many cards they pushed, whether the agent achieved any of the goals that we set for it. So you can see that there’s now a collaborator, Julia. So we’ve identified more members of the buying committee here. Here’s my contact information.
You can see it achieved the goal, it says, of educating the prospect. And also, you can look at the full conversation history.
So every message, every interaction is actually stored, so you can analyze it, you can find out what questions people are asking, you can find out What questions don’t you have answers for, or you, for instance, you don’t have content for, which is really valuable.
And then finally, we also have voice mode as well, so you can interact with voice or with text. Can you hear this? Yes, I can hear you a lot. Okay, great. Hey, this stuff looks great. Is there any way that I can get early access to this product? Nice! That’s an exciting next step. Let me think about the best way to help you request access.
Yes, there is a way. There’s a card called Get Early Access that walks you through requesting access, including the info you’ll need to provide. It currently integrates with HubSpot, so if you… You know, I’d really rather just book a meeting with someone, is there a way to do that? Got it! Let me think about the best way to handle that for you.
Yes. There is a meeting option. I can add a card called Discovery Call with Richard Dumas, which is used to book a discovery call. If you’d like, I can add it to your slate so you can open it and schedule from there. Okay, so that’s just a quick, quick demo of a slate. Again, it’s… it’s much more than a chatbot or an AI SDR.
It’s a complete buyer evaluation workspace, with a buyer agent, and it is persistent, and it’s collaborative. So let me just pop back over to my slides real quick. I don’t know how we’ve done on time here. I think we’re almost up. Away from Kurt’s size.
So I won’t go through the grader report for Hard Skill Exchange, but this is kind of the output that we would have seen there. Best thing is that you can go ahead and grade your own site, and again, it’s really meant for B2B companies, especially in the SaaS space. And more importantly, you can see, as an example with Hard Skill Exchange.
You might say, well, those are great questions, Richard, but they don’t really apply to my business specifically. I would like to customize the answers more. So just as I showed you in the demo, we capture all of the questions your buyers are asking, all the answers the agent provided, where the gaps are.
And we cluster those together for you to recommend new, digital content that you can create. You publish it, and then it powers the buyer agent, the Slate agent, and it also helps you with AI visibility, because I said when someone searches in ChatGPT, Often, ChatGPT is just searching your site for those same kinds of answers.
And, let’s see… Just real quickly, I think what you saw on the screen there as well, I just wanted to highlight, and we get this question all the time, you know, are you a chatbot? What’s the difference between a chatbot and a buyer agent?
So these are the criteria that we look at, and we think about typical chatbots, it just deflects all the time to book a call, book a demo. A buyer agent tries to Fill that gap that we discussed in the beginning to answer the real deal deciding questions. It’s got deep knowledge drawn from all of your content, all of your existing assets.
It’s persistent, it’s not one session only, and it forgets about you. But engagement compounds, because if you come back to the website, you’re sent to the same slate. If you’re sent an email, the link takes you to an asset on the slate. If the vendor runs a retargeting ad, it can send you back to the same slate.
So everything is saved, and everything compounds as you move forward. Assets, as I showed you before, can push all kinds of cards and content. It identifies members of the buying committee, and it’s collaborative. It generates analysis. It doesn’t just give static answers, as I showed you with the ROI analysis.
And it doesn’t just provide a response, it tries to push the conversation forward and get the prospect to think about things that they might not have otherwise thought of, especially if you have a provocative sale that you’re involved in. So that’s… that’s what I have for you today.
I think I’m… I don’t know how much time we have left, but, here’s some links, I guess we can post them as well. On sites, you actually can go grade your own site. We also have another tool that we use to test the chatbots, and what it is, is it’s a prompt that you can use with, Claude for Chrome, and it will test your own chatbot.
So the agent will have a conversation with your chatbot and see what kind of answers it comes up with, and it’ll actually grade it on how well it did that. And of course, we are still, pre-GA, so if you’re interested in early access, please talk to us.

Julia Nimchinski:
Phenomenal presentation, Richard. Thank you so much, and I want to come back to the numbers. So according to SlateCX’s report, 4% of the answers explain how the product actually works, and then 6% of sites explain why buyers choose them over a competitor. Why do you think those numbers are so crazy?
Richard Dumas:
I think they’re so crazy because… as marketers, we have become accustomed to making claims, and maybe even being overly hyperbolic, and trying to, you know, get buzz, get attention, as opposed to actually doing what the buyer wants, and answering the questions that they need answers for. And we have forced them into this mode of.
book a discovery call, even if that first call is just to get some basic information. So we’ve wasted a lot of time, we’ve wasted a lot of time with SDRs on these early calls. By no means do I believe that booking meetings or having meetings is going to go away. I mean, people buy from people, especially in B2B.
It’s just that the buyer is going to become much more informed, and you have the opportunity to inform them more if you provide these kinds of answers. And the same thing, I think, with the small percentage that actually tell you how they position against their competition.
It’s just been a… a real worry that you don’t want to down, you know, criticize your competition, for instance, or get into that kind of battle. But the truth of the matter is that that’s one of the main things your customer is asking. You know, they’re going to ask about your two or three main competitors, and why choose you?
Julia Nimchinski:
For sure. So, a lot of marketers are still withholding information, exactly like you mentioned, just, you know, to get yet another reason to speak to sales. What are your thoughts on just going… I don’t know, what’s the perfect ratio from your experience? What should they withhold and, you know, what should be transparent?
Richard Dumas:
I mean, we believe in transparency as much as possible, and I think that the world is evolving from one of being opaque to being very transparent, and that’s being driven, as I mentioned in the beginning, by The availability of tools to find these kinds of answers, and the vendors that are more transparent, are more likely to be on the shortlist than those that aren’t.
Now, I mean, of course, you know, there may be things that don’t work in your product, or aren’t available yet.
I mean, you don’t necessarily… no market or no organization is going to air dirty laundry, but that doesn’t mean that you can’t do everything you can to try to give your customers the tools that they need to get the information that they want.
Julia Nimchinski:
Richard, we have a question from the audience here, and… The question is, how do you stop the agent from giving a technically correct, but commercially damaging answer, say, about pricing, competitors, security, or product limitations?
Richard Dumas:
Yeah, I mean, that’s a good question. So, I know that one of the topics of this event is all about, guardrails and how do you control what the autonomous agent does. So there’s a couple things. It’s… you are still feeding it your authoritative information.
So, you wouldn’t want to feed it anything that is damaging, or that you wouldn’t want to disclose, first of all. Two. you can test the agent before you deploy it with these kinds of questions to see how it answers. And as I mentioned, I showed you a little bit there, all of the answers, all of the interactions are stored.
So, there are many tools out there today. There’s Langsmith, which we use, there’s Galileo, there are other products that basically help you with the testing and reviewing, you know, whether it looks like the agent is going off the rails in any way, or even placing guardrails while it’s acting in real time.
So that… that’s the answer there, but I recognize that it is an industry challenge. The more autonomous the agent is, the more trust you put in it into it, the more you have to think about, what kind of guardrails do you put in place?
Julia Nimchinski:
Another question from the audience here, and how can a CMO prove the buyer-answer gap is costing pipeline when the buyer will leave, never, identify themselves?
Richard Dumas:
Wow, that’s a good question. I think what we’re seeing is and I didn’t have this slide in my deck, but for B2B marketers out there, and I know this personally, things are just getting more and more challenging.
You know that the cost of a new lead is going up, Google costs are going up, there’s more and more competition in the market, there are more and more vendors, and everyone’s shouting at the buyer. So, everyone I’m talking to is having more… it’s becoming more expensive, more difficult to build pipeline.
So I think we’re all seeing that in our business, and it’s one of the reasons that people are here today and looking at tools like these. It’s like, what kind of tools can I leverage in order to try to reduce those expenditures and to generate more engagement?
So I think it’s really looking at the trend in your own business, and how much it’s growing, and how much pipeline you’re achieving.
Julia Nimchinski:
Richard, folks are comparing slight Cacs with other buyer agents and, you know. I don’t even know how to put the category correctly, so just… you… how do you think about it, actually? About the category.
Richard Dumas:
Yeah, so, I mean, we had one mind before us. There are lots of good products that are coming out into the market that are providing conversational agents for buyers to make decisions, so we’re certainly in that same category. I believe that we differentiate in a number of different ways. One is that it’s not session-based.
It is a… it is a workspace. It’s something that you return to. It’s something where you can save and store information, collaborate with other members of your buying committee. So those are our main ways that we differentiate. We also differentiate, kind of, on the go-to-market pricing. So we are planning to charge per engaged slate.
It’s kind of like monthly active users. So you know, you know, it’s all about building more and more engagement. You can see you’re not paying for any workspace where someone isn’t engaging on it. And all of those signals about their engagement are being fed back into HubSpot, as I showed you.
So, the score, you’re… everybody has either an activity score and a fit score, if you’re doing lead scoring, right? We think this is really going to drive up the activity score, because someone is going to do more and more activities while they’re on the slate.
Maybe also more fit, because it can qualify, find out what your title is, what industry you’re in, if you’re in their ICP. So we believe we’ll be coming to market around 20,000 to 30,000, ASP. So that is significantly less than many other products on the market.
Julia Nimchinski:
Phenomenal. Thank you so much for the amazing presentation, and what’s the best next for the community to support you?
Richard Dumas:
Well, right here, go to our site, you know, grade your own site, and as I mentioned before, there are plenty of options there to request early access. We’d love to work with companies.
We have a small, number of, kind of, founding customers we’re trying to, fill the slots for right now, and we’re going to provide specific hand-holding and attractive discount pricing for those first customers who work with us. So, please reach out if you’re interested.
Julia Nimchinski:
Amazing. Thank you so much.
Richard Dumas:
Thank you, appreciate it.