Field Notes #01 — Peak GTM Engineering: Are We Already Past It?

You guys are the power pack, so we got, we got us the bar like pretty high here. How I see this going, I mean, Jabara, Mitch, we, we chatted a little earlier about it. Let's just continue the conversations we're having, really. The spicier the better, I think. Speaking of spicy, did you guys see that the founder of YC tweeted something today and it was like, is GTM the new sales? Is that, is that like blowing up your, your LinkedIn feed?

It is.

Yeah. Paul Graham, right?

Yeah.

Yeah, so I think, I think he tweeted. LinkedIn's always a few days behind. I always recommend, I mean, I'm way more on Twitter, I'll be honest. But yeah, I saw that. He seems to think it's like a BS name, I think is basically his verdict. I mean, I agree, to be honest.

You agree?

You think GTM is a bad name for what we do?

Yeah. Yeah, I do. I think like the problem, like the problem for me with go-to-market engineering stems from the fact that it covers way too many things. And I think the reason it's covering so many things is because people are beginning to believe the ability to automate plus have agents inside Claude Code automatically equals infinite leverage, and therefore you can do all things. But the problem is, is that I don't know if you guys saw this come out last week, but they estimate that like 90% of corporate AI use is happening in sync, AKA a person is piloting the AI usage, not Async, which would be scheduled tasks and things happening in the background, right? More like an automated factory as opposed to a production line with people in it, right? And so the fundamental gap to me has like, if you reflect on the fact that there's all these things they expect go-to-market engineering to cover, could be customer success inside the CRM, outbound ads, content, you name it, run a podcast for us. It's that they think that AI is so intelligent and creative that it can take what you know and amplify across all of these vectors. But the reality is, is that any given person and decision-making power is a finite resource, and judgment and taste are finite resources down to the individual. You can only have so much skill and so much experience. Therefore, I think go-to-market is a silly name for covering so, so many things. And if we would just narrow the narrative of what it is, I think we would have a lot more clear hiring in go-to-market, agency hiring in go-to-market to desirable outcomes. But yeah, I think like trying to shove too many things in a box is always a bad idea.

I feel like we were just talking about this, Javiera, like yesterday.

Maybe this is contrarian. I don't know.

No, no, I agree. I mean, GTM as a role has expanded a lot and I was watching, I think a video, I forgot about which company the CRO was from, and she was mentioning that GTM has more than 10 roles under the title right now. And I kind of agree with that because it started with something else. And then with AI and new things coming up and having the capability to create a lot of other things, it's like there should have been more roles created, but because there were not, a lot of different tasks added to the roles that already existed. And GTM engineering being one of those roles that it absorbed a lot of different things. And as said, Mitchell, that if a person is kind of capable of maybe building an AI agent, it doesn't mean that they are going to be good at what GTM is at its core, because it's more about understanding business and strategy than being able to technically wire things together. So yeah, I kind of agree with this.

You think, you think the pinnacle of being a go-to-market engineer or GTME or GTM, whatever, it's still strategy rather than like technical chops though?

It is a part of it. And I mean, technical, being technical is good, but if you don't have the idea about strategy and you don't understand, like for example, you're really good at technical things based on how, like the times I've talked to you and the kind of projects that you, you showed me, whether in Cargo Cohorts or otherwise. What I think is a technical person is good at doing their technical job really well if they understand business as like To its core as well. Otherwise they won't be able to build things together. But yeah, maybe I'm wrong. Feel free to share your idea with me.

Yeah, I think, I think the one, the one gap I would put in your kind of take on things is that it's not necessarily like it's the problem I think really stems from like lack of clarity on what the desired outcomes are from the role. That's what I would say. And so where that, where that falls apart and you'll, you'll even see, I think Benjamin. Holley made a pretty long-form post about this, and I think he has some great takes on what it is to be inside an organization as a go-to-market engineer. Plus, I'm sure through many sales conversations with prospects, I'm sure you guys have heard all the horror stories imaginable from using Artisan, happy to throw them under the bus, unfortunately, or hiring internally and being like, we actually just couldn't measure what the hell he was supposed to be doing. Right. And so what Benjamin went on to say was, the next time I hold a go-to-market role, I will do everything in my power not to help non-technical people be uplifted within the organization because there's no way to measure the value of that output. And that stems from people thinking go-to-market engineer technical ability plus sales ability automatically equates leverage across all organizational vectors, which does not make sense. I think we are not at peak go-to-market engineering. And you said strategy, I would actually dial it back to judgment because I think strategy is a term we fall in love with. Because we're doing like a lot of marketing motions, but that's not everything that is covered in go-to-market engineering, unfortunately. Makes zero sense to me. Instead, people are shoving Claude code as a requirement into their job postings, but what they should be focused on is figuring out like, what is the outcome you want from a go-to-market engineer? And then who's the person with the best judgment I can pull into that role? Judgment over this narrowed priority domain, whether that is go-to-market strategy from a, you know, customer success point of view, whether it's inside the CRM and how we operate there, or external, going outbound, creating demand generation, creating demand capture vehicles, et cetera. It's all judgment, and people should be focused on training their judgment in the slice that they care about, because that's what will help them stand out to the right organization, whether you're an agency provider or an individual contributor that wants to work for a company.

Garrett, do you call yourself a GTME?

It's complicated. Legally, I'm a managing member and founder. But yeah, I think for simplicity's sake, I call myself a GTME, but it's probably just like a go-to-market advisor. I don't really, I don't really think like too much about like whether the title that we all have like coined for this set of skills is like worthwhile or like fitting of like the role. For me, it's It's like, can Garrett Mitchell and co. like figure out how to add like a ton of value in a bunch of different ways? And like, I don't really need like a title to do that. I just need to make sure that my agency and my abilities are like properly marketed and like people know that I'm the go-to guy. But I do think there's definitely like such breadth to the role right now. And a lot of that is like, there's people that I think historically wouldn't have been considered GTMEs per se, that because they are more confident with Claude Code now, they're, you know, hey, I'm a, I'm a GTME, I'm building cool stuff in Claude Code. And so I think it's a, it's an interesting, like, time. How's my audio coming through, by the way? Am I choppy or?

You're, you're great.

Okay.

Sound marvelous. Music to the ears.

I can't tell. I think, I think this, the Wi-Fi is like not great on my end, but, uh, but, uh, regardless, I think I think like with this year's, I'm planning on actually commissioning the next go-to-market engineering survey like a little bit earlier. And I want to be super dialed in on like what we actually want to discern and find out, both for folks looking to get into the category as well as for those that are hiring, but definitely trying to figure out like how are people compensated, how are they tracked, and, you know, what are the benchmarks they need to hit to be like a top tier GTME, because a lot of this can be, hey, I spend a bazillion dollars in tokens, or hey, I send a bazillion emails.

What actually is the metric that matters?

Is it return on spend? Is it just pipeline generated? Is it number of hours saved per sales rep? There's a lot of floating, I think, like cross-functional difficulty that we have in tracking this, this role or what people think this role is. And Ultimately makes it very difficult to discern like what, you know, versus us saying what is a top-tier account executive versus a bottom-tier account executive. And so there are different flavors of these people and their skill sets.

Completely agree. I think that points to peak go-to-market engineering having not arrived yet. And the reason for that is because if people don't understand the KPIs that they occupy, how can they relentlessly pursue peak, right? How can you peak at anything you don't fully grasp? For me, I think it needs to be broken up into more, more titles, or I think we're seeing like a lot of forward deployed engineers come out now as like the new emergent role. I think there's something to be said about a lot of crossover there, but I think the more clarity you have, the more, the less we'll see, like you said, I can build cool things, so I'm a go-to-market engineer. There's a huge, huge difference between cool and valuable. And valuable is, is only grounded in From the perspective of what the business's needs are, like what is valuable to a growth stage SaaS versus an equipment manufacturer or something like that is so different. Like if you've ever sold to like equipment manufacturers or logistics companies, like a simple change to like an ERP automation, like SyncUp is like, my God, that's like a million dollar value. Like just a simple little lift there, right? All comes down to judgment. In the world of AI, like people are like, humans are becoming less valuable. No, humans are becoming significantly more valuable, in my opinion. And the reason they're becoming significantly more valuable is because people want to, one, interface with humans, right? Not agents. And if you want to interface with humans, problem, human time is the most finite resource there is. You select where your time goes. Very meticulously. Number 2, if your time is occupied by AI slop, there's an unfair exchange of value there. You feel like you've been cheated out of your finite resource. And so human judgment and good human-to-human interaction is becoming the most valuable, which is why I would carry on to say that I think sales is actually the most valuable fundamental skill leading into a good go-to-market engineer because sales revolves around the KPI that matters the most, which is let's get some fucking deals over the line here. Right? That would be my take.

I, I mean, yeah, I think you're right. What we did is, I mean, what I do is I take like the big, big data and agentic RevOps projects now. And like, some of these deals come to me. And they're like, okay, we need to build a CRM from scratch. And I'm like, okay, what is your ERP? What does your data look like? And they're like, yeah, here, like, share the screen, sign the NDA, boom. But it's like, hours and hours and hours of like, of looking inside of their system, asking them questions and doing this stuff. And they're like, yeah, like, why aren't you just pitching us? I don't know what the hell the problem is. But I don't think like, if you're able to sell, I don't think if you're able to ask your questions correctly in the right order to hit the right notes, I don't think that you can provide a good solution to a core problem. And I think like, historically, I don't think we're at peak engineering. I think like, Javeria, when, when you won The Clay Cup, that was like peak engineering. But I think like that was one year ago. And I think like now everybody has Claude and everybody's— now everybody's a co-founder to Claude, which is really funny. With my projects, it's like I literally sit with them for like 6 or 7 hours just to figure out what the root problems are. And then I put the proposal together. If anybody that has ever asked me for a proposal to like their problem or like outbound CRM or agent builds or anything like this, I just say, I was like, well, I'm not in the business of Making proposals, dude. Like I'm in the business to do business, right? Like let me figure out what the problem is if I can help you. And I just learned this somewhere online like a long time ago. It's I think you guys are nailing it to say that sales is like sales is still important, but I think like questioning reality is still like what the judgment is still reserved for. Like you know, like I can come up with a strategy. Like I have skills in my repo. I'm sure you you all have different skills. and your Claude repos and stuff like this. But I think that's just different lenses. Like everybody has a council inside of their repo now, or at least I think you guys would. But I don't know, I'm like ranting. And I also think that people are not ready to buy from AI. I think AI is, I think selling a tool to AI to make its job easier is one thing and having it, you know, do deterministic judgment over like a webpage or like a product page or a service page to weigh the odds of like, yes, will this solution actually fit the problem that we're having. Yeah, that's fine. I don't think that for $1 million, $10 million deals nowadays where I don't think that they're gonna do fucking PLG, go on a website and click, yeah, let me pay you $1 million, send me the invoice. I don't, I don't think that happens. I mean, maybe Artisan can prove me wrong, but I don't, I don't think so.

Like, could you imagine SpaceX selecting their humidification system for, for their chip, new chip factory off of just like an AI agent being like, yes, boss. This is the one.

I think reasoning— well, I think reasoning stands to differ. I think you can use AI, and that's— I mean, I use it. I use reasoning models to be able to say, yes, like, based on my priority matrix of solving problem 1, 2, and 3— ah, damn, he's gone. Based on problem solving like 1, 2, and 3, this would be the top 3 models or, you know, tools that we'd recommend, right? Like Claude, Clay, Cardo, whatever. Or AC number one, AC number two, whatever, whatever.

I mean, I don't think that it takes away the judgment. Assistant, right?

I don't. I use it for more or less like edge case mapping.

Hmm. What I'm saying is like you're not gonna be like agent great go proceed with a contract for that humidification system.

No, no, no, no. Yeah. I mean, it's it's like you have like in the legal in the legal department like we're working with some like lawyers and things like that. Like they would. They are still required to check that those contracts are printed in the right way and that the signatures are valid and to QA everything. I think that in our sector, I think that a lot of people just throw spaghetti at the wall and just hope it sticks because they're blindly trusting the tool. I think that's a whole, I think that's a whole issue by itself.

Does Claude manage you or do you manage Claude? That's something I ask my team all the time. All the time.

Yeah. And that's why I also tend not to like, like courses. Like, I know a lot of people are like doing like YouTube courses where they're just like, and then it's like 3 months old. Like, we were talking about this, like, I want to start, I want to record a course. And then you said to me, it's like, yeah, well, once you record it, it's out of date, basically, because the industry moves so quick. So then that leaves, who do you trust in the industry, in our industry, to be able to say, yeah, this is working, this isn't working, what is it rooted in? And, you know, what data do we actually have historically to be able to say like, this outbound campaign works? Or what is, what is Jordan's model? Uh, the PVP matrix or whatever that we're talking about.

Permissionless value proposition.

Yeah. Like what works? Why does it work? When does it not work? Let's poke holes in it.

It's just that I think there are new things almost every day. And one thing that I have been noticing is that everyone is so much in a hurry to have all the tools in their tech stack somehow. And that is also a part of things. And you have to understand that what you really need and what you don't need, and if what is What I have right now is working for me, then I will make an informed decision if I really want to transition to something else. And as Mitchell was saying, that does Claude Code run you or do you run it is an important question because using it mindlessly without knowing what it is doing, it is going to mess up everything. So even most of the things that I personally think a lot of models do right, especially reasoning models, they do it right because we use our logic. I think still, if I'm not giving it the logic, the logic it comes up with is not always accurate or the best. So giving it what it needs to perform better is very important, and you need to be the in charge. You need to know your tech stack. You don't need to have everything. You don't need to use everything to get the results. It is going to be something different every day, but I think GTM engineering has become more of an experimentation role where you have to experiment with a lot of things, and then you let the data decide for you what's working, what's not working. But Yeah, that's what I feel about it.

How fast do you feel like you have to iterate on things though? Because for me, I feel like if I run an ABC test, I'm running it every 2 weeks. And if it doesn't stick, it doesn't stick. Then I'm, I'm gonna rotate the campaign out. I'm gonna rotate the angle. I'm gonna test different copy. I'm gonna test a different solution. I'm gonna see a different tokenization. You know, like, are you doing it quicker than 2 weeks? Are you doing it every 2 months? Like, when do you— because I have a problem Like, Mitchell, you're so much better at this than I am, but I have a problem with context management and not just context management. I have a problem with relevant recall of context in an efficient way that I haven't broken yet. I'm not great at that. I'm really good at systems architecture, but trying to be able to pull back the right information in the most efficient way is something that is still evolving. And I look to people like online to like guide me and I'm like, yeah, okay, okay. Like, let's look at, let's look at that progressive disclosure. Amazing. I don't, I don't know how, how fast you have to test nowadays.

For me, I don't know if this is directed towards me, trying to answer some questions on LinkedIn as well, because I know they're not coming through to the Riverside. For me, tests are just volume-based. So if we get a throughput of anywhere from 1,000 to 2,000 leads, then we're going to move on and start testing something new. And the one thing I will say there is it comes down more to total bookings than anything else. So as long as we're getting one meeting over the line every 1,500 contacts, we're gonna continue iterating on that. If we're not, we're gonna pivot as fast as possible. And I think when it comes to like context around tests, I mean, it's an interesting question. I guess it depends on if you're leading with like a true offer to like make a transformation, or if you're leading with, look, there's this very clear pain in the market. Usually for me, it's going to be around like a new regulation, to be honest, depending on the market. And I figured out the fix. Here it is, right? And you're just leading with some sort of value or you're leading with pain. If you're going that direction, you're probably going to have to give it a little bit more time. If you're going the offer direction, you can spin up and keep changing offers like every 48 hours and just pipe through different tests, right? And then once you find the one that's hitting like that minimum benchmark, building on it is usually pretty simple, right? Like I'm going to start with something that makes you money always. If not, it's going to be something that avoids, avoids the government repercussions, right? That would be number 2, like avoids massive, massive pain that will destroy your business, which is ultimately a money issue. And then it's filtering down to language. How can I make this more about them? You'll see this across my YouTube. It's a fairly similar system we run. Kind of over the last like 3 years. And we haven't changed that. And I would actually give pushback on this idea that things are evolving too fast. It just, it depends on where your focus area is. If you're, if you're running outbound, the only thing that's really changing is how do you get into the inbox, to be honest? Like the strategies that work, they work in cycles for sure. Right? Like things, things leave, things come back. Right? And it's really just about what is the pattern interrupt right now. But the simple things hold true. Like you want to position against the big bad thing. Like if you're targeting a professional, a local professional home service business, let's call it like a commercial landscaper. You're going to want to position against getting leads from marketplaces that you have to fistfight over. You're just going to want to position against the big bad thing, right? Um, and that holds true forever. You want to position against the big bad thing. And where do you find out what the big bad thing is right now? It's usually going to be Reddit. Tapping into Reddit data is super freaking valuable. And it's usually going to be third-party review locations like Facebook groups where people are going to leave reviews on their experience with things. Avoid like company reviews. Avoid, to be honest, like G2 reviews I think are like fairly manufactured at this point. I'm sure you guys have received messages from friends of businesses that are like, hey, could you just, just leave a review here about how you feel? Really about me, not my tool, right? So going to first party, finding out what's different. By the way, if anyone wants to do that, there's a great GitHub called Last 30 Days that'll make it easy as pie to access everything there is to know about anything over the last 30 days. It'll give you a snapshot. And if you just put in your context of research there, you're going to find out some pretty magical things for your campaigns. And if you do that, you're going to sound like the market. And if you have an offer that corrects that issue, You're going to get interest. Like, I really don't think it's that complicated when it comes down to the message. It's just how do you get to the message that's changing over time? And you could honestly be like Kellan and be very stiff to change and win really frickin hard because he focuses on judgment, taste, right? And continuing to nurture that judgment and taste. And ultimately, if he really needs to catch up at some point, it's a consult away with someone that's spent every day tirelessly staying on top of like the latest and greatest.

I think if you don't have a good offer, you're just dead in the water though. Well, especially for email.

That feels like a narrow thing to say because your, your offer comes out of the data, right? Like your offer is the data, to be honest. It depends on the business and their ability to adapt for sure. But, and there are businesses that are dead, but I think people think that like my, my offer is just like lead generation. Well, lead generation as an offer can be tailored in 1,000 different ways depending on who I'm speaking to. Like saying I can get you clients versus I can say, like we were talking about this morning, I could say I can get you H-1B visa holding contractors for your landscaping during the busy season coming up. It's like, fuck yeah, that's exactly what I need. I don't need clients, right? I don't need new hires. I need these guys. So I think people need to broaden their horizons on what an offer is and what like, it's not just like I had a video about this recently. It's not just The Hermosi equation, there's the added layer that surrounds it, which is who's it for? And that's what takes something from like demand capture to demand generation. Yes, but like a Hermosi offer is really just something that you stick on a landing page and you then have ads that are tailored to whatever audience that Meta will like serve them up to. And then they hit the landing page, right? And they've already shown interest. They're going to get captured by it. Doesn't need to be super particular. I will My setters will get you 10 to 20 leads in the next 3 weeks or your money back. And that offer's gonna crush through ads. But email's a different ballgame. It's one-to-one. You're putting your reputation on the line to a degree. You're putting your business's reputation on the line to a degree. And to stand out in a sea of 400 emails a day hitting these guys, you're gonna have to be a little bit more demand capture, especially to break through that trust barrier that exists because there's no face associated with it like there is with an ad. That's what I would say.

Yeah. I think that the companies and plays that I've seen like perform best or campaigns that perform best are like, A, the teams that are leaning into like marketing first and foremost. That was always like a big strength of Unify's, of Clay's, of like, if you can get your name out there, like there are like cross-dimensional benefits associated with that. So I think a lot of that comes through when you see like a trusted name in your inbox. Email is also like to some extent an awareness game. And then on top of that, like I think that staying current, like sure, I think right now there's definitely a perception that things are moving faster and I think things are moving faster, but the basics of things that you need to do like are not changing. You still need to make sure that variables are normalized, that you identify the primary current job of someone, that you warm inboxes appropriately, that you're tracking performance. That stuff like can be sped up potentially, but like you need to make sure that these systems work. And I think the onus is honestly more on individuals now than it was prior, where a lot of these things are delivered to you as software and in some type of commoditized platform. And now you have people that are saying it's actually on the user to actually interrogate Claude and interrogate these tools and say, hey, like, this is a problem that I know that it is— there's a non-zero chance that my system is likely to face. What are we doing about this? And you need to really, like, make sure that you uncover things that you might be missing otherwise. And I've, I've, like, written about this a bunch, but, like, silent failures is like a really, really big issue now for people just trying to do stuff in Claude Code. And so figuring out how to monitor all of that, especially as I've moved more and more of my workflows off of Clay and into Claude Code, has been challenging and like a really interesting problem. So I think that is definitely like figuring out how to make your brand more recognized and/or pick like the right angle, like that offer, but also For people that are, you know, claiming that they can just like build stuff. I think there's a lot of like post-college grads that are like, oh yeah, check out this website I built. And it's like, that's great, but do you actually have like a fundamental understanding of like each composable module that goes into like whatever equation you're trying to solve? And if you don't, then you're going to end up spending a billion dollars on some dog crap data vendor. You're going to end up emailing a bunch of people that left their job. You're going to end up not moving people's contacts from the company that they're currently at and associating them with the, the new company that they work at. So there's a lot of small pieces that I think will slip through the cracks, and that those are the areas honestly where like the true colors show. When I think about like how to evaluate people, even that like I want to work with or hire, is almost like a lot of this stuff can be answered with AI now. So I, some of my interview questions have just been like, what are the strangest problems you've run into while you're working with Clay? Like, tell me about like a classic problem where Clay's not actually like fitting for you. And if someone says like, hey, I wish I could like run multiple, multiple validation vendors instead of choosing one, or like I run into like a cell size limit, those are the keys to me typically. And I'm sure a lot of that be pulled off of Reddit and forums too. But like, if you actually talk to someone about it, you can tell that they've been deep enough in the weeds to— I don't know if objection handle is the right word here, but like objection handled their own workflows at a level where you can tell they've been in the weeds of it.

Yeah, that's— it's interesting that you, that you ask that type of interview question because when I interview people, when I mentor like prospective DTMEs or even people in the field and who I talk to, I just say like, what are you ambitious about? And what are you curious about? And like, what is, what is the one piece of work that you're really proud of? And what is the one piece of work that you're very technically proud of? And why are they different? Because for me, like I can be really happy building some architecture, but if it doesn't work technically, well, that's not great. But if I'm really ambitious about trying to solve that problem, well, I see no reason why the problem can't be solved. And I think there are a lot of People also on the, on the completely other end of this, Garrett, like I think that the people that you hire and that you talk to are different than the ones that I coach and mentor. I think there's a lot of people on the supply side of the market, not in this room that are chasing this like invisible $10,000 a month goal just because it looks shiny, which I think is also a big problem of like what's going on. And this is like the Clay, the great Clay-Claude debate. I don't think that they're mutually exclusive at this point in time though. I think there's a lot of people That we all know cannot actually do either one of them. That's a hot— there are, I think there's only like 10% of the market that actually knows what the fuck they're talking about.

There are people that—

I hope I'm one of them.

I can't even think of who you're talking about.

There are people that teach courses that I know don't know how to use Clay.

Ooh.

Yeah.

I see what you're saying. I think—

How does that make you feel, Garrett?

I think I might, I might have a hint here of what's—

It's like, I don't, I don't like tie my sense of pride or worth to other people being charlatans and pretending to be something that they're not. I think that this is objectively a very— I think it's difficult. It's hard for me to conceptualize what would've happened with my career if I had stayed in finance and how people who come from finance are able to evolve and do other things. But it's been really interesting to see this Clay learning curve. I remember I pretty much like berated everyone in GTM Cafe to like help me get from 60% to what I imagine now is close to like 90, 99, 95% proficiency. But I think, you know, I, I think that testimonials and like real results of like real people are what matter. And that is like what is gonna speak for people's reputation. And I think there's, you know, I, I think a lot of the courses in general, it's like you, you can't really cheat. Uh, progress. It reminds me of like the in-game, like, uh, I'm like a little bit of a video gamer. So like when you can go into like Fortnite or whatever and buy a bunch of things to like accelerate your, like, like you get the best weapons or whatever, it's like you kind of just need to like be living in the tools to actually be proficient with them and taking a class for like a bunch of made-up scenarios. Like that doesn't, that doesn't represent any real Like any real like deployment or client engagement, it's like, how do I, how do I circumvent this problem? What bug did I run into? And how do I need to think about it a different way? It's like much more iterative and like running around in circles. And it's, it's not always like as linear as, hey, today we're going to learn how to like run sources of people, companies, and signals. Then we're going to step into qualifying them. Then we're going to enrich them. It's like, it would be amazing if it were that simple, but if it were actually that simple, AI would fully do it. And I think we're moving in that direction, but I think there's plenty of reasons to, to say why that's further off than, than we might think.

I think it's very far off, but I'd like to ask you guys, and maybe this will be beneficial for people. You mentioned your proficiency in Clay moving up to 90, 95%. What do you think is the number one skill to be building right now as a go-to-market engineer, if you're interested in getting into this industry or this role, I should say?

My take would be to develop taste to have an idea how businesses are different and what you need to do. Because as Garrett was saying, it's not one template that you can apply on all clients. And whenever like we are starting a new project and we are talking to clients, I always think even if they're in the same industry, it's not the same. Every project is different. Every project requires separate layers of research. And then you figure out what should the intelligence layer be looking like for this client? What should the data layer be looking like for this client? Do we need to use Clay? Because for every client, Clay might not be the best solution. So do we need to use Clay for them? Do we need to use something else? What would our data providers be? You can't be using the same data provider for everyone across the world. Then for signals as well, because I think a lot of people talk about very few signals, but for every industry it's different. And the campaigns that I have seen performing better are the ones where we have this idea of What is the right signal? And if it shows this is the right time, then who is the right person to talk about this thing? And then what is the right message for that person? So yeah, it's not linear. I agree with that. And I think there are a lot of layers and it's different. So if anyone is willing to be a GTM engineer, they need to understand this. I usually recommend everyone to go ahead and read some case studies, to talk to GTM engineers who have been building these systems. for companies and see what they have been doing. I usually just connect with GTM engineers from different companies just to understand what they are doing, to learn from their experiences, because there is no book, no course that can teach you everything about GTM engineering. And every person that you meet in this space who has been doing great, they will have something unique that you can learn from them. So that should be there.

Yeah, it's, it's a, it's exposure, 100%. That's like, you just need to get as much exposure as possible.

Reps.

Every, every, yeah, every single month, like I update pretty much all of my documents to like account for like a new thing, whether it's like a new data tool that I think is better or like a new ordering of like how, like the ordering of things in the waterfall. Maybe it's like a new repo that Mitchell has posted in GTM Cafe that I need to like integrate in some capacity in some of my projects. I think I remember when I was like spending whatever, probably like 2 grand a month on like Clay for like each of my clients. And Kellen or someone was like, dude, like you gotta stop using Clay credits. Like you gotta figure out how to use Lead Magic. You gotta use Surfer. You gotta— and so it's, you know, I feel like that was a big evolution for me of like, okay, we figured out how to get the work done, but now let's like rein in the costs. And then, you know, now I'm, now I'm getting like even more creative with things just because I know that Claude can Help me do things in ways that, you know, otherwise I'd be doing with like AI within Clay. And I'm sure their workflows product will allow you to build code nodes and all that fun stuff. But I think it's exposure and reps is like the differentiator here. And if you're working on, you know, some like not real like deployments that are, that are active and in production, I think it's just like not, it's not going to like help you push the ball forward. And so for the people that are trying to jump in, it's like, Offer to help for free. And that's what I did for like a bunch of friends when I got started, even like post-Unify. I was like, I need to figure out how to use Clay if everyone's like raving about it. And then, I mean, some of my, some of my initial engagements were like very, very, very small fractions of like what I, what I think I can charge now. And that's because I wanted to like try to provide value and, and learn and do something of value to people that might not have access to it otherwise. And so Trying to, you know, you don't, you're never gonna come outta the gate being like the most well-paid person and just trying to give yourself leeway to do someone a favor price-wise and, and also give yourself the ability to, to learn a ton.

I'm gonna say probably the number one skill that I learned, or at least that I think people should learn about getting started with clay, or at least in the industry, is just like understanding limitations and how they apply to things in different manners. And I'm not even just talking about like, like how to use Clay. I'm thinking like more of like what Clay could not do 4 months ago. Like, okay, a few months ago, now it's 50,000 rows. Okay, now they've removed it, but now they've added CLI access and they've added all of these different things. That was a limitation before. Okay, if we want to get really fancy inside of things, how do we constrain models or like LLMs to not go and hallucinate? Let's learn about temperature controls. And I think like, This is fundamentally the hardest thing to learn because there's so much knowledge that surrounds everything. But it's the number one question that I ask to the companies that come to me and they say like, well, like, what about HubSpot? What about Salesforce? Well, what about something custom or like N8N or something? And I just tell them, I was like, well, yeah, you can use it, but you're going to be limited in this regard. Here's an alternative solution. Like here, weigh the pros and cons. And I think like that comes with reps. That comes with understanding, like being in the shit, digging through the shit, trying to figure out what actually does work versus just looking at the new shiny object that everybody is talking about on Reddit. Of course, like it is a shiny object because it's like a good tool, but it is definitely the number one skill that I look for when I start to judge people. It's like, are they able to understand and articulate through their reasoning on why something would not work, including why their own strategy would not work? Because I would be the first one to admit, be like, yo, I fucked up. I dropped the ball on this strategy. Like, we're going to have to pivot. Here's why I think it's going to be better. Here's why I think it might work. Here are my hypotheses. I think just having that self-awareness, even awareness to be able to ask the question of like, what is this tool good for? Okay. Everybody will say what it's good for, but not a lot of people will actually outright say what it's not great at. Again, like it's you versus the big bad evil, like the gospel, you know? But I also, on that note, I don't think that there's one solution to every problem. I think that's a bad way to look at things, especially in this market, like in any market, like with math. Yes, there's one correct answer, but in science there might, might be multiple ways to get there. This is engineering. This is not like, you know, throw spaghetti at the wall.

I think another thing that I would like to add to this is documenting whatever has been done for the GTM motion. Because I remember when I started back in 2023, anything that was being done for GTM or in terms of GTM Motion, nothing was as much documented as we can right now. Having the data in a way that it's accessible and where we can look at it and we can see this is what has been tested, it didn't work. So we don't need to repeat it again. And now we need to move forward with something else. Documenting the successes and failure in the right way is also very important. And you said somewhere about about context and trying to fetch the relevant parts of the context is also part of it where you doc— like we have to think in terms of like, okay, this is what I have tested. This is what didn't work. How do I edit in my context that when I build the next strategy, it is automatically available over there that I don't need to use it again. And now I need to think about something else. Or I mean, like, for example, after 6 months, now you know what worked and what didn't work. At this time, the data that you have about what worked, you just need to double down on that. And it's the easy part. And your percentage of experimentation should be more on doubling down on what worked than a small percentage on exploring new stuff, looking at maybe some other strategies, some other signals, or some other things that you want to build around it. But proper documentation of the GTM motion is important.

I think all of these hint at like reps. And I guess I think everyone's answer is pretty similar, to be honest, like judgment. I think what I was trying to go is a layer deeper and try to understand like what skill is the best skill to learn to begin training that judgment. I think Spencer, you were saying like there's just so much to know when it comes to, you know, AI temperatures. You have, you know, progressive disclosure that needs to be put in place. You need to have observability to audit the pathway that the agent took and understand why it's being an idiot and at any given moment. And then you have to have, you know, loops at the end of it. Right? You have to have an evaluation framework. You have to have an agent that proves that the work was actually done, and then it needs to go back to the agent to fix any crap that it's trying to output. This is awfully complex, and I, I, I don't think that you need to understand that depending on the lens of go-to-market engineer you are, right? I think it's super important for what you do, Spencer, and what you do, Juveria, and in many ways what you do, Garrett. But if we're just talking about like a go-to-market engineer that generates leads, I, I would bring it back to, I would even go one layer deeper, like below sales and just the ability to talk to people with some level of emotional intelligence and guide a conversation. Like I actually think that's layer 1 of go-to-market engineering. And the reason I would say that is because even when I was a total doorknob, to be honest, when it comes to technical ability, because I really dove into this with zero technical background. I think where I won was, was my ability to just interview clients effectively and extract their latent knowledge, like latent knowledge that they have over their market that AI is never going to have and make that a unique advantage. And the ability to guide those conversations, yes, but also pair what your client says with what like the latest and greatest research tells you. And if I were to bring this back to To like how I try to test people, I think a lot of the technical is very teachable these days as a result of Claude Code being able to build whatever you can articulate well enough, at least when it comes to automation, I would say. Definitely not copy. Definitely not. So for me, when I look to test people, it's far more exciting for me to be like, look, you're selling this, share your screen and show me where you're gonna find information that's gonna help you with your campaign without AI. Like literally just, and as you go, tell me exactly what you're thinking, right? Oh, you know, I, we're targeting this. So I know that, you know, Thomas.net is going to have a few businesses that I can go check out in that industry that are going to be super relevant, right? You know the directory, you're, you're in the game for sure. Reason their way through where to find, you know, some interesting points and then figure out like an angle, right? Just an angle. You don't have to do the copy. You don't have to do anything else. I think if you have that skill and the ability to talk to people and really differentiate like the garbage in the research from what's actually relevant, just by like, this is the hardest thing to train, like intuition, this judgment that we're talking about, this comes from the fact that like, Javiera, I don't know your background fully, but this comes from thousands of hours and hundreds of different skills that you've built, books that you've read. Like my framing for go-to-market comes from the goddamn Art of War, like Sun Tzu. Like that was my original framing, right? Know thy enemy better than you know thyself and fear not the result of 1,000 battles, right? I was like, fuck yeah, that's marketing in a nutshell. Let's ride that wave, right? That's judgment calls that are built on so much experience. So like if I were to try to boil it down, I think it's just like talk to people, understand them, have good conversations. And then if I were to do a sub-skill, The ability to just go find stuff, like find interesting shit through Google and you'll probably win pretty hard. Again, not on your side of stuff, the technical side, which is like very CRM, like how do I save salesperson hours? How do I round robin this shit? If we're just talking about like figure out a message that turns someone on towards your product, I think that's the important part.

Another thing that I would like to add to this where, uh, Mitchell, you said that you talk to your clients asking them where they would find There's one more thing that I do with my clients. For example, if they are going to an event, when they come back from the event, I have a call with them the very next day. And I get on that call with them and I talk to them. Okay, you match who at that event? What did they say? What did you say to them? Is it going forward? Where is it going? When is your next call? Or if they have a call with someone, I ask for recordings if they are comfortable sharing with me. And most of the time they are, because obviously then I will have more context into their business. And I always tell my clients, you know your business better than me, and I can never claim claim to know your business better than you. I can tell you that I know my GTM strategies, but I don't know your business. So I need as much sauce from you as possible from your business to understand it where I need to actually understand it to build something that actually, that's actually relevant. And the best results that I have had for my clients are for those clients who have been talking to me at least once a week where they are sharing who they are going with. One of my clients recently booked a meeting with someone at Walmart. The same client got response from Vercel. And the reason is we connect every week and we talk about what they are seeing, where they are going, who they are meeting with. And looking at their business from their lens teaches me more about their business than I can ever find from Claude. So that is really, really important.

I love that, by the way. And for anyone listening, like that is, if you want to figure out how to have like a zero churn business, that's the recipe. Like that, my God, you were at an event last week. Oh, you had a dinner? Like at the event, let's go through like what you guys discussed on the dinner, right? Like those little things, especially if you're like on top of it and like the very next day you're trying to extract it from them, it just blows clients away. Or like client mentions small thing and then you show up and like, you're like, you mentioned you're just, you're having a hard time going through, you know, I had a client, they had a hard time going through like the App Store and Steam to figure out like recent game launches. And I just turned it into like a slash command, literally a slash command. And I just hit the button and then all the information would just like shove into their CRM under the record. And they're like, oh my God, we can just do follow-ups now. And it's nothing, nothing crazy. It's like the simplest thing. We already have the tools. Those small things will keep clients sticking around forever. Honestly, forever.

I remember when I used to go to events and I sold into Marriott actually when I was in Dubai. And, and I was like, oh my God, only thing I need to fucking do at this event is set up my next meeting. That's all I needed to do. Like, just go there with the conversation setter and be like, yo, we're gonna walk away and the last thing I'm gonna say to this guy is gonna be like, when are we talking? 2 days, 2 weeks, or 2 months? I don't wanna piss you off. I don't wanna piss you off next week when you're getting texted by like 20 other people. So like, if I text you in 2 days, 2 weeks, or 2 months? What, what's better for you? Because I'm not going to piss you off. I'm not going to waste my time. I don't want to waste your time. And then it's just a matter of like, yo, what's your WhatsApp? Here's the QR code. I'm going to text you right now. And then like going through that and like, I remember I came back from that event. I was like, oh my God. But just setting that up now, I think that's a great idea. Honestly, Jabari, that, that's like great that you talk to them right after that. Not even the pre-event, but I think the post-event, that's great. I'm going to steal that from you.

And Spencer, that's the human element. That like anyone can win on. Like what you just said there, it's the human element, 100%. Actually having empathy. Like I was on a call yesterday and like the people on the call were like clearly rushing into a decision. They're like, I'm down. Let's do, let's do the $5K a month package. Let's rock. And I was like, whoa, hold on. Let's actually pause here and look at, so you guys said you spent the last month talking to a bunch of people. Like walk me through like what happened there. And like, yeah, we've got, you know, 11X, we've got Artisan. We spoke to this other agency, so on and so forth. I was just like, did you guys figure out if you even want to own the system or outsource it? And they just froze. And I was just like, okay, so let's start there. Let's go through like why you might want to own it, why you might not. And what we sold on is like their market is so small after going through the back and forth that they really need to go heavily multichannel. Like if they just go email, like I think it's doomed to fail. Right? 100%. I was even like direct mail. And so because I don't do cold calling, I made an intro to a competitor of mine that I think is fantastic. If anyone knows them, BuzzLead. And I think that human element pays forward an insane amount. I've had people come back with referrals that never ended up working with me just because I helped them figure out what was best for them. And I think that needs to not be forgotten in translation with all these KPIs and, you know, shoving meetings down the throat of certain people, or, you know, I'm gonna pick on Arison again, or locking people into insane contracts that you cannot escape. The ZoomInfo kind of protocol.

I was As well.

Just going to say, like, it's going to catch up to you. Like, I think if you focus on the human element, you win. And I think this is actually, I'll pay homage big time. There's something Kellen Kasper absolutely taught me early on in my journey.

I think another thing is that if you tell your clients that you own everything, I always tell, like, whenever I'm on sales calls, I tell them that you own every, every piece of work that I'm doing for you. When, whenever we part ways, whether you don't get the results or you want to bring things in-house, I will transfer everything to you that is with me and other things I'm already setting it up in a way that you will be the owner of it. However you would want that data, if you want a customized database, I can build a Supabase for you, I can hand it over to you. And if you want to bring someone in-house, I will make sure that after 3 months, 5 months, 6 months, whenever you want to bring someone in-house, I will not only transition things to them, but I will give them a week or 2 weeks or even a month, they can always reach out to me on Slack, on WhatsApp, anywhere, and ask about anything that they don't have access to or they don't understand.

They can spam me on Slack at any time before 5 PM or else the wife will get mad. That's my go-to line.

That is good. My husband is my business partner as well, so he can technically get mad, but he is much better than me in balancing work life because he He is a video gamer as well. He just values his personal time. But I think for clients, this is always good. Even sometimes I sign some deals where I know the client has already told me that we are building it with you because we have no idea about how to do it, but eventually we want to build it in-house. And when we take those deals, it's already decided that after a few months I will help them transition, find a GTM engineer, train them, or whatever. But I think the The, the way you build a business is by being human and by thinking best for your clients. At a point, it's not about money, it's about providing good service, and money is just a side effect of good work. So if you are determined to do good work, money will never be a problem for you. I think this is good advice for people who are starting in this space. Just try to put up good work. When I started, I did free work. for some people as well, and they are still my clients, so it paid off.

Couldn't agree more. Anything else anyone wants to finish on, or I think that's a perfect note.

I think that's a pretty good note.