E61: GTM Strategy: 5 Insights from 500 B2B SaaS Companies (Jeremey Donovan, EVP Sales + CS at Insight Partners)

Jeremy Donovan is the EVP of Sales and CS at Insight Partners, where he advises over 500 B2B SaaS companies on their go-to-market strategy.

So even this week, I think I had 6 or 7 conversations with CROs about what's working, what's not.

Jeremy breaks down the counterintuitive trends he's seeing in the data.

One of the surprises in the survey was that some of the top-performing companies are actually expanding their outbound sales teams, which blows the mind because everything you read on LinkedIn says the opposite.

We dig into the build versus buy debate and why Jeremy thinks buy Buying software right now is risky business.

I think a lot of buyers are afraid that they're going to be the dope who buys the point solution tool only to have one of the foundation LLM companies just be able to do that capability pretty quickly.

We get into a ton of more interesting topics in this episode, and I ask Jeremy what he thinks the best revenue leaders in the world get right.

When there's a CRO vacancy in our portfolio, that's the number one thing I'm looking for is a—

Welcome to the Revenue Leadership Podcast. Podcast. Momentum is a powerful tool for turning sales and customer conversations into go-to-market intelligence. Using Gen AI, it extracts, analyzes, and automates customer intelligence across your GTM org. The best part, it's not a new platform for your team to adopt. It integrates seamlessly with your existing stack, or it can straight up replace your conversational intelligence tool. At Owner, we ripped out Gong and went all in on Momentum. It writes back to our Salesforce directly, Captures forecast and churn risk, autofills CRM fields, shares product signals, and tracks sentiment. Companies like Cursor, Zscaler, Ramp, and ElevenLabs use it every day. Check it out with a free trial at momentum.io. Today's guest is Jeremy Donovan, who is the EVP of sales and customer success at Insight Partners, where he advises around 500 B2B SaaS companies on GTM strategy and Uh, Jeremy's had an unusual career path. Started as a semiconductor engineer, Master of Science in, uh, Master's in data science, MBA from Booth, and then almost 20 years, uh, on the, in the analyst world at, uh, Gartner and GLG, um, where you were, uh, SVP of sales strategy at the end, uh, GLG, and then operator side. So almost 4 years at Salesloft across solutions engineering, rev ops, and eventually revenue strategy. And now in this role as an operating partner and advising a bunch of these companies and insights. So really excited to unpack a bunch of those learnings. And you have a really interesting vantage point as an analyst, as somebody who's connected to portcos, as somebody who's been an operator. And one of the things that is great timing is you published a survey recently that we're going to unpack. It was a survey of 150 CROs and some findings around where AI is actually working, where it's not, how folks should think about it next. And really looking forward to this and being a big fan of your content for a long time as well. So thanks for joining me.

The timing is good too, by the way, because independent of scheduling this, we've actually been interviewing a lot of the CROs who responded that they had adopted with high value a bunch of different AI use cases. So even this week, I think I had 6 or 7 conversations with CROs about what they're doing. So a lot of fresh off the presses information about what's, what's working, what's not, where people have adopted things or where people have even abandoned things.

Awesome. Okay. Well, I'm looking forward to digging into it all. So as I mentioned, you know, you're, you're advising a bunch of companies in the Insight portfolio. You've done this survey. You're, you're you're closely connected to these CROs. And so I'm curious, at a high level, what do you see as the gap between the companies and leaders that are getting value out of AI versus the ones that are coming up short today?

Yeah, well, I'll start by saying that I think it's early days, right? If you look back at the course of 2025, in early 2025, there was hype everywhere, right? And I pulled some, some other survey data that folks had produced that was— we're talking about astronomical improvements in win rates and productivity, right? And, and I think as we got further into 2025, we, we really landed in, uh, the trough of disillusionment, if you will, in the hype cycle, right? I, I saw a lot of stuff on LinkedIn and elsewhere where people were saying— and podcasts also, like that I listened to yours and tons of other ones where people were saying, ah, it's not really working in sales yet. And I think, like, I'll start by saying, you know, that, um, at least with AI adoption from top-performing company CROs of top-performing companies versus CROs of, like, you know, average to below-average performing companies, there's not actually a huge difference in which use cases they're adopting. And, um, in which order. So I would— that's a— I mean, maybe it's disappointing, but there's not a— there's not like a silver bullet that's obvious yet in terms of what they're doing. I would say that the ones who have been more successful, talking anecdotally from the individual conversations we've been having with CROs, is they are— they've developed a system to farm ideas from the bottom up, right? Because you had a lot of reps and other people who just got interested and excited and started testing things, and then they would pluck the best ones of those and then start to build a mechanism to syndicate that across the team. Another one is maybe stating the obvious, that a lot of companies realize they had to shore up their data foundation and AI, right? The garbage in, garbage out thing. If you don't have a clean data foundation, you're not going to get anywhere. I think the thing that is switching towards the end of 2025 and into 2026, and I'll pause, the thing that is switching is I think you're seeing a lot more top-down now versus bottoms-up. Like we tried to do a lot of things bottoms-up and there's only so much you can do. I gotta be careful with this one, right? Like I was gonna say there's only so much you can do in like the foundation LLMs, but every day what you can do gets more and more impressive. But when you get into agent capabilities and things that require orchestration across different systems, like if you want to get in your Snowflake or whatever, your, you know, you know, Databricks, Snowflake, whatever, like your instance. And by the way, we are investors in Databricks. I always have to disclose when we're a known investor in a company, but whatever your data warehouse is. Yeah, whatever your data warehouse is or any other systems, right? Like when that happens, at any company of that's of any scale, like security gets involved and you can't just point your, you know, random AE can't just, can't just point their agent at those systems and have it mine data out. So I, I, I, it's a hypothesis that like 2026 is going to be much more the year of top-down sales AI innovation because you need that systems access in order to get the real value out of it.

Yeah, I, I think there's a lot of truth there, and our approach has been almost purely top-down. Um, we have— we made like really aggressive early investments in AI across the business, and especially in GTM, but it's largely being run out of our data team, and now we've got a few applied AI folks within that organization for that exact reason, because you only get so far with chat. You can only get so far in a chat interface that has, you know, some limited MCP access to certain things. But, you know, if you really want to get value, well, you need to be able to combine— I mean, this is the context graph thing that everybody's talking about today, but you need to be able to pull in information from Salesforce and potentially Snowflake and then do external research and then be able to tie these things together or else the outputs are are just sort of thin. And one of the things that I wanted to talk, I wanted to ask you about is this idea of governance and orchestration. So it's a topic that we're talking about internally a lot. How have you seen, and this ties into your top-down hypothesis, how have you seen companies be able to give the power of these tools to a broad cross-section of people without significant like security security, privacy, governance, governance concerns? This is, this is like the topic du jour, I think.

I mean, it's— the answer is they don't get around it, right? I mean, I think there's— I deal with two types of companies, um, and maybe companies in two dimensions. So the, the companies I support are in the $0 to $100 million ARR range. We have other teams— the team that I'm on supports $0 to $100. We have other teams that do $100 $1 million ARR+ of our 550+ portfolio companies. So one dimension is like the size of the company, and absolutely, as the companies get bigger, and certainly as you get over, you know, $20, $25 million of ARR, stuff is pretty locked down. And then the other dimension is whether or not these companies are serving regulated industries. So one of the CEOs I was talking to the other day, you know, sells into financial services. And, um, you know, they— we have others that sell into healthcare, we have others that sell into, um, uh, you know, other regulated industries. Doesn't matter how big they are, they are locked down, you know, from the beginning. So I don't really— I, I think this is a— you need to include security and privacy as your, as your partner. The good news, and this is another best practice of, you know, like top-performing companies, and this gets at the top-down thing also, is that the, the best, fastest moving, most agile companies have appointed a human who has good organizational influence and authority to be the AI point person for internal AI optimization. It's really important to separate those things because, right, we're dealing with almost all of our portfolio companies are, are B2B SaaS. So we, we have to, have to be really careful when we talk about AI that we're not talking, you know, whether we are talking about AI in the product Or we are talking about AI for operational efficiency and, and, and growth, right? So in the case of, of, you know, these companies, they have a, an AI czar basically. And it's, it's a bit of the old chief of staff, you know, there's still chiefs of staff floating around, but we're seeing a lot more AI czars and those AI czars could have come out of RevOps. They could have come out of PMOs, project management organizations, right? They could have come out of finance. Like they're popping out of different places. They're just people who really they showed the passion early on and then they're helping to basically project manage these complex things including privacy and governance through, you know, through, through implementation and deployment.

Yeah. Anything else that you see in terms of AI adoption in the top performer versus average companies?

Right now, most people are like, it's not a, there's not, I would say like a huge difference. The, Some of the themes, I guess, that I'm seeing, you know, this— a lot of this stuff is starting at the top of the funnel and moving its way, its way down funnel. So I mean, that, that could be a thing that the, the companies who are more aggressive with this, right, are starting to not just do the top of funnel inbound lead engagement and outbound personalization and research and so on, but they're moving into those mid-funnel and bottom of funnel use cases like, right, figuring out risks and opportunities, using it for RFP responses, going way down funnel into customer success to do churn signal detection, right? So it might just be a matter of like, yeah, I was about to say more of the use cases, but I almost feel like doing fewer better right now is a better approach than trying to scattershot over a whole bunch of different use cases. Yeah, that's what I hadn't— I, that's what I hadn't explored in the data, but that there's a little bit of that, I think, is focus. Focus matters.

And, and this is maybe me talking my own book, but how, how much do you see, because I think this was the, your quote at the very end of that survey news article. How much do you think it's just like the discipline, like companies that are disciplined with execution and can project manage and deal with the details are winning versus, versus not considering this is such a like systems engineering problem.

Yeah. And that, well, that's a bigger theme, right? Like I'm often asked, hey, if you were to go back into an operating role and you couldn't do any research, right? You just had to get in there and you had to do, you know, you could do, we could run one play. You know, what, what would you do? And the thing I've observed in working with these companies over the course of the past 4 years, and then, you know, the prior or whatever, 25, 30 years of my career is everybody knows AI or not AI. Like we all know the playbook. I'm not going to tell you, no one's going to tell you the what or the why or even the how. That is, that you're not going to know. The best companies have insanely disciplined execution. So getting more specific, right? If I were to go into a company, what's like, you know, some of the first things I would do? One of them is just deal reviews. Like if I could do one thing, it would be incredibly disciplined weekly deal reviews using, you know, in the survey we asked what the most common sales and deal sales qualification and deal inspection framework is. It's MEDDIC or MEDPIC or whatever variation you want. Like especially if it's in an upper mid-market to enterprise sales context, the first thing I would do is, is like get rigorous deal inspections on MEDDIC, you know, number 2, incredibly disciplined pipeline generation, right? With time carved out to do that. You know, and you could kind of go number 3 is, is ensure that you're really going after the ICP that you think you're going after because, you know, you could waste a lot of time on the wrong ICP, right? So you can go down the list. But, but so that disciplined deal execute— sorry, that disciplined execution thing, what I do a lot of interviewing when there's a CRO vacancy in our portfolio. And that's the number one thing I'm always looking for is I want to understand their operating rhythm and how much they adhere to that. And I don't just ask them, I do backchannel reference checks and I go ask the teams that they worked on, what was it like to be on that team? It's because I really need to know what that discipline was. That's success and failure. So it's the same thing with AI. It's like if you just throw it out there and you're not kind of, you know, passionate about making sure that people are using it and you're not using it a lot of times yourself, especially with these smaller companies, it's not going to stick. Another one of the CROs I was just interviewing, um, like he personally was writing a lot of the prompts and, uh, you know, you could argue that that's, it's a smaller company, right? Like a larger company CRO, that may not be the best use of their time, but like in this smaller company, it was a good use of his time. And that he helps train and enable and set expectations that people are using this. And that reinforcement is really critical. That's part of the discipline.

I would argue it's a great use of time. Like, I do this myself, you know, like I'm, we're moving into a cloud enterprise plan for the express purpose, like my most Governance and security is a piece of it, but like my most important feature is centralized skill management. So I can create and update skills and not just have to, you know, copy paste my markdown or send my markdown file to somebody because I'm, I'm whenever I'm using my Claude skills, I'm using it. The output's not quite right. And then in that same message, I'll be like, oh, actually, like, can you update the skill to do this, that, and the other thing. And being able to do that and then have that now be centrally improved for everybody using that same skill, I think is really compelling because, you know, you think of the leverage AI gives you. We're in performance review season right now. Performance reviews can normally take, could take, you know, 20 hours for a manager. They got 8 reps. You want to do it right. And yet, you know, we have all of our one-on-one transcripts. You've got all of the notes that you've taken about your rep. You've got all of your coaching sessions. And so with a really well-done prompt that you've iterated on, you know, that can save, I think it'll save 75% of the time that it takes to do all this. I still want, you know, the managers to like hand edit and craft and make sure that it's really personal and tight. But, but like, there's a bunch of grunt work going and reviewing all your notes and remembering a bunch of stuff and sort of synthesizing the themes. And so it was my BDR director who created the prompt that then we shared more broadly. And like that, I think is a great use of time. Yeah. It's, I mean, I, yeah.

Yeah. I mean, as companies get bigger, I think it's just right. Like the CRO might have a right-hand person who they trust, you know, to do that stuff. I was going to yes and. What you were just saying though, which is I would love for AI to be the death of the annual performance review. And instead, right? Because the AI can give you real-time continuous feedback, which is what's most valuable, right? I mean, every, right? Tons of research that practice in a feedback-rich environment is what matters. And the more, you know, the more immediate real-time that feedback is, the better. So, you know, there's an increasing number, you know, I think the first wave of tools is kind of like, it'll post-process your call, right? Like there's a ton of conversation intelligence solutions that'll do that. And then, you know, maybe we're moving into a world where you get real-time guided selling. You know, we've had that as, as a promise for quite a long time. It's almost like augmented reality, right? That keeps coming, coming and going. But what if you're on a, you're on a call and you've got a little side panel I know Copilot is a little bit charged term these days, but whatever, an agent helper, whatever terminology you want to use. And if maybe it's not flooding you with information because that's going to be very, very distracting, but maybe it's just giving you like just the nugget you need and imagine it's subtle that it understands if you maybe said, if you misspoke on something, right? Or if you were pausing and hemming and hawing a little bit trying to remember something, or you get asked a detailed technical question that you would need your sales engineer there for, but you don't want to delay the sales cycle. Like, we're gonna get there, right? I mean, just like, you know, those of us old enough kind of remember the days of, you know, everything— seeing you, you were really limited in compute, in storage, in communication networking bandwidth, right? And then all of a sudden, compute seemed effectively free, storage seemed effectively free. So that, that's As I think about AI, that's what I think about too is we will advance upon a world in not too long from now when the memory context window is effectively unlimited, when the use of the AI is effectively costless. Funny side story from just yesterday. So One of the things that we do is we, in the diligence process, right, we segment things and my team and I will conduct a sales and customer success diligence on companies that we're about to invest in. And I was, we were doing one and we re-record the calls and I wanted the transcript and our call recorder did not join properly. But fortunately we had a recording and we had the recording in video and audio, but not the transcript. Okay, so what do we do? One of us like took the entire video and fed it into Gemini and got a transcript. And I was shocked like, because that's a huge file and within 2 minutes, right, he got a transcript out of that. I used a service that I've been using forever that, you know, has an AI transcription. So my thing cost me $14.50. His thing. Wow. Yeah. Which is, you know, like in 2 minutes I was, I was, you know, whatever. And then his thing was free. So like, yeah, this is again, that you can take whatever, this is an hour long video. So I don't know how many gigabytes that is that you could just dump that into Gemini and like 2 minutes later get a transcript is pretty, I don't know. I'm amazed. Maybe I shouldn't be, but I'm amazed.

I find myself, I'm amazed every week. I was just slacking with, um, or texting with a buddy last night 'cause he was asking me cloud code questions and, uh, he was like, man, this is insane. I'm like, we are living in a truly insane time. And yeah, it's, it's crazy.

It is. Yeah. I don't, I mean, you could stop me if you, if, uh, if these stories are not relevant, but I have a buddy who built he, he, he like, yeah, using Claude code, he built kind of a RevOps Claude agent, right? With tons and tons of, of skills. And he realized that you could, you could basically build a virtual clone of yourself also using that. So on, on one of the weekends I was on a call with him and he onboarded me 'cause I, I still code and So he onboarded me with his like kind of Claude Code package of stuff. Yeah. And then I down, you know, I pointed that at all my podcasts and transcripts and I have a revenue playbook and all these different things and like started asking it questions and it's kind of a virtual me, right? Yeah. So yeah, pretty incredible. And then the other story just from this last week about how powerful this stuff is, is we get a lot of questions from portfolio companies when they're, when they're doing planning, right? So they want to figure out, okay, how much of my, of my budget should go towards sales and customer success and marketing and R&D and this and that, right? So basically an income statement based on your rule of 40, your, you know, your growth rate, your efficiency. And so I wrote a little, you know, I did some, some data analysis and then I turned that into a calculator. And I, I did it first in PHP, very old school, But, you know, one of, I just kind of had that available and then I said, okay, I'll convert this to JavaScript, right? So I just dumped it into JavaScript and then lo and behold, because of our systems, this is an internal tool at this point, because of our systems, I can't run PHP or JavaScript. It needed to be in a Microsoft Power App. And I'm like, oh, I don't know how to do a Microsoft Power App. So I just did the same thing. So, so I met with our engineering team and I said, okay, you know, like we're gonna send this out to the, to our, you know, outsourced coding people. It'll be about a month. I'm like, hold on, hold on. And I just put it into, into, I was using Cursor in agentic mode and I just said, hey, build a Microsoft Power App out of this. And again, whatever, 60 seconds later I was done. So yeah, it's crazy. It blows your, it just blows your mind.

My VP of RevOps built a similar type of agent that we call Cerebro. And it's hooked into all of our stuff and you can just like point it at a Slack thread where it pulls out the business requirements, highlights what questions still need to be answered, writes the PRD, can actually go and build that thing in our sandbox environment. And then you could go like do the testing and deploy it. It's, we actually think it's going to like completely transform how we do RevOps. And allow the RevOps team to do much more strategic, interesting, forward-looking stuff because this can just chew through the basics.

But you have to—

Go ahead.

There's a subtext of something that you're saying, by the way, which is like, I described, I told you a bunch of stories of like, you know, where whatever a buddy of mine built or I built, same thing for you, right? Like someone at the company or a buddy or you built. And that was also a theme that came out of this, which is there's a lot more built than I've seen in past waves of technology. We all kind of know it. I don't think I'm telling you anything you don't know, but yeah, the skew towards build is pretty surprising. I mean, one of the spaces where we've seen a lot of like mid-end funnel use cases is in RFP response. And right, there were a whole bunch of RFP response companies in the pre- Gen AI days, and they have morphed, right? And, and they have strong solutions, but it's now become— I would say it's 50/50 now of people using, you know, packaged RFP software, response software, and, and then the other 50% just building it, you know, using— by uploading, you know, they, they'll point it at their, at their knowledge base and they'll point it at their prior RFP responses and they'll point it at their security guidelines and this and that, you know, and then boom, they're done. So, yeah.

So I want to, there's a few threads I want to pick up. I want to get back to buy versus build. That's a really interesting one. But you were just talking about this idea of like augmentation and it's in the survey as well. Like people are seeing success, people are seeing success augmenting reps versus, you know, this promise of replacement. And even when it goes to coming back to that conversation around performance reviews, like I don't really want to pull a person out of that. I don't want to pull the manager out of that. I actually want to think about what is the most elegant human in the loop so that the output is as human-driven as possible with, you know, the craft and taste of the sales leader embedded in it. But like, you know, you've got all the advantage and scale of AI. So maybe just talk about what you saw in the survey results about like people being successful with augmentation versus automation thus far.

Yeah, I mean, it definitely skews at this point towards, towards augmentation. And it's the use cases that, that you would expect, right? It's at the top of the funnel. It's research, it's personalization. And take outbound prospecting for, you know, for a minute, right? So one of the surprises in the survey was that like some of the top performing companies are actually expanding their outbound sales teams, which, which blows the mind because everything you read on LinkedIn says the opposite, right? And I was— that was mind-blowing to me. And the reason— I think there's a reason for that, which is this whole correlation-causation mix-up, which is I think it's because they're top-performing companies that they are getting higher response rates to their outbound prospecting. So I don't think it's like the AI necessarily. I think it's that they've got great product-market fit. And then once they've got great product market fit, step on the gas, right? Like no question about that.

But it can also be this Jevons paradox of sales where, you know, if you're, if AI makes every BDR twice as effective, the decision for me as a CRO isn't like, okay, great. I'm going to get the same number with half the reps. No, I'm going to say, okay, let's keep the same budgeted reps and get 2x 2x the funnel, the economics of investing in outbound on a relative basis to paid marketing or a bunch of these other channels that are quite saturated and it's been well discussed, rational actors look at that and go, oh, let's allocate more budget here.

Yeah. I mean, yeah, yeah. I mean, my expectation is Well, let's stay on outbound and let's transition from it. So like staying on outbound, right, is, is, is probably been in the SDR game for, I don't know, 15 plus years, 20 years now. And, you know, it was really shockingly effective early on and it really is pretty, pretty hard right now. If, especially if you're, you don't have obvious product market fit and a clear, and a clear brand. Um, so what does that mean? It means that like the, the, the, what I've seen is basically that the efficient threshold for where outbound SDRing outbound prospecting in general works. It's just moving up dollar amount-wise, right? So these days you probably get about, um, you know, 6 qualified ops per SDR. Actually, not probably, we just benchmarked this. 6 qualified ops on average per SDR per month. And if that's what you generate— this is outbound, like a pure outbound SDR can do 6, 5, 6, you know, like it's not much more than that. There are exceptions, right? If you have a great product market fit, There's exceptions.

Yeah.

But like if that's the case and, you know, you assume a, you know, a certain, you know, a certain win rate of whatever, 25% on that, which is probably still pretty optimistic even, even for a qualified op. And, you know, then what you find is it's the threshold is around $40K or $50K ACV for it to make sense for outbound prospecting. That threshold just keeps, you know, as the effectiveness drops and drops, that keeps marching marching up. The other comment about this outbound thing is, is also benchmarking. Like, something like 70-ish percent of all meetings get booked by a phone. Um, so you're still doing the multi-touch thing, right? Phone, email, social, whatever. But most of the ops are getting booked by a phone. So that's one of the reasons why it's also an augmentation, pulling this all together. The whole thread together is like, that's why it's one of the augmentation things, because it is currently not legal in the US and in Europe and probably elsewhere. I'm not sure what the regulations are in Asia-Pac. It is not legal to call somebody with an AI if they have not opted into receiving the call. You can do it if they're a customer, but, and if they've otherwise opted in, but cold outbound via AI is a no-no. You can't, like, I think it's like, I can't remember the exact dollar amount, but it's significant fine per call. Huge fines. It's like $5,000 or something per call.

So per call.

Yeah.

Yeah.

Yeah. Yeah. Per call.

And I don't see that changing. We, it would just be, I mean, your phone would be now useless if you said AI calling is, is allowed because every, every smart company will say, great, well, I can have a million calls be made every single month to every, every, you know, every part of our ICP.

We'll do the same thing to the phone that they did to email, right? Like we'll just destroy it. Yeah.

And email is, yeah, you, you, email is so bad now that, you know, I use Superhuman for my email and it's great at screening, but I miss a lot. Like Superhuman, the, the, the bar has to be so high that I miss a lot of emails and it's just because my, my inbox is torched.

Yeah, I think it's life. I think it's life. And then, then people find other channels to, you know, to try to get to you if it's important.

Yeah. What has that 70%— do you know how that has trended over time? Is 70% higher than it was 5 years ago?

It has to be. I don't know the answer, but it has to be.

I would think so.

Yeah. Just because like back in the day, right? I mean, going back to the way, way back evolution of this, if you put someone's— I remember the first time I saw my first name in a subject line, like I read that email. This is a long, long time ago because there was no dynamic tags at that point. So the, some human had to type my name in. And then once I realized that dynamic tags came along and I, the collect and also the collective universe of people, then we said, oh, delete. Right. And then, and then, and then the next wave that came around was a little video in your email. And when I got the first video in the email saying like, hey, Jeremy, blah, blah, blah, blah, you know, like, and holding the little whiteboard. In front of them, same deal, right? It's like, wow, they took time to do that. This is incredible. And then AI could insert their first name and whatever in there.

Yeah.

Yeah. Yeah. And then it's like, so I think, right? I mean, there's nothing, what is that book? Cialdini, Influence, I think.

Influence. Yeah.

Yeah. So it's like, this is not a new thing. It's we will, there's reciprocity, like we will reciprocate when another human puts effort into engaging us in a thoughtful, valuable way. And the reverse is also true is like, if you phone it in and just like use AI to try to contact me, I'm going to be upset, right? Like, not only am I not going to respond, I'm going to like I'm gonna flag you as spam and, and hope that your emails never come through and your calls never come through. So yeah, this effortful thing really matters. I mean, you're probably, you probably get this too, is, um, I don't know how many LinkedIn connections you have. It's probably like 200,000 or something crazy.

40, but, uh, yeah.

Yeah. So, but like, okay, you have 40,000 LinkedIn connections and you probably get SDR calls all the time. How in the world do they not know who you are? Right? Like, that's unforgivable.

I know, I know.

It's unforgivable.

It is, it is really interesting. The, I come back to disciplined execution. It's just attention to detail. And, and, you know, there's some leader at that company that is not, not paying attention to the, the you know, the specifics of their system. It's like, oh yeah, just like give people a bunch of lists and let them call and say whatever. And, you know, we do a bunch of work to, we've got these machine learning models that score the leads and its propensity to buy plus the deal size. And then we enrich it with the information we want the BDR to use, you know, the closest customer that they can name drop and, you know, some analysis of that person's digital presence that the BDR can can leverage and all of those little things just like add up. We say shavings make a pile, shout out to Shannon Millar, who I got that from. And you do all of these little things and sort of it stacks up, but it's, you know, it's a grind and you need that attention to make it all work because I mean, I, the effectiveness of outbound has such a massive gulf. Like if you talk to the companies that are really doing this with a lot of precision, so like my team spends time with the Ramp team because that's a team that has a lot of diligence in terms of how they've built the system. And so we like to share notes. Their economics are awesome on BDR.

I mean, and share on their behalf for sure.

But then conversely, you go look at a bunch of companies and you're like, Okay, so you are losing money on every single BDR and you've still been doing it. You're at this for like multiple years. And so when I hear you say 6 qualified ops a month, you know, that's, to me, I look at that, I'm like, that's tough math. That's really—

Yeah, we have some, I mean, we've had portfolio companies where, you know, let's say they had a $20K ACV and when we moved from, we the collective we, right? Like moved from growth at all costs to efficient growth. Everyone started to look at that and, and you know, that, that losing money on every call didn't make sense. And we had a bunch of companies who, you know, if their ACV was too low, they just got rid of the inbound, sorry, the outbound SDR and then took that money and redeployed it into inbound channels and redeployed it into channel partners and like went from there because it was just never, if you have a $20K ACV, it's just not going to work. Like, it just— okay, you're gonna— you, you'd brand, you have to be so exceptional. I mean, but it's not just really good. I mean, like, what really good at it means, it's— I don't think it's the quality of the SDR. I think it's the qual— it would be the brand and, and the problem you're solving.

You know, is this a problem that people care to solve? Like, we're, we're only in 8 to 10K ACV, and our BDRs on a monthly basis generate $70K in closed one new revenue. So we have like a more than 10x, uh, cost to close one ratio. But, you know, we've dialed in the way—

sorry to interrupt, but it's velocity also, right? Like, I assume you, you have a one or two call kind of close, right? That, like, if you're Square or you guys or whatever, you know, like Boom, you got to velocity, then it works. But you have to do the math. Yeah.

Yeah. That, that is, this is something that I share with people. I think the, the magic of our business is, you know, $8,000 to $10,000 on 2 calls over 3 days is pretty attractive. You, you can, you can make that work. But, you know, like I, I, I see this when people ask for advice, but when founders ask for advice, like, oh yeah, my, my ACV is like $17,000. It's like a 45-day sales cycle. Or like a 90-day sales cycle.

I'm like, dude, and there's a 2-week POC and all the rest.

I was like, that does not work. I was like, you have to find a way to charge 3 times as much, or this is a PLG business and you gotta go like totally refactor. 'Cause the, it's, it's, this is sort of the Mark Roberge stuff from Science of Scaling. He talks about, you know, there's product market fit, then there's go-to-market fit. Like your business model needs to work with the right economics before you could scale. But oftentimes we skip, we, I see companies skip that. They're like, oh, you know, there is PMF, our retention is okay, and we can close some deals. But you know, like you do the, your economics are never going to be amazing in the beginning, but they have to be serviceable and you have to have a clear path to solving solving for a, you know, 3, then 4x LTV CAC. And I think like the point, one of the meta things I see in a post-AI world is like, you can't win as a point solution. There's too much competition. Every incumbent can spin up that feature. And so it's, you got to play the platform game because the sales cycle like necessitates it.

Yeah, yeah, yeah. Well, and we saw that, I mean, another aspect of that was in the survey, we asked the CROs about like how important tech stack consolidation was, which kind of speaks to the platform thing. And it is up, I mean, it's confirming, I think, some of these, that's what we wanted partly, that was just confirmed what we were hearing out there, but absolutely, they're eager to, you know, to consolidate. It's not just about money savings, it's also about just tool explosion. And yeah, and you know, a lot of times I think the reason you get shelfware is because you just forget that I need to log into the 20th system, right, and, and use this thing. And, and, you know, to the extent that you can, you can log into one thing and get, you know, 10 different pieces of core functionality, you know, that, that's key. And, you know, we're seeing it already in RevTech, right? Like, there's a bunch of companies out there who are, uh, who are consolidating and/or— and it could be a buy, you know, it could be a buy or build. Like, I can think of— I'd be really careful talking about companies, but I can think of one company, right, where they had mostly built. And, you know, they built out forecasting and they built out conversation intelligence and they built out, you know, all the pieces. And then I can think of a bunch who are buying their way and merging their way into this.

Yeah. So let's hop into that topic. It's certainly one I wanted to hit. So what are the conditions that would tell a revenue leader whether or not they should build or buy? And what are the relative strengths and weaknesses of those approaches?

At least on the AI side, if it were me, I mean, frankly, I'm probably going to build at this point. Yeah. The, the, I, I, I'll, I'll tell you why I, I think I see more of the building, uh, on, on revtech related stuff right now. I think one of the reasons, well, first of all, having built a lot of stuff in the past, it's always faster and I'm being hyperbolic. It mo— almost all of the time it's faster and better to build because you're going to build exactly what you need and it's just a lot less friction. And that's only gotten faster now with like agentic vibe coding. It's just so freakishly fast. So you're going to build something faster and better. And it's always been true independent of AI that a year later, most of that stuff stops working because nobody maintained it. And the world just moved on. Like we're going to see that too. We're going to see a lot of stuff that gets built and that we're going to see a lot of stuff fall by the wayside as things go. But that's one thing is like, it's why there's a preference I think for buy versus build. It's just so freakishly easy to build and you're going to get a better solution. The other thing I think is like, I think a lot of buyers are afraid that they're going to be the dope, you know, who buys the point solution tool only to have one of the foundation LLM companies, you know, just be able to do that capability pretty quickly. And we're seeing so much of that. I, we talked about the RFP use case earlier, right? Is like, you know, you know, I mean, there are instances of where the point solutions are great, but there are plenty of instances where you can build a good enough solution, you know, with a, with a DIY effort and, and some prompts. So I think there's, there's just that, like, I don't want to buy the op— the instantly obsolete technology. So, so to me, those are the Those are the two big things, but it does, I think as these things, this is what I was saying with the top-down thing also is as the integrations become much more complex and potentially as the functionality, you know, becomes more heterogeneous and I'll explain what I mean by that in a second, then the point solutions or platforms like the off-the-shelf stuff become more appealing. The heterogeneity piece is, we've been talking most of this conversation about text, right? Generative AI. But a lot of the revtech value is unlocked by predictive, traditional predictive machine learning. And, you know, there are some solutions out there that are starting to put those things together. And, and like, what, what's he— what do I mean by that? Okay, let's take customer success. Um, if you want to figure out like health, and this can be true for opportunities also, but let's take existing customers. If you want to figure out customer health, that's a predictive, generally a predictive AI problem. You might use some generative stuff to comb through transcripts and stuff like that, but like the algorithms that you use to figure out whether, you know, the probability that something's healthy or not is predictive. Yeah, as a traditional ML. All right, but now I'm going to take that and I'm going to say, okay, this particular account is not healthy. Now I gotta put in, now I gotta generate a workflow. So now we get into agentic and or, you know, guided workflow mode. And then I need to draft messages, right, on behalf of the CSMs, either that would just go out or that they would review by way of augmentation. Now we're in the generative world. So my, my point is like that, that, that's the heterogeneity is like the heterogeneity of the traditional ML. Predictive AI as well as the GenAI, like that stuff starts to get complicated. And that's another reason why you might, you know, you might start to skew towards platforms and endpoint solutions, just packaged software. Yeah.

And what is— because I don't think every company can build, like from what I've seen. And so what are, what are the signals that you as a business can build versus like what type of companies should just, you know, buy even with the trade-offs?

It's a, some of the, I don't think that I have a perfect answer here, but like the more regulated you are, you probably should buy because you want someone who's SOC 2 Type 2 compliant and has done like all this stuff. And if you try to build that it's going to be really, really difficult. So I think the more you skew towards regulation, I'm waffling in my mind about the bigger you are, the more likely you are to want to buy something and the more likely you are also, by the way, to want a platform as opposed to a point solution. That's definitely true. But yeah, like if you're a big company, you know, you want to use your capital for what you do. Right? So you want to use your capital for R&D of what you do instead of using it for, you know, the, the operational efficiency side. So I think that's an instance of where you're also probably better off buying. Um, little companies would, you know, whatever, they're going to be scrappy and do what they're going to do. But the bigger you get, I think they're much more likely to buy.

And, and do you think the technical prowess of a business matters? Like, you know, if you're— if you are, um, selling a technical solution into dev and engineering, or if you're Cursor or, uh, Ramp or Rippling that have like really, uh, impressive technical teams, is like that an indicator? Maybe build versus if you're like a more commercially driven organization? Because I feel like who's going to build this thing? It's not traditional RevOps most of the time, it's something else.

Yeah, but you'd have to carve off, and that's part of what I was saying earlier and did go through my mind. It's like, okay, you're a bleeding-edge tech company. Are you going to carve off a pod of engineers to build a revtech widget or are you just it like— and those engineers are, are so freakishly expensive right now. Yeah, like, I don't think so. I think you're gonna— I think you're gonna— if you're being economically rational about it, you're gonna buy. Yeah, because— yeah, yeah, it's an opportunity cost for you. I, I think the people who don't consider that opportunity cost, you know, they, they, they should be.

Yeah, I have friends who, who are leading go-to-market at some of the most sort of like technically impressive companies out there in the AI space. And I'm always surprised by how much they do buy. I was like, dude, you guys are XYZ company. Like, why wouldn't you just build it? Like, we can't get an ounce of engineering capacity because we're in a battle to the death with whoever. So I think that makes sense.

Yeah, yeah. I was at Salesloft for 4 years before I joined this and my little disclosure, Insight still has a stake in it, as do I personally. But I could not get engineering, you know, when I was like running RevOps, I could not get engineering resources. And I will credit our CEO Kyle Norton and our COO Rob Foreman. They were like incredibly thoughtful about capital allocation and our CFO Chad Gold, like incredibly thoughtful about capital allocation. And like I completely respected their no. So like I built as much as I could because I, I could code and, you know, I had at least one person on the team who, you know, had some coding background and, and, and, but I coded a ton. And, and then when I hit my limit, I, I bought. And then the other thing I did was I often prototyped, which this is another common thing, right? It's like you prototype. Internally, and then you go outside. So I did a lot where I would prototype, and the whole point was like, this thing should not go into production, even in our internal environment. I'm just, I'm just, we're going to run it for 6 months, and if it sticks, then we're going to go out and find a vendor who can do this better. And sometimes we didn't, and we just kept running that thing for, for years. But, uh, we probably shouldn't have. Like, we probably should have tried to, tried harder to go find a better Yeah, I, I sort of think the answer is you need the—

you need somebody in RevOps or your data or BizOps teams to go and say, I'm the AI guy now, I'm going to go learn to be sort of a vibe code master. And so it's not engineering capacity, but it's, you know, somebody who is learning to be a pseudo AI engineer now because, you know, you you can do enough. That's like a little bit of how we've dealt with it. We've hired two applied AI people who are not traditional, like, engineers by background, but have like done a bunch and built a bunch and are just interested in these problems. And it's working because same sort of thing, like, I think I would have a hard time getting engineering capacity away from our core product just because of how important it is.

A lot of us are probably in that boat, and I do think it's a superpower, right? Is a lot of organizations, right? Don't have somebody who can code. And I think it's trickier, definitely trickier for them. I would say most remote organizations I deal with don't have someone who, like they can, they're learning live coding for sure, but, uh, they don't, you know, they can't code at scale. Yeah. And not super scale, just like enough for the RevOps organization.

Yeah. Um, okay, I want to, I want to transition to one of these other, um, areas I wanted to pick out. So, uh, in that survey, uh, CROs expect between a 5 to 15% productivity lift. And so obviously very modest compared to the promises that like many of these vendors are touting. Um, why, why is the lift so low? And, and, and what should we actually expect?

Yeah, yeah, I think that's the right thing. And I'll tell you, like, I, I, in my mind, it's like I ask myself— we're often asked to quantify the impact in our portfolio companies, or we ask them to quantify the impact of AI, and I, I rail against that because, because it's, it's like, what is the ROI of, of sales enablement? We've been trying to quantify that for, you know, the, the first super formal sales training— or the, sorry, the modern era, the first super formal sales training was Xerox in 1968. And at least since then, we've been trying to quantify. And then you can go back to, you know, uh, Um, uh, you know, uh, who was it? It was, uh, not maybe Burroughs Company. Oh, NCR. It was National Cash Register, NCR, in the late 1800s. Like, you go way, way back. Anyway, um, like, I, I just— it's not quantifiable for operational efficiency. So I think part of this is like, I can't really say, but I can say, okay, I gave everyone ChatGPT or Gemini or Claude or whatever And I think we're all like 10% more efficient. So I think there's a little bit of that hand-wavy thing in there. The other piece is, is you think about a CRO and they're saying like, what am I going to commit to, to my CFO and my CEO and my board? And I think they want to be really, really conservative about like how much productivity lift, because if I say, okay, AI is going to make me 10% more productive, That means that instead of hiring, you know, 10 people incrementally, I'm going to hire 9.

Yeah.

And, and that, then I'm that much more likely to miss my, you know, to miss my target. I, one of the CROs I'm working with recently, like, on the, on the, the good news is he has, he knows exactly how much people will produce month by month of their ramp and into their 10-year experience. And he nails his forecast like nobody else. The bad news is the productivity is not, you know, he's not rule, they're not rule of 40 plus, you know, they're rule of 5 or something. Okay. A big reason they're rule of 5 is because they're so like, they're just not efficient. And you know, I think you need to get kind of But, but my reason for bringing that story up is like, he would not accept hiring 10% fewer people because he knows he'll miss his forecast by 10%. So anyway, that's— I think that's— there's like a conservative— the conservatism there. Um, I actually think that, that like 5 to 15%, whatever, I think that's about right. I, I personally wouldn't commit to something bigger, nor do I see in the data like a discontinuous trend in efficiency. So one of the cool things of having all these portfolio companies is that we can track like net new ARR per ramped sales equivalent. And there absolutely is a steady, or even ARR, you know, net new ARR per FTE across companies. And there absolutely is a steady, slow and steady productivity increase that has been happening. We don't know if that's AI or not. It could be other efficiency moves, right? Because you can't know the exact cause here.

Just coming out of the ZERP era, inefficiency is— Yeah. Yeah.

And a lot of people are doing a lot of things to be more efficient, but it, at least it's improving. Probably AI is a factor there, but there's no, there's no kink in the curve yet. So I would be reluctant also to commit to something significant until the kink in the curve really starts to happen. The kick in the curve is scary, by the way, because it means like machines are replacing humans. So from a social impact point of view, that's scary. Yeah.

I mean, unless going back to that Jevons paradox thing, unless reps getting more efficient means more companies are viable, the economics on these roles get better and people actually want as many or more of them. I think it's going to be a shifting more than anything. People are, our roles are going to shift around an organization. Like you see, you know, large swaths of customer support being compressed and eliminated. But oftentimes those roles are showing up in other places. Like the head of McKinsey was talking about this. They've reduced their like back office roles by 25% and increased their customer-facing strategy consulting roles by 25%. So it's like a big job dislocation, but like net-net, they're even but not growing.

So, right, so yeah, yeah, yeah, but not growing. And they're like, presumably revenue is increasing with flat headcount. So it's like, yeah, yeah, like Microsoft—

I can't remember the exact numbers, but Microsoft has added like some number of billions of revenue and they've frozen headcount for 4 or something years. Satya was on All In talking about this, which was, you know, I think an indicator of where people are going to go. And so what do you think it's going to take for some of the promise to be realized? Not like, you know, 10x-ing, but like, you know, a 30% lift. What do you think needs to happen for that to be unlocked?

I wish I had the crystal, the crystal ball, but I do think it's coming a little full circle on the conversation. I, I don't think it's going to happen through the bottoms-up tinkering use case approach. Yeah, I, I think the unlock is, is like to go, you know, downstream into deals, downstream into, um, customer success, uh, and, and expansion. And leverage the multi-systems approach as opposed to me throwing prompts into a GPT. I think that's the precondition. I don't know what it's going to— I mean, if I knew, I probably would go do that because I love tech, but I don't know. I do think it's a system solve though, as opposed to a prompting solve. Yeah.

Um, so on the, on this topic of, um, job displacement and, and job changes, um, everybody's saying now the SDR is going away, the BDR is going away. We haven't seen that at all yet. Um, what do you think changes about that role over the next, like, or what would you actually, more interesting question, what would you change about that role? To maximize leverage and grow?

I've heard of two things that I steal 99% of stuff. Like I don't want to take credit for stuff I've heard. So I've heard a couple things that I do agree with, both for SDRs and SEs, by the way. So it's worth chatting about both of those roles. For SDRs, Outside of this correlation causation thing around like if you're a hot company, then sure, you step on the gas and you hire more outbound SDRs. But otherwise, I think, you know, I think you can absorb the outbound SDR into the AE's job. And AI accelerates that, right? Because now you can have discipline around, you know, who you engage and how you engage them. It just, By necessity makes it easier to consolidate the job again. The other place it's happening for sure is in inbound SDRs. Like inbound SDRs are going away fast and they're going away fast because even independent of AI, like just regular lead to account routing stuff, why would you have an AE, uh, an SDR in the middle when you just want to get speed and just get people directly to AEs as fast as possible? And then in instances where, like, those are instances where you, you don't have a lot of tire kickers. In instances where you have a lot of tire kickers, you do need to do a little prequalification. And there again, because people have opted in, like, you can actually have the AI, you know, have the conversation with them in real time and qualify them in ever sophisticated ways before you send them to an AE. So that trend for sure is like absorbing into the AE. So I, I think independent again of the product market fit thing. I do think that the role is increasingly being absorbed back into AEs. And then the other one, and I wish I can give credit to who, where I heard it, it must have been on a podcast. Somebody was talking about the kind of an evolution of the sales engineer role that because of technology and augmentation of these things, AEs are going to assume more of the SE's responsibility in sort of the low to mid-end part of what SEs do. And then at the super high end, right, you've got this trend towards forward deployed engineers, towards FDEs, of what sales engineers like could otherwise be doing. So it does seem that AI is an interesting threat to the to the SE job as, as the low end goes and gets absorbed into AEs and the high end, you know, is moving towards FDEs. Yeah.

And again, this is that same like dislocation from low value to high value that, that I think we're seeing in a bunch of places because, you know, like the, the FDE, that's just a fancy word for like a solutions engineer, an onboarding engineer, whatever.

Exactly.

But when I think about how many tools we've bought where we like haven't felt like we've got the support we really wanted, and I just wanted somebody to sit with us and really like go line by line through something. It feels like if companies can save on the low end and you don't have SEs just like on a bunch of demos answering questions that the AEs really should know the answers to. And those, and those roles can be deep, more deeply embedded in customers. Oh, that's like net-net a huge win. And, and sure.

Yeah, I think it's, it's the other thing, by the way, you were talking, we asked me about like the unlock. The unlock is also a guy I keep saying like I heard this from somebody else, but like, um, so far what we've done is we've tried to optimize the jobs to be done within a, within the existing role structure. And within existing processes. And the unlock is to say, okay, maybe we don't have the roles right and maybe we need to like not just make the existing process better, but we have to develop some new processes that are, you know, more AI-first, AI-friendly, whatever you want to call it. That's, you know, is likely the other the other unlock here and there. That's real creativity and innovation, I think, to figure out what those new process designs are going to look like.

This was one of the things that was my favorite takeaway from my episode with Jordan Crawford last week. He was talking about how you need to look at a role as a basket of tasks and then unbundle it completely and say, what of these tasks are AI tasks and what of these tasks or things that are good for humans. And then once you unbundle the tasks across, you know, call it 3 roles, BDR, AE, solutions engineer, just to keep things easy, you unbundle a bunch of stuff. You say all of these things are really just like AI tasks and we need to go build those workflows. And I like saying workflows more than like agents or like AI employee, which I think is corny. But like once you, once you can have those done by those workflows, then, you know, what are the tasks that are there and what is the, What is the grouping that makes the most sense? I haven't heard somebody say the SDR role will get absorbed by AE. I think that's really interesting and true. You know, if you can have, you know, your copilot running with you on a demo. So when technical questions come up, you know, you can quickly hit a button and get an answer to that from your little, your ride-along. That's a great way to improve the buying experience, make sales reps more valuable and more strategic, and then have SEs do like a different set of things, like the deeply embedded forward deployed engineer. And so I think we're going to go through some version of that, this like unbundling.

My, yeah, my process thing also, I wanted to call back to something that you were kind of hinting at is, and the fact that like you might need different, governance structures and operating structures. So yeah, take, you know, we have a lot of portfolio companies using Clay. We're not invested in Clay, but we have tons of them using it for, you know, data enrichment and whatever. What we're hearing from them and I, you know, I don't, I don't know, I'm not close enough, but what we're hearing from them is, is like it's not a tool currently for individual contributor like AEs and SDRs. It's a tool that is some for someone in RevOps, a go-to-market engineer or whatever. And I mean, I know this from trying to consolidate certain things like lead research and account, you know, moving accounts around and so on. When you centralize that stuff, like it's, it's, there's, you, you create risk. So I mean, by way of story, you know, we, this is ages ago, but we centralized lead research and we had a guy who was like the manager of the lead research team and he had 5 or so librarians that we called— they were librarians at the time, but like lead researchers. And, you know, most of them were contractors, but basically he was in the critical path. And like, he got sick, not bad, you know, like he recovered, but he got sick for a week and a half and just knocked out of commission. And when he got knocked out of commission, are we, we went into like desperation mode because he was in the critical path. And I think these go to like, that's the thing with go-to-market engineers are also in the critical path. So like, I'd be really scared to consolidate with maybe just one person in that function. I would be more likely to, you know, use an agency in that, in that case, right? So that there's backup. And I don't have that. I don't have— I was just talking about portfolio to a portfolio company about this you know, this week is like, you kind of need an agency to, to, to cover your risk there. And, and that's the kind of thing where I think it's hard, right? It's, you gotta plan for that and figure that out. And that's complicated. And I think that's part of the value realization of this is, is, is these types of, of, uh, you know, centralizing some of these things is hard.

Yeah. And, and there's a good chance we're gonna see that across the business. Like, I, I have a feeling that you're going to see much, uh, much more vertical organizations where leaders are way more impactful than they once were. Where if you are an AI-native VP of sales and you can like much more tightly and broadly orchestrate your organization and be in more things You're doing less busy work. I think you're going to see that key man risk accelerate because you're going to have certain people that are really AI native just be so much more high leverage and important than ever. You're seeing this with engineers. The dispersion of engineers from, you know, people always used to say the cliché, you know, a normal engineer versus 10x engineer. Like that 10x engineer is a 20x engineer. And maybe even like a 50x engineer now. And that seems hyperbolic, but like we have examples in Owner where there's engineers that like just ship an ungodly amount of stuff still at a very high quality bar because they've been such great adopters. And yeah, you're more reliant on those individuals now than ever.

Yeah. And I think in REM ops too, right? I think you can have, like a RevOps person who knows how to vibe code, or, and, and to code, and, and to integrate systems, you know, like, and has the collaboration ability to work with security and governance, and, you know, understands how to make sure this stuff gets deployed and used. Like, that's all 10x RevOps stuff.

Yeah.

And, and, um, yeah, I would presume I don't know, I would presume hard to find, but maybe every RevOps person is like doing everything that they possibly can to get as smart as possible on this stuff as fast as possible.

I can tell you that is not what I'm seeing and not what I'm hearing from folks in the market.

Yeah, I mean, I'm getting old and I try to keep up, you know, like I'm using Cursor and, but one of my buddies who was walking me through cloud code, like, I realized I gotta spend a weekend at least. Yeah, just kicking, you know, hitting my head against a wall and, and mastering, mastering Claude code. Because, um, yeah, because I don't know, you, you mentioning, uh, Jordan Crawford, I think he talked about, what is it, like, uh, dangerous? There's some mode you can put Claude in where it's like you're authorizing every permission. Dangerous authorization or something like that. So I gotta learn all this stuff. Yeah.

But it's so fun. Like as soon as you, as soon as you get things like the basic infrastructure done, your MCP integrations are working enough and you've got like a skill built, then like it's just, it's just a blast. I just wanna spend every one of my non-scheduled hours doing this. I'm building like an AI chief of staff for myself right now. That like sort of works. There's little pieces. It's not autonomous enough yet. But, you know, like I just see, I can see a path to making myself twice as productive. And people ask me all the time, be like, how do you have time for all this stuff?

Podcast and this and that and the other.

And I'm like, it's just AI. Like I have work, anything I do, anything I do repeatedly has, you know, prompt templates and snippets and projects within whether depending on the model. And now with this AI chief of staff, I'm like, I bet I can go to 50% fewer meetings.

Yeah.

Because there's a lot of meetings I go to that I'm like, ah, like I said one or two helpful things here, but like I could have just read the outputs and then asked, I can just talk to the transcript. I can just read the output, ask a couple of questions to the transcript, message that person. Hey, you know, I noticed this and this, here's like a little tweak on it. And that's sometimes all you need to do as a senior leader. It's just like, just make sure nobody's making the wrong decision. Those are important changes. And then just like tweak people a little bit here and there. And that's a lot of the job. And I spend 20+ hours a week in big, like, org-wide meetings to do that. And I see a pretty clear path with this, like, AI chief of staff to be able to monitor all those things and hire— like, I'm uploading a bunch of my decision-making frameworks and mental models, and it learns from how I like give feedback in meetings to try to emulate me more and more to, to like, you know, give myself that extra leverage.

There's, there's a negative externality in this, which a lot of people talk about, which is the, the damage, potential damage it does on critical thinking. And I'll give you a real example of this. Like one of our, one of our portfolio companies was using a tool to, I can't remember if it was homegrown or if they licensed one of the ones that's out there, but they were using a tool that, you know, took call recording transcripts, extracted the MEDDIC framework out of it, and then pasted that into Salesforce. And here's the good news, like the good news, bad news. The good news is whatever they were doing, again, in-house versus tool, did a perfect job, near perfect job. The bad news is when the sales managers were doing deal reviews with the AEs, in the words of the CRO, he said, they said, she said actually it was one of our women CROs. She said the AEs were getting stupid. Like they couldn't, they didn't know their deals because previously they were really thinking hard about, you know, the metrics, the economic buyer, the decision criteria, like blah, blah, blah. Like they were really thinking hard about it and they just knew the deals much better and that they were on top of the gaps much better. So, you know, this CRO is ripping out that solution because it made her reps stupid. So that she actually like took an AI tool out of commission and, you know, there probably is an augmented better way, but that full replacement of that was not was not effective. So it just, yeah, just, I worry about that too for myself is I now ask myself on everything, how might AI, like how might I use AI to do this? And I push it further and further. Like I've coded my whole life and I used to use AI just to like refine a function or a couple lines of code. And now I just describe the entire program.

Yeah. Then it was autocomplete and then it was, you know, build a part of it. And now it's just like yap. And tell them what you want.

Yeah. I just write, I use Markdown to describe the entire program and I press go and then it works. But I worry, like maybe my critical thinking is working on the architecture side of the mind.

I was just about to say this. Yeah.

But I feel, I try to do something about this one by reading. I always read a lot, so like I just can keep reading long-form stuff. And then I used to write and I'm thinking I gotta start writing again because that's another way to just keep the critical thinking muscle strong.

Yeah, the— I, I do think the critical thinking is just in a different way. You're thinking, you know, engineers now spend some of their critical thinking time as product managers and as system, you know, as, as solutions architects, more than just like the engineer on their little piece. I feel like I'm, I'm using more critical thinking than ever because I'm like really sweating through each component and how it's going to work in this, like, building this system, um, to eventually like let me work on higher levels of abstraction. But it's very easy to go the other way and just like outsource your thinking to—

you know where I think there is high value is I remember, um, in the course of my career journey, like, I was a manager at a super young age And I messed that up royally. It's too long of a story to go into, like how I messed it up. And then I got put back as an individual contributor. I was like 23 years old or something. And then a few years later I had the opportunity to manage again. And I remember the hardest thing I was dealing with was like, because I'm so technical and I love the work, you know, so much, I was having a really hard time delegating. And it was only when I was so overloaded with work that I had to delegate that I had the breakthrough. And I think, I don't know if I can completely generalize that, but I think a lot of people who go from individual contributor to manager have to have that breakthrough somehow that they learn to, they basically learn to distribute the work and then be able to reassemble it. So the buildup here is I think that might be on the, going into the critical thinking side, like I think this may give a lot of people a lot of opportunities to build that skill sooner because if you start having to think about how to architect job distribution to a set of AIs, that's just as complicated, right? If not more complicated than doing it with humans. So that could be a, I'm trying to find the silver lining. Like, that could be a really good silver lining to the critical thinking pieces, is that we learn at a much earlier age and at a much more sophisticated level how to, how to like systems think. Yeah, about, about this.

Yeah, it's, you know, it's the same evolution that engineering went through, that computer and computer, uh, engineering went through. Like, you used to have to actually like build things on the chip or actually choose the knobs and tubes or the punch card outline. And then, okay, now somebody else is going to program the chip and you just need to program at this layer of abstraction. And now it's a different layer and a different layer. And so, because open source, you can just get a bunch of the answers and you're just now building on top of that. And somebody, people would say, Well, that makes you a lazy software engineer. Well, but it doesn't. We're just doing a different thing. And now it's about managing agents and agent swarms and like all these other things. I mean, I'm an optimist at my heart.

Yeah.

I want to ask a couple quick hitters and then I can get you out of here on time. What's the one thing that you think the market believes about AI in sales that you think is wrong?

I, right now, I think it's that it's having a discontinuous impact on productivity, right? Like it's not yet having a discontinuous impact on productivity and we haven't figured it out yet.

And what have you changed your mind on in the last 18 months in AI?

I think the mind change for me was I felt, at least from my personal usage or with my work usage, um, I felt like 18 months ago or whatever, yeah, I forget what your exact time frame was, but I felt like AI gave me really good answers on things that I didn't know well. But it wasn't when it was on something that I felt like I had deep subject matter expertise, I felt like it didn't give me a great answer. And I've noticed since the last generation of models, AI is giving me answers that are as good as, if not better than things I would come up with myself, which is both incredibly satisfying and incredibly nerve-wracking and just terrifying. But yeah, so my mind change is like the quality of the output, I think, in, in even incredibly esoteric questions is, is beyond what I would have thought it was capable of doing.

Yeah. Um, all right. So now we're going to get into our quickfire. I'm curious to hear your answer on this one. What do you think separates a good CRO from like a truly great one?

Discipline to execution. That's it. Nothing else.

Love it. Um, what's the most common advice you give to like first-time new VPs of sales or new like heads of sales?

Um, I, I give them advice on hiring that like be incredibly thoughtful about the, the hiring process. Um, and the dis— and, and the way that you structure you're hiring, because if you hire the wrong people, you will torpedo the business.

Yeah, yeah.

I've— I—

you and I have always thought about this in a similar manner. And when I read, um, Leading Sales Development, I was like, oh, there's some good little— there's some good extra nuggets. So I, I pulled in some of that stuff into our hiring system. It's— that's actually one of the books I recommend the most for people managing BDR teams. I think it's one of the single best books out there on building BDR teams because it is like such a systems engineering-centric version.

I appreciate it. Yeah, like if I do write another book, I'm going to take a similar approach. Um, I really like— I've studied sales hiring for so long and I still don't think I cracked the code, you know. But yeah, definitely IQ is a huge— I, I, I've definitely come to realize like assessing IQ is a humongous piece of that. And then I think the other unlock is, and we saw this actually going back to the CRO survey, that, you know, the number one criteria that CROs look for is a track record of success. And the only way you're really going to be able to assess the track record of success is through a backchannel reference check or having worked with that person before. So yeah, like, I, I think, yeah, you can't just hire, right? Like, I, and I've done a lot of data analysis. You just, there are certain no-nos of, of people, you know, that at least at this moment of their career, you shouldn't hire because you're just wasting time and money.

Yeah. We're going to do this big analysis of all of our hiring and try to do a regression between the scorecards we use in the interview process versus rescorecarding everybody when they're in role.

Yeah.

I'll let you know when this is done. I'm sure you'll be interested to see the output.

I've done it a bunch of times. It's a great exercise.

Yeah, it's so good and effective. What is the hardest lesson you've had to learn in your career?

I don't know if I saw that one coming, but the hardest lesson I've had to learn in my career, I think it's probably— I'll tell by way of very quick story. So one day, Kyle Norton, right, the the former CEO of Salesloft, came to me and he said, "Jeremy, I don't think you really want to manage people. I think you want to build shit." And it was painful when he said it because I had been thinking about this for years. And I went home and talked to my wife and I said, "Damn it, he's right." So I think what that gets at is just like, accept it. And different, different stages of life, you know, like I'm at a stage of life where that's where that is the case. Like I want to build stuff and, and I, there's a huge value in leading and managing people and architecting that, but it's just like not for me right now at this point in my life. And yeah, that's, that was hard. That was hard.

Cool. What's the best thing you've read in the last 12 months?

I constantly reread A River Runs Through It. Oh. I'm an avid fly angler. I constantly reread The Qualified Sales Leader, John McMahon's book. I constantly reread a book called Art and Fear, which is just a book about art. And even though I'm not an artist, it's still like I'm a creator. So yeah, I reread Anna Karenina. Like, yeah, there's a whole bunch of books I reread. So yeah. I read, I'm like a book a week or a book every 2-week person. So it's sort of like I read a lot of, a lot of stuff, but it's the stuff I reread. That's the best.

Very cool. Jeremy, man, this was awesome. It very much lived up to the hype I had in my head. I knew you'd have a ton of really insightful stuff to share and certainly you delivered. So I appreciate you doing this. Awesome. And yeah, well, I would love to stay connected on other stuff moving forward as well.

Always, always, always.

All right, man. Thank you. Thank you for listening to the Revenue Leadership Podcast. If you enjoyed it, don't forget to subscribe, and you can find a link in the show notes. And be sure to leave a 5-star review, share it with your network, and please join me next Wednesday for another great conversation.