AI-Powered Demand Gen with Guillaume Cabane, Co-Founder and GP @ HyperGrowth Partners

Thrilled with our guest for today and to see everyone. Welcome, welcome. As always, we are thrilled to have you here at Workshop Wednesday. Please make sure to drop where you're tuning in from so we can see where in the world you're from. G, I'm guessing you're joining us from the south of France.

That is right, just above the city of Cannes where you have the film festival these days.

Wonderful. And we are gonna see Ji next week when we are in London for SaaStr Europa. There are still tickets left. So if you are on that side of the world with us too, please make sure to purchase your tickets so that you can see us in person. I'm gonna drop our Workshop Wednesday code for you guys because there are still a few tickets left. But thank you so much for joining, Ji. We are thrilled to have you.

Thank you. Thank you. And yes, SaaStr Europa last year, I was there, was the best event in Europe. So like such a great like, you know, minds meddling and like, you know, vibe just so good.

Amazing.

Well, thanks. And we didn't pay Ji for that little plug either.

So no, I don't get paid for speaking.

Thank you guys so much for joining. And Ji, we are going to get things kicked off. As always, we're going to have plenty of time for questions. Ji has an awesome slide deck for everyone to go through, but he's going to go through his slides pretty quickly and then we're going to have plenty of time for Q&A. So please make sure to drop your questions in the chat. Chat throughout the session, and I'll make sure to get you hopefully queued up to be able to ask Ji your question live. But we are really excited to have him back for another amazing session. Ji Cabane of Hypergrowth Partners. He is a SaaStr fan favorite. And today he's going to be talking about a topic that Jason actually spoke about on last week's Workshop Wednesday about incorporating AI into your demand gen, into your marketing. So super timely, super talk— topical right now and everyone's trying to make sense of what's going on in the ecosystem. So we're really thrilled to have Ji's presentation and help us sort through it all. Before we get kicked off on the slides, would love for you all to get to know him a little bit better. So Ji, let's get kicked off and would love to hear a little brief intro of kind of what you're working on in your day-to-day right now and anything else you want to do to kind of set people up for your presentation today.

Sure, sure. So, you know, everybody calls me Ji and that's, that's great. I love it. And I've been doing growth for the past gosh, more than 15 years now. And these days I advise companies like post-Series A on how to grow faster. And as an example, I'm currently advising Reddit, which you'll see in the slides. I'm going to show examples of that. Previously, I was advising Ramp, the fintech company out of New York. So some pretty like fantastic companies.

Amazing. And you're probably going to preview this a little bit, but we love to ask everyone what is the current SaaS tool or platform that's your favorite right right now?

Yeah, so I mean, obviously focused on like my topic of choice, which is like, you know, demand gen, especially outbound. I think probably these days it's a battle between Clay, which is like everywhere, and probably Unify, which is my former mentee from, from Ramp that's building that, is doing also very well. So I'm not going to pick favorites, just like between your kids. I'm not going to pick favorites. So those two are great.

Amazing. Amazing. Well, without further ado, I want to make sure we have plenty of time. I'll let you take it away and dive right in.

Awesome. Thank you so much, Taylor. So, you know, as we're discussing, my name is Ji. I'm currently at Hypergrowth Partners where I'm the co-founder and JP. I've actually previously launched the growth teams early on at Drift and Segment and Gorgeous. And yeah, I work with great companies like Reddit and Workstream and so many others. We have like We've worked with over 50 different companies at Hypergrowth. So, you know, if you're in the Series A stage or later, you want to grow faster, like, do, do reach out for advisory. With that said, for today, I want to cover the evolution of the B2B stack and, you know, where we're coming from, especially going back to, you know, 2015 and where we're now and how to pick the tools given on your stage and your scale and your needs. And then I'm going to go into like what you all want, which is like the practical example and workflows. So with that said, let's get started on the evolution of where we're coming from. And the reason why I'm sharing that is like I've been here for a while, you know, I've been doing marketing for more than 20 years now and I've been doing outbound and DemandGen for almost all of that time. And so I have a pretty good perspective. But you know, what's going on with AI is like it's heavily taking over the sales role. And so this presentation and this deck is geared towards outbound and AI taking over sales and how can we, you know, make the most of this switch. So, you know, if we go back to something like, it says 2018, but probably even before, like probably even like 2015, it's a world where the outbound was run by the reps, by the SDRs, and reps were in charge of everything, the contact list, like building lists out of Sales Navigator or ZoomInfo, maybe if they had a license, but also the email copy, the sequences and the mailboxes. And he had this one-to-one relationship between the humans and the mailbox, like one rep, one mailbox. So as a consequence of this very manual process, the output is a low volume. Of email sends, low volume of engagement, and a fairly high cost per contact engage, or high CAC. But you get a pretty good response rate. You know, 2015, 2016, 2017, we would get maybe like high single-digit response rate, positive response rates, which is why we had huge SDR teams back then. And so if you move forward to, you know, the end of like 2019, towards 2023, I'd say that bottleneck of having like those huge and that cost of having those huge SDR teams started to be identified and solved really by those platforms like Apple Market and Apollo, but also by having more technical people handling the infrastructure, being, having the ownership of that infrastructure still within sales, but you had like more like sales ops And, and those platforms have both the audience product as well as the sequencing product. So in Apollo and Apple Market, I'm sure you know, you can go and like find the contacts and build the sequence, right? So I'd say much more efficient, but the reps are still in charge, right? More efficient, but let's say they are empowered, but they're in charge. Well, obviously, you know what? No, you know what happened? What we all know, which is outbound volume exploded. All right. And while the cost per contact engaged, the CAC decreased, but so did the conversion rate because everybody was spamming everybody. And so we got to a point where we were below 1% engagement rate at a fairly high volume. And you can see like all those curves, like are across the board there. And so we were at a turning point at the end of like 2022, 2023, kind of like outbound was like no longer working at that scale. But now what's happened, which we were all surprised by with, you know, the coming of age of like real like generative AI tools, it's that sure, like we're going to go into it. The workflows are all automated. The copy is fantastic. But what I wanna talk about is that the process is entirely out of the SDR hands, right? It is, they're out of the process. The modern stack covers everything they used to do. So like the audience building, sure, that's automated. It's based on signals nowadays. The copy generation, automated. The outreach and the response handling, automated. It's entirely within the scope of growth of marketing, growth marketing with technical people. What's fascinating is that the conversion rate is back to single high digits. Now, whether that remains the case or not, I'm not sure I can predict. Well, maybe I can. Like, it seems unlikely it's gonna stay high because it's high right now 'cause most companies have not adopted the modern approach yet. And so there's like those, this few number of companies that have this great copy, great outreach, great sequencing, and the rest don't. Eventually everybody's gonna catch up. And, but what it means is that there's a, there's a time window right now to act and reap the benefits. So how do we do that? Right? So, you know, where do the SDRs end up in the, in this AI world? I guess they don't. I guess they don't end up in this AI world. They become AEs, closers in ops, or they, they're displaced entirely. All right. For now, the closing is still done by humans. Humans mostly want to talk to humans to buy something and sell something. Though there are industries I know of, companies, products that are being built for specific industry sets like dev platforms where engineers would prefer to buy from a bot than from an AE. So you're gonna see some evolution on that in the next coming, you know, months. So something to look forward to. Okay, so we looked at history. Now, which tools should you pick? What's the right tool for you right now? What do, what do I recommend? Well, first I want to say that there are, there are 3 focus areas you should look into. The first one, it's the, the TAM. So the audience building, right? TAM is for total addressable market. Audience and scoring the audience, break it down into like multiple segments based on the forecasted value and forecasted cost to close them. The second is the intent, what we call signals, right? To, let's say, laser focus the efforts on a smaller subset of that audience to increase the conversion rate with better copy, what we call an excuse. And the last is, well, the action, right? How do we send an email or LinkedIn and whatnot? Once you've looked at those three buckets, those two pillars, you can decide, well, do I want a platform that does everything for me, right? Like a done-for-you, right? Or do you want to build, let's say, a composable stack? I do both depending on the situation, right? What's clear is that a lot of the early done-for-you platforms like, you know, the Outreach, the Salesloft, and even the, maybe less, but even the Apollo, like those are becoming a legacy, right? In Apollo, you'll find a touch of AI. Those are becoming a legacy because that's very important. Their pricing is user-based, so they're constrained. They are, they're incentivized to have more reps on the platform, which is not the world we live in today. Just keep that in mind. The platform you want to invest in is a platform where The pricing is incentivized to have less humans, not more. Okay. The second thing is there is a volume where it makes sense, uh, maybe a percentage of your pipe to invest a lot more than just a platform and control the different components, which I call it a composable stack. So let's look at that. So first you have the, let's say, making the reps more efficient. So we have the outreach, the puddle, the input market. Those are for reps. So there's no code really. And they work well up to like maybe like 100,000 emails, maybe a bit less, maybe like 50,000 emails, I'd say a month, which is high, but you'll see we can go much, much higher. There's little customization possible in there, little, let's say, of the AI craziness we can put in there. If you want to have a bit more control, the newer platforms would be something like a Clay, which I use a lot. We've talked in the very, at the very beginning, Unify and Commonroom for the signals. Those are very different. Like Clay is like a spreadsheet, and I'll show you some examples, spreadsheet-like for outbound, great to test and discover. Your audience, your signals. Commonroom is a marketplace for signals, and Unify is an all-in-one modern AI platform. But if you want even more control and more volume, the things that I would use at the large companies like the Gramps and the Reddits would be Cargo plus Waterfall, which is, you know, something which is much closer to a Zapier or Make.com. For outbound purposes, and you can just stitch all of the components that you want. So let's look at a couple of the examples. Speaking of Clay, here's a great example on like just in minutes, how do you estimate the audience size, the number of accounts based on a set of criteria? So in this case, you know, I'm looking for companies that work in SaaS with specific company sizes, maybe specific keywords, right? So if we're doing enterprise B2B SaaS, MarTech organizations and kaboom, for that set of filters we find 4,000. Now when you get the TAM size, you get the list of companies, you can quickly validate that it's the right kind of companies. It's really hard to go anywhere else and like do that, inspect move it to a table and play around with the table. So this is like the best tool for that, right? And obviously you can build a scoring system in there. So you can say, hey, like, you know, based on those companies, based on the fields that they have, have a point-based system to like sort through the companies. Now if you want to laser focus a large list, and you want to have a good excuse to put in your email, you know, whereas in the past I had to build signals one by one, like pull like Clearbit and G2 and do all the API integration. Now there's like marketplaces where you can do that, one of which is Commonroom here. And so in Commonroom, you can just like in a few clicks, like activate the integration from like, you know, a Reddit, a LinkedIn, a Slack, Twitter, like G2. It's just all there and just it will aggregate and find all the companies that are showing intent based on your filters. You can see the example of the intent here. You can actually aggregate the intent data per account and get an intent score, which is something I struggled to do even like 3 years ago. Right now you can have a bell curve of intent with maybe a decay model and push those accounts, those contacts to whichever engagement platforms you want, which is fascinating. It means that those tools are displacing most of the value part of outreach out of outreach. Outreach is used only to send email, but all of the intelligence is out of it, lives elsewhere. Now, if you're in the composable stack camp, then a great signal I'm always using is job changes, right? Whether it's a champion moving campaign, so somebody who's moving from a current customer company to a prospect, non-customer company, right? And you wanna reach out, or whether you have a gate blocker, which is the opposite. Somebody said no to buying your product and eventually they move out. And you want to reach back to the same company. That means you're scraping, monitoring LinkedIn changes all the time. You don't want to do that yourself. Well, then you can use a platform like LoanScale here to, um, track a set of accounts, a set of users, or maybe even target accounts that are opening roles for a possible champion based on the job description. As you can see the bottom right. That's like just so great. You can't imagine the amount of work I've had to do to make this, you know, happen just a few years ago. Speaking of the composable stack, the best, so high volume approach, which I covered in the 2x2 earlier in the axis, is something like a Cargo. So as I said, it's very similar to Zapier, Make, or n8n, it's entirely dedicated to B2B dimension automation. So as an example here, it has direct integrations with Waterfall, which is probably the best— well, actually the only contact data aggregator. What that means is that it enables you to have multiple vendors, data vendors into one API and just pay the bare minimum price and just have one endpoint. And you can have the the ZoomInfo and the Apollo, the Clearbit, the Lusha, PDL, all like one API point. All right, so just going back here, you know, you can have a workflow where you're gonna get the signal from, you know, a loan scale, hit Waterfall, have the best data coverage, and push it back to whichever outbound system you want. And this is where the intelligence lives. This is what a platform, a company, sorry, like Reddit or Reampl, all the high-volume senders sending more than 100,000 emails a month are using. This is what Gorgeous uses to send like 300,000 emails a month, because you have a lot more control and you're not dependent on any part of the stack. So let's jump into some examples that I worked on recently here. So as I said, I recently became the advisor for DemandGen. To the Reddit advertising business, the SMB advertising business, and a great company. And they came with the goal, we, they said, want to get to 10% of US advertisers on Reddit by the end of the year. And that's a tall order. Like, yeah, sure, it's Reddit, but like, we're talking of 10% of the entire US market. Okay, so how do we do that? Well, one of the first ideas is to scrape the Facebook Ads Library, Google, LinkedIn Ad Library to find who's advertising right now that would be a good fit for Reddit based on Reddit's ICP. So, you know, gaming companies, SaaS companies, and reach out with a good excuse about, hey, we saw your advertising on Facebook or on LinkedIn. What about, you know, replicating the same ad? Well, that's exactly what we did. As you can see here, this is screenshots from Clay. Now we use Clay to prove the feasibility of the concept before we invest too much. We just wanted to go there. And you can see here there's a GPT prompt to find the Facebook page, which is harder than one would think, accurately at scale. And then get the Facebook Ads Library match, pull all the data from the Facebook Ads Library and inject it in an email. And that's just fantastic. It worked. This was a couple of days of work and with a couple of days of work to get it fine-tuned and nice, it's just running. It's humming. I can do like a couple of thousand, it's a highly targeted, highly like customized, personalized emails per week. It's not super high volume. It breaks once I went over like 50K rows on Clay, but this is just magnificent. Like, this is just so good. Like, I saw your ad on Facebook, YouTube, whatnot. Boom. Summary of the ad, which is AI generated. All the ACWAs here are AI generated. And there's a screenshot of the actual ad on the side. And then we're even injecting the relevant subreddits. That are fit based on the industry and the company data that we get. With this, we are getting pretty good. Obviously, this is like, I can't show the actual data, but we're getting really good reply rates, really, really good reply rates. It's amazing for a stack that costs us close to nothing. We're talking like a couple thousand dollars of entire stack. That's almost 100 times less than it used to be before. So fantastic. Similarly, similar end-to-end, but with a platform now, you can do the same thing all in one within, right within Unify. So you can see here within Unify, so the opposite end of the spectrum, no code, in a couple of minutes, you can get all the list of companies visiting your website. Built an intent level over the past 30 days, get this bell curve, as I said, of intent, all the people, the prospects, and create an engagement, right? And say, hey, based on the company's accounts that have seen my pricing page with a champion that changed job and on G2 saw my competitor, and are not on Salesforce, then move them to Salesforce and engage. That's just like, wow, this is so good. Those accounts here are just so, so qualified, right? And so, you know, you can build and stack playbooks one after the other. You can see here, like, it's doing a couple thousand companies at once and you can have playbooks at the accountable but also at the person level. You can have playbooks based on seniority, based on role. So that's like the all-in-one approach. And if you look here at an example from Unify, you can see here that the prospects have fantastic reactions, right? Why? Because we're doing this like typical of this reveal loop approach for which I personally am famous, which I kickstarted in 2017. This is becoming a standard approach, but still getting that contextually within minutes, high quality, it's hard to pull off, to be honest, without such a platform. Okay, so with that said, I shared a couple, a bunch of examples. I do want to give a couple of takeaways and then some thoughts, and then we can move into the Q&A. So takeaways should be obvious. Like, outbound lives within, you know, growth marketing ops. Like, live with it, embrace it. It's the way it is. It's not going back. Okay. Um, it's all about moving the human, uh, prior human workflows of audience, uh, finding the TAM the signal discovery, which used to be like on LinkedIn, and the outreach, which used to be in, well, in Outreach, to an entirely automated workflow, which can be built-for-you platform or let's say low-code or some code based on the volume and the personalization that you want. And yeah, that's, that's it. That's where we are right now. A couple of links here, which I'll leave open for a second, of the tools that I've shared. Unify, Clay, and Waterfall, and LoanScale, and Cargo, and Apple Market. Great platforms. Before we move to Q&A, one last thing, some closing thoughts. I was with like a, a CMO yesterday and we shared a couple of thoughts I wanna share here. One, the tech cycle is accelerating. It's obvious, you know, looking after, you know, at, at the cycles I presented earlier, the first cycle was like 7 years, the second is like 4 years, and now 3 years in into this AI cycle. And the products change what, every quarter maybe. So, so how do we as marketers, as technologists cope with that? How do we make a tech investment, a bet knowing that that decision that product we're buying, maybe on a yearly contract, can be superseded next month by something way better. It's very difficult. Second and last thought, as we discussed, we're now in an era of very high quality automated outreach. Soon, if not already, indistinguishable from the best human SDR. Really, I can see it in the conversion rates. It's like pretty much the same thing. How will buyers react? It's not just outbound that's at stake. It's the core concept of reciprocity and how people perceive the entire email channel as a communication channel. People respond to people. What happens when buyers lose trust in email as a communication channel because they think it's all just like AI bots. And how will we handle that? That's it. With that said, thank you so much. I'll go back here and we will go into Q&A right on time.

Great presentation, Ji. Thank you so much. We have a lot of questions as well. I'm going to start off first with some of the more tactical questions that we're getting, just like on the actual workflow examples you gave. And then there are some more like bigger picture practical questions, especially for some folks that are a little bit earlier stage, but we'll start with just some folks again that had some specific questions about your workflows. Mike, I'm going to pull you up. You had a question about phone calls. Mike, are you there to jump in and we can bring you up?

Yeah.

Thank you, Taylor. Thanks, G. That was pretty awesome. My question is for, you know, you brought up like Salesloft and stuff like that back in 2018 and, you know, grab your 50 leads every morning and, and all, and all that kind of stuff. A lot of those cadences involved, you know, phone calls and different voicemails that you would leave and stuff like that. And certainly, so one, just in general, what about that part of like an SDR's tasks? And then, and then, uh, another like question has to do with it is there's probably some correlation between like deal size and current effectiveness.

Yes. Yes.

Like what you're talking about. And so that, that if you could kind of include that.

Sure. 100%. So it's funny because like I kept this for like more the early stage audience, but great questions. So on phone calls, you're absolutely right. There's a comeback of phone calls because of what I just said in closing thoughts on email. 100%. It is industry dependent. There's some industries where phone calls just work better. Especially more like traditional industries. There's tools, for example, like Aurum, R-U-R-U-M, sorry, that makes phone calling at scale a lot more efficient, still using humans to do the actual conversation. I've tried the AI platforms to place and have the call, the voice. For now, they don't work. We've placed 427 calls in the most recent example last month, and we had a zero success rate, not 0%. We had not a single call went well. Okay, there's just too much delay. The AI just doesn't pick up the, you know, what the person is saying. It's just, it's just weird. Okay, it's going to get there though. Like, it's eventually going to get there. It's a question of like, when does it get there? We're not there just yet. For now, people will What's very obvious is the audience, people, US consumers will get used to AI voices on inbound calls. When you're calling United or Comcast, you're going to get that instead of a phone tree, which is a better experience. And after some quarters of that, they will be used to like the outbound phone calls. Yes, we did do your second question, which is the deal size, 100% at Reddit. At Ramp and almost everywhere. The examples that I shared are for SMB and mid-market. Above mid-market, so the enterprise territory, um, we don't use it because the risks involved of like wrong copy, wrong person, yada yada, outweigh the cost savings. There's just, there's not enough enterprise companies for that to be worth it. Like in most companies I work, there's like a couple of thousand real like enterprise accounts, full stop. And just with like a couple of reps, you can just cover that neatly and just too much history. You can't right now load all of the context of like, you know, 20 different, you know, business units. So if you're selling to Delta Airlines, this is not going to work. So yeah. Thank you.

That totally answered my question. That's similar to what I thought you were going to say. So that means you did a good job in your speech and your talk, given that people could kind of guess at what the answers would be for their own questions. Good job.

Thank you, Mike. We are going to move on to a question from Ash. He has some questions about how you include— what details to include in the email. So, Ash, I'm going to bring you up now. To ask your question to Ji.

Oh, hey, thanks, Ji. We do a lot of outbound ourselves for clients. Appreciate this talk a lot. I'm curious, you know, nowadays you don't really have the chance to include a lot of links and things like that in emails to have, you know, because they're filtered out from IT spam filters. The example that you used, you actually, you know, scraped things like Ad Library and things like that, and then you included an image in your outbound email. To that, to that audience. I'm curious, how does that impact your deliverability? Is that, you know, things like that working? And then a corresponding question to that is like, you know, what, what is your answer to how we bring audiences into sort of, if you will, a digital sales room or some sort of piece of collateral or content beforehand, before they even engage with, with you afterwards?

Okay, okay, great questions. So Traditionally, I agree with you. I avoid links. Every time I see, you know, a company I'm helping where they have a direct Calendly link, I ask them to remove it and go for a response instead to make it more human, right? I never, in my, as a human, which I'm supposed to be most of the time, I never send an email to somebody else that I don't know with like, hey, how about you book a Calendly link? Because there's no, connection. I never do that, right? I try to see, is there appetite on the other side? So it's just not human-esque. So we try to avoid that. That said, it depends on TAM size. So can you, can you, do you have a benefit of like a lower response rate, more efficiency, right? It depends on domain authority, especially for including images. Now Reddit has one of the best domain authorities in the world. And they send gazillion of legitimate emails to legitimate users per month. Like, I don't know the number, but like, it's just called, it's like enormous.

Okay?

And so it's so big that me sending 10,000, 50,000, 100,000 emails of outbound to businesses per month doesn't even register on the spam filters because the, the domain is whitelisted. It's Reddit. And so my strategy changes. And every time I can hide my emails behind the forest of legitimate emails, it also works for Gramp, you know, which sends a gazillion, maybe less than Reddit, but a lot of like, you know, add your receipt and like here's a transaction per day to legitimate users. And so in all those cases, I'm like, hey, I can take more risks, I can do things differently because it looks, if, if it does get through, obviously it looks more human. 'Cause nobody thinks these days that this is automated, especially for Reddit's audience, which is SMBs. So like, it's like we're talking like small SaaS, game developers, indie hackers, all the people who could advertise for Reddit communities. Like, it looks very human. So I can do the screenshot. This is, hey, what about this community? Have you thought about it? This ad would play very well. So it looks human, which is what I want. Yeah.

Yeah, that's perfect. That's really good. Yeah, that's a really good clarification because I think, you know, from Reddit to RAMP, that those things might work, but they might not always necessarily work for, you know, small to medium-sized enterprises, which are still, getting their emails warmed up and not necessarily hitting the inbox.

Yeah. And volume-wise, if you're a small business and you don't, you can't hide behind that force of legitimate, you know, emails, then you are going to have to build a quite a large number of like new domains, warm them up. And I would say these days probably stay below 60 emails per day per domain. Which feels— sounds low, but that's where we are. So I have people who are building 100 domains with like 1 to 2 mailboxes per domain to get that, you know, that's where we're at these days.

And for what it's worth, social is working out really, really well for us.

Yep. And for your second question on like digital sales room, I don't do much of that. That's more in the enterprise territory. But I have found, which brings, you know, kind of close to that, which is there are some audiences which are not warm enough. We don't have enough signals. And so we do want to nurture them, send them content. And so we do things like, for example, building free tools and do the outbound on the free tool to create the engagement and the brand awareness and see if there's engagement there and then bring them closer to the main paid platform, that works. And so trying to like find tangential opportunities, signal opportunities, companies launching, hey, like this free tool helps you like do this thing. So that's the closest I've done, which kind of works.

Thank you so much. Appreciate it.

Thank you, Ash. We have another question very related. So Ash, this was a good segue. So we have a related question from Tito. I'm gonna bring you up right now to ask your question live. And it's probably gonna be a few more details about the domains and creation and how to kind of scale up. So Tito, I'll let you dive in.

Thank you, G. First of all, I appreciate you doing this. I've been following you for a long time, and I've been in the sales development space also for 12 years. Um, it's good to have a sounding board today. I actually want to pick your brain a little bit and see if you're seeing what I'm seeing. So if I understood here correctly, you know, you're working with, with companies whose average deal size might be very small, and some of your, you know, capacities and tools that you're talking about are based off of you know, the 50,000 emails, right, being sent and all that. It's very high volume. You talked a little bit about the enterprise being different and it being much more high volume. But I've actually started talking about this with other people. Oh, I was trying to share screen. I wanted to share some of the emails that I've been getting. Is that, is that possible?

Can I do that, Taylor?

Let's see if we can show you.

If not, no big deal. But here's what I'm seeing in my inbox, G. I'm getting a lot of emails and now they look personalized, but I can smell the AI so quickly. Like it says, hi Tito, I'm excited to see how your leadership has influenced the innovation of B2B.

Yeah, they're not great.

You know, and I can smell that. So I think similar to in the early days, the first time I got the hi Tito in 2004, I was like, Somebody's emailing me. Now I'm desensitized. You send me— I don't care. So I agree, right? Good. But what's next? Because there's no—

wait, there's two things. One, you're not— you're not— you're different. You're not the usual buyer. In most cases, you work in sales and marketing, and so you get a lot of those emails. You learn how to identify them. And you know how they're built. There's many, many industries where that's not the case. That's one. Second, a lot of those emails either lack somebody who's really creative, like crazy creative like me— I can give some examples of that— or a more modern AI platform, which is happening. So for example, like we at Ramp did a sequence for CFOs of SaaS companies, which they should be able to smell BS like from pretty far, right? And we had no signal. So it's cold, cold email, which is the worst situation. And the email we built is we scraped very typically the LinkedIn page. Nothing surprising there. We found the college. Nothing surprising that we did not want to say, hey, you went to this college because that's not great. Okay. What we did is that we paid a freelancer to identify What is the most— sorry, what are the college sports team of every US college? Right. Then we pull the data on the coming upcoming sports games. Okay. And we send an email to the CFO and we said, hey, Tito, I see you're from like this college. The, you know, basketball team is going to go play against this other team. And I'm betting you $50 that they're going to lose. It's $50 on me or you do a call. All right. Right now, nobody believes that this comes from a human. And I see Rob says that he got this email. Yeah. So nobody— why? Because it's so out there. It's just so different. It's like me emailing you, especially in the US where, you know, all CFOs went to college and they're all people who went to college, especially at that level, keep very strong ties with the college sports team. Nobody believes it's from an AI, but it is. Correct.

But what happens when, you know, that's the high tido of 2004. Now in 2024 is all these other things.

Yeah.

What happens when this merges? Like, will I, what I've been doing is I know—

That's at the end. So exactly what I said at the end is you, you will either, there's gonna be two kinds of people. There's people who are gonna say, hey, I don't care if this comes for human or not. I no longer care. The only thing that matters is whether the email is valuable and I get value from this. And B, there's gonna be a second part of people who are gonna say, hey, I no longer trust this form of communication 'cause I don't know if it's from a human or not. And so I will retract into my bubble of like, hey, this is not trustworthy. But that's gonna happen to email. It's gonna happen to phone calls. It's gonna happen to all other things. And so What we express as reciprocity, which means when you are convinced it's from a human, you will do some efforts, that will happen less and less often. That's the concern.

I'll share one last thing here. I, I wanna see if you're seeing the same thing, but I'm desensitized for, for example, to direct mail. When I get a piece of paper in the mail, it says, Tito, offer for you. I'm like, you know, throw it in the trash. So I think email is going the same way, but what I think will happen is, for example, we do a lot of outbound. In the enterprise space. And what's working is we go to an AI tool and we build, let's say, a song for the company, and it's how G and Tito should be friends. And then I send you an email that says, G, I made you— made a song for you, check it out. And it's on YouTube.

We upload it, you know, like, I think long term, long, long term, all of this, you'll be convinced it's from a robot. I'm going to share two things to close this out. One, you're pulling my leg on the stay on my example, on the physical metal example. So I got the shared list, like everyone in the room here, think about this, the 100 people here, think about this thought experiment. You're Tito, you have this great office, home office, and you have a mailbox and you get junk mail in the mailbox on a daily basis. Okay. You take the junk mail and put it in the trash. You don't have any remorse, no second feelings. It goes in the trash and you move on with your day. Okay. Now, as you pick the Next day, the pile of junk mail. On the top of that pile of junk mail, there's an envelope that is obviously human written with scribbled with an old handwriting. And there's a stamp that's licked and put on the envelope and there's your name on the back of it. And it's obvious it's from a human and from an old grandma. Now, how likely are you, Tito? How likely are the other people in this room to take this envelope and put it in the trash? 'Cause you don't know who it comes from and you don't care. Zero. There's no way. There's no way you can open it. Why? But why? Why?

It's, it's, I, I see the ink. This is pattern interrupting because now AI is the pattern and only a human can break the pattern by being human.

There's two things. Yeah. There's two things. One, you have reciprocity. You feel for the person that went through the effort of writing to you, to this curiosity. Okay. So those two things are happening. That is what we try to replicate with the AI tools. We're trying to trigger your reciprocity sensor and your curiosity sensor.

But if I get 20 of those exact envelopes, like, exactly, one doesn't work anymore.

I agree. But that's the point. You've got to be, you've got to be the person who's on the edge, who's always doing the thing that the others aren't doing. You gotta be in advance. And so that's why you've got to be on the leading edge of looking like a human and delivering more value in your communication, where all the others are part of the folks who are like trailing behind and sending you like postcards, which are like printed.

But you're either the leading edge of AI doing the things that nobody else is doing with AI, or you're going to be the human that's doing the human things that AI can't do. And I think on the AI being on the leading edge, It's excellent for SMB and I agree with you and I love your presentation, by the way. It's excellent. I got to think about more of that. But I also think that the human side, humans need to keep evolving into what can the AI not do yet? If I do that, I open accounts and that's what you do in the enterprise. Yes, that's my belief.

Yes, we agree. Okay.

I got to buy you a beer or coffee today.

I want to go chat. You're incredible.

You have to come to London and Ji will be there to dive in more.

Let's grab a beer slash coffee, whatever your drink of choice is.

Yeah.

I'm flying to London tomorrow actually, so I'll buy my tickets. I'll go see you there.

Yeah, awesome. Thank you so much, Tito. Tripp, I'm going to pull you up. We're getting a lot of questions of just people talking about testing. So like, how do you kind of get it launched and going? So Tripp, I'm going to pull you up now to ask your questions, and I think it covers a lot of other folks too.

So I'll let you dive in. Sounds good.

Hey, thanks very much for the presentation. Really good, inspiring stuff. So you hear me? Okay, cool.

Yeah.

So we've got existing investments in existing tools, the ZoomInfo and, you know, Gong and kind of all the rest of it for outbound. And so, and an existing motion that's based on intent and all this kind of stuff. So my question is like kind of thinking about how to kind of find a test case or kind of a very, you know, what would you recommend to kind of the minimal viable product version of, you know, this kind of approach? I also, by the way, really appreciate all the nuance you provided because we're in a market where it is a kind of constrained TAM. We're going after a niche. And so it's a tight community and we don't want to, you know, kind of you know, kill the golden goose here in terms of creating a bad impression or kind of burning our brand authority in the space we're in. So we want to kind of go about it with a lot of empathy towards the buyers in this kind of tight community.

Yeah, that's fair. Yeah. So for all those cases, I think that's something like, if it's to test and you don't know whether it's going to work, you don't know what audience angle fit is. You don't know what the exact message for the right audience, you don't know how to break it down. You know what signals work. You want to test a lot. That's when I use Clay. And that's— Clay is honestly like, I love this. One of the few companies where I don't even own equity, which is rare in this space because I should own equity everywhere. But like, I don't here. And I love them because I use it a lot with my portfolio companies because it's great to test. Downside, it doesn't scale super well. Like you want to go past, you know, the 50,000 rows and the enrichments and all the columns, it tends to break. But that's okay because by that time I've figured out and you know, they have a credit-based system. You can use all the vendors based on credits. So you don't need to go through like all the, you know, the SaaS purchasing cycle for that, which is a pain. So you just put like a couple thousand bucks to work on Clay and you're going to test the vendors, the audiences, the playbooks, send a couple hundred emails here and there, and you'll know, right? And at the point where clay breaks, if you have a good response rate, you're able to get to parity with the human SDRs, then it's probably worth it to invest into a composable stack. Absolutely. Right. But you gotta find those playbooks before.

Awesome.

Thanks, thanks.

Thanks, Tripp. We have another question from Gaurash, a little bit more about enterprise. So, Gaurash, I'm going to bring you up as well.

Perfect. Hey, great session. Love it.

Thank you.

I have two-part question. The first one is— should be easy for you. So let's say I'm targeting chief information security officers or chief data officers, Global 2000, of leaders, right? So, uh, and the question is really related to intent signal and the real-time data from those signals, because everyone is invested. And I won't list the name of the tools, lots of ABM tools or platforms and things like that, right?

How—

what you proposed is phenomenal, right? I love it. And how do you see the real timeliness or lack of it?

Yeah, from the signals, and to address that audience Now that's a great point, and I'm just going to add on top of that before I answer. The real timeliness, freshness of the information is a valid point for the signal, but also for the contact data. People in tech jobs change jobs, you know, every 18, 24 months, which means 1/24th, you know, of your database changes per month, you know, at best. Okay. And so if your data vendor scrapes LinkedIn, you know, every— the entire LinkedIn every 6 months, then you got 6/24 of the database, which is out of whack at any point in time, which is terrible, which is why you have bounce rates. Now, given the bounce rate should be in the low 2 to 3%, you can see how 6/24 is a problem. All right. And so what I recommend is that your investments on the stack, especially on the data part, scales with the ACV of the contract, right? Which most people don't do. So for example, for not for Reddit, ACV is too low, but for Ramp selling to CFOs where an ACV is in the 5, 6 figures range easily. Spending $0.10 doesn't make sense. We should spend multiple dollars to get the proper data, which we do. We absolutely do, which means we pay data vendors at Ramp to scrape LinkedIn real time at the moment we are about to send the email, not 3 months before, not 2 weeks before. In the workflow trigger, once we have all the signals, we recheck LinkedIn to make sure real time this person is still in this job. At this company. Okay? And the signals, the higher the value, the more we invest on getting our own source of signal so that we can pay to run faster on a smaller subset or to create custom signals. That's even more valuable. As platforms like Commonroom start to become commonplace, everyone's gonna use the signals and people will be desensitized like Tito was saying earlier. You have an incentive to build custom signals. Example of which, for example, when I was at Gorgeous in the CS e-commerce space, we built signals to monitor weather disruptions to USPS and FedEx warehouses. So we could reach out to e-commerce stores and say, hey, your packages to your customers are gonna be delayed. Are you ready to handle the influx of CS tickets? And often we'd be able to send that email and get in the inbox of the CS leader before they knew that they would be disrupted. So we would inform them, and that email is valuable. They said thank you, and then the storm disrupts the, you know, logistical chain, and we're like, great, we gotta talk. Custom signal. So invest more to get higher quality data. Alongside your ACP.

Cool. Thank you.

Sure.

Thank you so much, Karesh. We're going to close out, Ji, with one final question, going back to your first point in your presentation about the future of SDRs. So I'm going to bring up Toshin to ask your question.

I was afraid you were bringing up my wife, who's a salesperson, so there's an SDR.

Toshin, are you there?

I'm here. I'm here.

One second. How's it going? I'll bring the wrath of your wife. So, um, I started as an SDR and it's where, you know, we gain a lot of skills to essentially close sales. So my question is more of like a two-prong approach. Um, one thing I'm kind of recognizing with AI and adoption at these larger companies is that they're just not doing it, like, or they're very slow to it, which is why I still think the role will exist or at least evolve. So it leads into the next question of what are your thoughts about how the AE role is going to look at these larger companies?

Okay. Love it. I agree they still exist. 100%. I mean, I'm working with companies, can't name them, but there's like hundreds of SDRs. What's happening, it's why I think we're on the verge of a cataclysmic event, is that I'm starting to see a CAC inversion where it used to be that SDRs were more efficient than automated outbound automation for most cases, and you would stack both. For different use cases. Okay. What's happening now is that the CAC of the channel is lower with automation. And so the companies that rely on humans are going to be spending more, have less return on investment. That switch is happening right now. It's happening very fast, and that's going to create a lot of pressure to make the outbound channel— just forget like how it's done— the outbound channel more efficient. And so that's gonna be cataclysmic because like there's no way humans can compete with those fully automated platforms 3 months, 4 months, 5 months from now in the US at least. You could outsource it. That's a different story. Let's say like US-based SDRs, there's no way. There's no way except for like if there's a couple of industries and like sure, like generally speaking, while you're selling to like SaaS and whatnot, like there's no way on the SMB mid-market sector that you compete with those AI platforms. I just don't see it. All right. Which creates a good question of like, how do you train AE is getting to your second point, because generally the, you know, the way to go into the world was to start as an SDR and get trained there. Out of college, but there's no SDR role anymore. All right. So that's going to be an interesting challenge that's happening. How does the role evolve? I think it's, it's a good change for now, which is they're going to be focused more on the closing part of the process, not on all the chasing, getting people to join calls and be on calls like this is going to be back to back on calls, just the closing. They can have an AI summary of all the context where they space processing contact company came from, what's the need, what's the signal ready for them with? Like, if you think of like how Gong does it, like there's a prompt for the human. So the AI is prompting the human like, hey, like, say this on the call and it's going to work. This is our forecast of the best things to say, right? So that's what's happening right now.

Yeah, thanks, thanks. Good thoughts, appreciate it.

Sure thing.

A follow-up question to that, Ji, um, like, what is your best advice for someone who's an SDR right now?

Get into an AE role, get into ops, get out of the SDR role. Uh, it's never, it's never been a career, to be honest. SDRs, it's not, it's not a long-term career. No one's like a long-tenured SDR and a good SDR. Those two things don't exist. But what's gonna remain for a long time is people who are creative, build those workflows, build those playbooks. And so you can either become really creative, really good at creating all those workflows on the ops growth side of the world, or you can go and become really good at the closing and like working with humans. Just don't stay in the middle. Just don't try to like, you know, bang your keyboard like a monkey as fast as you can because you're not going to win that battle.

Amazing. Gee, you have so many amazing comments and feedback in the chat of people just really loving this presentation. Super insightful. Again, very timely. So thank you so much for spending some time with us in advance of Europa. Again, Gee's going to be with us next week in London, so please make sure to join us. Also some folks that he dropped, I'm just talking about as well, we have the CEO of Gong too will be on stage with us. So really great. And Ji, we'll see you next week in London and everyone else will see us next week for Workshop Wednesday, but you can watch all the sessions at Europa live on our YouTube. So please make sure that you're subscribed and then we'll see you back the week after Europa. So thanks everyone and we'll see you soon.

Thank you so much.