The Science of Scaling & AI - Mark Roberge

Proven Podcast - Charles Schwartz

In this episode of The Proven Podcast, host Charles Schwartz interviews Mark Roberge, former HubSpot CRO, Harvard Business School professor,

Key takeaways

• AI is accelerating productivity by increasing 'selling time'—the percentage of a sales rep's week spent with buyers—from an industry average of 25-30% to as high as 75% within the next few years.

Main topics

• AI-driven scaling in sales organizations
• The future of human-AI collaboration in go-to-market functions

Notable quotes

"If you're doing a new ERP assessment for a large global organization, I just think AI today can do a much deeper and more accurate assessment... without being contaminated by steak dinners and golf outings."

Conclusion

Mark Roberge emphasizes that AI is not just a tool for efficiency but a fundamental shift in how

Transcript preview

Speaker 1 (0:00) Welcome to The Proven Podcast, where I don't care what you think, only what you can prove. Normally, I go into a long list of who the person is and what's going on. Sure, I could talk about Mark exiting out of HubSpot, and the fact that he's run a fund, and the fact that he's a professor at Harvard, and the fact that he's been rigorous and broken down the science of scaling in such a way that I've already changed how I operate my businesses after I got off the call with him. But that's not the exciting part of this episode. This episode is absolutely my favorite of the year because he breaks down not only where AI is going to be for the next 30 minutes, but where it's going to be in the next 30 days and more so the next 30 years. How it's a fundamental change of how we run our orgs, how it changes leadership, and how it changes scaling across the board. This is one of those episodes where you stop, take a breath, and then re-listen to it. That show starts now. All right, everybody, welcome back to the show. Mark, I'm excited to have you on, man. Speaker 2 (0:51) Thanks, Charles. Great to be able to jam with you here. Speaker 1 (0:54) So for the four or five people on this planet that don't actually know who you are, which is wild to me, can you give a little bit of heads up? Who are you? What have you done? Speaker 2 (1:01) Oh, come on. There's Speaker 1 (1:02) a lot of people that don't. Speaker 2 (1:04) Yeah, I guess at the foundation, I'm a tech entrepreneur. I've just found very like variant ways to express it. So in the first part of my career, probably about 15 years, I started three companies, three or four. The last one is the one I'm most known for HubSpot. I was the fourth employee there, um, joined as the first sales person, uh, and then B was the founding CRO through the IPO over a nine year period that ended back in like 2013. Um, in the, uh, uh, the goal to take a minor break. I was fortunately recruited to join the faculty full-time at Harvard Business School to teach sales to the MBA program, build and teach the sales program, which I still do 13 years later. So that's been a blast and has been quite instructive to some of the things that we're going to talk about here and the pattern recognition that comes from that post. And then eight years ago was approached by a gentleman at Bessemer. venture partners to start the first VC firm running back by the best sales and marketing leaders in tech, which we have done. It's called Stage 2 Capital. We have four funds that we have raised and deployed across 150 startups. And that also has been quite instructive to what we're about to talk about today. Speaker 1 (2:21) Yeah, so there's a lot there. There's HubSot, there's Harvard, there's Stage 4. There's a bunch of stuff you're doing that's moved around. You and I have started at kind of the same start line, but then you outran me within about three seconds. So thank you for that. Oh, come on, Speaker 2 (2:34) Dan. Speaker 1 (2:35) Just different expressions. Yeah. Speaker 2 (2:37) And actually I forgot Charles, which partially we're talking about today too, is I've written two books. One, the sales acceleration formula 12 years ago after HubSpot and more recently published the science of scaling, which we can, you know, weave in today. We'll talk a little bit about it. And I'm, I've donated the proceeds to all. to both of them to different causes. The most recent one with signs of scaling, 100 % is donated to mental health. So we might be able to take a little tangent on why Speaker 1 (3:02) there during our discussion. Let's go right into that tangent if you want to. Let's talk about why you're doing it. We'll Speaker 2 (3:08) do Speaker 1 (3:08) it toward the end. It's all good. Speaker 2 (3:09) Yeah, yeah, yeah. Speaker 1 (3:10) All right. So everyone talks about the scaling. and how important this is and all that. But most people are terrified of AI right now. And they don't understand that artificial, it doesn't mean artificial intelligence. It means always incorrect at the moment, but that's going to change in the next, you know, 12 to 36 months. How do you scale and leverage AI versus just being terrified of it? That's going to eat you alive. Speaker 2 (3:31) Oh yeah. Everyone's talking about that right now. I have to do a lot of thinking about that. It's a part of my job. It's also a part of my just natural curiosity. I love tech and the way it can change society for better and for worse, you know, lean into the better and mitigate the worst. And that's part of what we're trying to do. Let me give you a construct. Um, probably it's hard to know how long of a period this construct is. I think like when we had the shot, heard around the world in November of 2022 with the release of ChatGPT, the version that went viral. I mean, at the time, if you remember, people were predicting like no humans would have jobs in 18 months, you know, like, so I think like historically so far, I feel like. Silicon Valley has been probably a little too aggressive on their prediction on timing. Okay. So we have that backdrop. This is going to take longer than the techies are predicted. But it's going fast. I mean, every quarter, it's like, it's crazy to think back three months ago where we were and where we are today. And so. Speaker 2 (4:35) Let me give you sort of like a longer term view of four potential phases that are specific to go to market but can be extracted to other functions. Okay, so phase one is really the elimination of all work besides humans talking to humans. So you still have human sellers talking to human buyers, but a lot of the other work goes away. That's pretty much... measured by selling time. You know, selling time is an industry standard that measures the percentage of a seller's week that is spent with a buyer or prospect. And over the last few decades, best-in-class orgs have gotten that to like 25 or 30%, which sounds embarrassing and low, and it just kind of shows the upside that we have. And I do think even this year, some best-in-class adopters of AI in the sales org can get that to 75%. And that's going to be a profound improvement in productivity. Okay, so we'll circle back to that because that's the shorter term opportunity that I think you're asking about, Charles. Phase two, I do think at some point we will have AI agent sellers. We're starting to see some signs of that in the tech. I think it will start with like what we see today in product-led growth, very transactional, maybe SMB simple sales. I don't think it'll ever... get there in like the big million dollar deals, I think you'll still have humans involved. But I think we'll start to see a trend in that in phase two. Phase three will have AI agent buyers. If you're doing a new ERP assessment for a large global organization, I just think AI today can do a much deeper and more accurate assessment of the needs of the org than a 100-person global committee. Um, and can assess vendors without being, uh, contaminated by, um, steak dinners and golf outings and, you know, Taylor Swift tickets, whatever. Um, and so, so I think we'll, we'll have that in the, in the mix later on. Um, and then finally, if you want to talk Star Trek, um, I think the, the phase four is the, the functional boundaries within an org blur. Um, I, I think, you know, If you extract the organizational design back to first principles, the reason why we have a finance department and a marketing department or product department and engineering department, a sales department, is because of human limitation. You know, you don't see a lot of people who study finance and then go code and vice versa. And so we build these departments around our, you know, what we've studied and the experience we've built up. Which has advantages, but also disadvantages. Those boundaries create inefficiency. You know, finance would love to be closer to revenue. Product would love to be closer to customer support. You know, like there's alignment. I think in phase four, those will blur and you'll start to see organizations that almost operate more like GMs of business units. with that are very agent enabled very cross-functional um and so like whatever we can talk about yeah but let me circle back charles to your short-term question of like um you know what what does it look like in like the next year and you know i i show up to a lot of board meetings and people are like hey good news like we are so ai enabled in our sales org Like we are so AI advanced and I'm like prove it to me, dude. Like, how are we measuring this? And so that's what I'm kind of been working with the ecosystem on is how do you measure in 2026 how AI enabled your sales team is? And my working hypothesis now that I have a decent amount of conviction on is the two input measures are one, selling time, which we talked about. So first off, measure selling time, right? So like use AI to like... See how often your reps are either in meetings with people with a calendar integration or on Zooms with people with a Zoom integration. How are we going to do it? And try to get that to 75%. And then the other one is your rep to manager ratio, which historically, at least in tech, has been seven to one. And that varies if inside or outside, whatever. But like today's AI can coach reps better than humans, most human managers. And so you can push that to 15 to one if you've adequately enabled AI as a deal support and deal coacher. Okay. So the combination of selling time going from 25 to 75 % and the combination of the rep to manager ratio going from seven to one to 15 to one, I think at least will double productivity per rep. So if you had Marty and Jane producing 250K a quarter for the last three years, they'll be producing 500K at least, which will be profound. Now that's like pie in the sky. I'm happy to dig in Charles to very specific use cases and how it works, but I'll let you take it from there. Speaker 1 (9:52) Yeah. So there's a lot of different things there. I like how the silos or the orgs are starting to blur. So like marketing sales, and then you've got tech and you've got accounting. And I still love that HR still has its own silo on its own and no one still wants to talk to HR. It makes me happy. That's just me ripping on HR. From there, when we get into the AI side, There's a lot of people who are freaking out about not only the future, but what's happening right now. And before we even jumped on the call and started recording, we talked about how what's happening right now isn't what life's going to look like politically or economically in the next 30 years. And this idea of just, can we survive through the next garbage that we're dealing with right now, the influx of how things are changing right now, politically or economically. How do you see that also playing into it? Because everyone's talking so much about AI right now, not understanding that we have huge ways of how we operate across the world from data centers to politics to power to all those other things are changing pretty intensely. So how do we survive the next four to five years? Yeah, Speaker 2 (10:53) let's go there, Charles. I rarely get to on podcasts, but dude, you're a guy that we can do this with. I got, we, Speaker 1 (11:02) we have to take off. I got to take, I should probably take the vest off because I'm, Speaker 2 (11:07) when I'm about to do this clip, I am not Mark Roberge, managing director of stage two, professor at Harvard. Yeah. I am Mark Roberge, United States citizen and early. Okay. So this is just me guys. All right. And if you try to twist it, shame on you folks. Um, so yeah, I mean, it's crazy. How much effort, time, capital is going into building AI and not... Thinking through and how society can adapt. Correct. And that's troubling. And that's part of why I'm donating the proceeds of my book to mental health, because I do think that all of us in tech need to do more to balance that. Don't just build, but help society adapt. We can't delegate this to Washington or academia. Um, not because they're not qualified, but because they're just not close enough to it to really understand it to the level that you need to, to, to look ahead and help society adapt. Um, and. With each technical revolution, let's like we most recently lived through the internet one, we do come out the other side as a species evolved, but it doesn't come without its scars. Right. And, you know, we're still experiencing them from the internet wave, specifically in social media. And they're going to be worse if they aren't mitigated in AI. Speaker 1 (12:39) Yeah, I see. Speaker 1 (12:41) Yeah, when we go, we talk about social media, like, oh, we're going to be more and more connected. And then if you look at the patterns over the last couple of years, people who want to do early exits, which is a nice way of saying, you know, often themselves radically increased because loneliness increase and isolation increase and bullying increase and all of these things, because we're being fed dopamine hits that are coded very specifically based off human behavior to make us react a certain way. The positive for that for the companies are we're glued to these little screens. Because if I would have told you, because I'm old enough and 48, if I would have told you when I was 15, 16 years old, that I'm going to be staring at a screen for 10 hours a day, I'm like, you're out of your mind. Go outside. The graphics are better. But as someone who was a Microsoft trainer, I got to see it up front. I'm like, this is fundamentally changing how human beings exist, how we interact with each other, how we interact, how we make money, how we connect as human beings with each other. AI is just the amplifier of that. Where I used to say alcohol is an amplifier. If you're a jerk before you drink and I give you a bunch of alcohol, you're going to be a bigger jerk. If you're a goofball and I give you a bunch of alcohol, same thing, same thing with money. AI and tech is that force multiplier as well, where it just takes off and it scales it. And I do agree with you. It's evolving us at a rate that I don't believe we as a species can keep up with fast enough without some help. Speaker 2 (13:54) Yeah. And I think like, if you want to go there and this is an area where we can talk later more tactically about the principles of the book on like how to scale revenue, which I. Speaker 2 (14:06) I'm one of the most well-read, I think, in the world, if I could say that humbly, and I can speak to you in depth there. What I'm about to speak to, I'm not certainly at that level. I'm not an economist. I'm not like a politician. But I am a tech entrepreneur. I think I'm decent at vision. And I've also have been curious about this for many years because I don't want to bring a tech into the world that harms society. So I think it's important that we think about this. My personal opinion is you have to zoom way out to like almost a multi-thousand year view. I do think this movement around AI is not comparable to the transition to the internet. It's comparable to when we went from a nomadic species to an agricultural one. It's comparable to when we went from feudism to democracy. And if you look at the deep histories there, Um, there was a lot of skepticism as to whether that could even be had. And there were multiple generations of massive pain, um, in the transition. If you look at the history of moving from nomadic to agricultural, there was tremendous skepticism. It was like, where are you going to find enough animals to eat if you just stay in the same place? And how are you going to protect yourself from like attack? If you stay in the same place, everyone's going to know where you are. And they tried it and everybody went back to nomadic because not only did those two things happen, everyone was dying of disease because as a species, we'd never lived in close quarters with animals for so long that we didn't adapt. So people gave up on agricultural for three generations. But after living through a hybrid. they eventually figured out and got there. Now, when you look at, you know, feudism, we got kings and queens with like walls and we were protected. Now someone like Adam Smith comes around with the wealth of nations and is like, hey, if you make people free, if you give them free will, everyone will be better off. It's a thing called capitalism. And they're like, what are you talking about? You're going to let the peasants decide? You're going to create mass hysterical, like craziness. And yeah, United States was this wonderful experiment. Was it easy? No. In 1840, we were killing each other in a civil war. Do you think people thought democracy was working? Right? So when you go through these massive ships, if not mitigated with... massive intelligence, you go through generations of pain. Yes. And the pain that is coming upon us is we have a very difficult situation. And I'll speak specific to United States government. And some of this can be, you know, applied to other government entities. I believe in our long-term mission, which is human rights and freedom. I believe in that. And I think we... have been the world leader on that mission for centuries. So I want to see us continue to drive that mission. And it is questionable in the short recent years how well we've done at that. But like, that mission is going to be tricky because... Speaker 2 (17:40) It's our version of democracy and capitalism. Our economy is likely not compatible with AI, a