Building the Foundation for AI in GTM

There's absolutely nothing that makes me believe that I couldn't find 100 people out there that know how to like program an AISD artist. Go to their website, research it, write a message that pulls from the website, and then send that message through our email server out to this list of people. I get those emails every day and most of them are terrible. That's not going to work. Welcome to GoToMarketScience. There's an art and there's a science to go to market. In this podcast, we talk about the science by interviewing CROs, private equity investors, and other sales and marketing experts, as well as talking about what we learn every day in the trenches helping to build GoToMarket engines. Welcome back to another episode of GoToMarketScience. I am Rachel Buchert, the marketing manager here at Unescore Consulting. And with me is Eddie Reynolds, our founder and CEO. Hey Eddie, how's it going? It's going great. Awesome. We excited to dive into this with you. Like always, I love doing these things. It's really fun. Yeah, I'm getting the hang of it more. You sound more natural. Yeah. You ask quick more natural questions. And I don't know, for whatever it's worth for anybody listening, this is just fun for me because it really challenges my thinking, I don't want to come on here and sound stupid. And so I have to really think through stuff, because we write this content ahead of time. But then we go on here and we get to just dive into it. That's really fun. Yeah, practice makes it perfect. Yeah. So today we're talking about how to build the foundation for AI in Go to Market. And this is based off of the A topic that we wrote a newsletter on a little while ago. And that newsletter, we started off talking about the AI SDR, that saster created and how it was outperforming every human rep that they had in two weeks. So we wanted to kind of discuss, you know, how did there's work? How do people try to make something like that? How do they fail at it? What are the foundations that you need to create something like that for yourself? So Eddie, why do you think that story hit such a nerve for people about Sasters AI SDR? Well, first, I don't know if it did hit a nerve or not. I assume it did. It hit a nerve with me. That's all it was, right? I didn't really check to see if it went viral or not. But I think everybody is obviously thinking about AI right now. I'm going to go to Market. I went to Pavilion's CRO Summit in Denver a couple months ago. And all anybody wanted to talk about was AI. I mean, I think that's unsurprising for anybody. But at the same time, at that summit, a lot of what they were talking about is it's like, everybody's thinking AI, AI, AI, but what do we actually do with the AI? A lot of people are like, well, let's go buy some chat GPT licenses for everybody. Okay, cool. I mean, what is that going to do? We need specific use cases. And I think replacing salespeople with AI is kind of like the logical next step. Like, why don't we need all these salespeople? Let's just replace them with AI. I don't know anybody that's necessarily done that successfully. But then you start to like break that down and then you say, well, okay, the most junior position in sales is the SDR. So in theory, that would be the easiest thing to replace. What are SDRs responsible for doing? They do research and then they log out bound activity. How do we do that? And then a bunch of people tried it last year and failed at it. And then Saster comes out and says, hey, like we put a bunch effort into this, like a lot more effort than I think a lot of folks have. And we made this a success. And so that resonated with me because I'm like, well, here we go. All we need to do as a go-to-market advisory firm is find one thing that works. And then if we know how to make that work, we can go and bring that to every B2B SaaS company that we work with. And I don't think the AISDR is the only example, but it's nice to just have something tangible. And we can use this as an example in this content to talk about why it worked, why other people failed at the AISDR. And what that means for whatever other initiative in AI, we want to implement whether it's something in marketing, some other aspect of sales or something in CS. This is so important too because there aren't a lot of really big success stories with AI, circulating right now either. You know, lots of people are trying it, lots of people are like selling tools and products and evangelizing certain things. But I think this Saster story is the first real success case study that I've seen using AI to this kind of scale. Yeah, I mean, I think what's behind your statement, like the definitions of that are maybe a little bit murky. Because like right now, like everybody's using AI and everything. From that perspective, I think what you're saying is not accurate. But I think what you're trying to say is like, this is the first story that, at least you and I have like seen where somebody's like, "Hey, we've leveraged AI in a really, really meaningful way to fundamentally replace like an entire workflow or not even replace because they talk a lot about human and loop, but to fundamentally augment in a really significant way, something that the human was doing." Whereas like, you've been using clay for quite a while. Oh, yeah. So you're using AI, where you use AI. Yeah, I mean, that's what I meant though. Yeah. In a big way where you're like replacing core aspects of a team, not just helping day-to-day tasks. Yeah, I think that that's why I resonated with me a lot. So why do you think so many go-to-market leaders are jumping straight to the tools and prompts without doing any of that foundational work first? Well, first I'll say, I don't know for sure that they are, but it's a fair question because I think that that's sort of the standard. I mean, you put yourself in the shoes of a busy CRO and they've got a lot on their plate. They've got a higher new people, new salespeople, new leaders. They've got to hit it number. Oftentimes they're trying to help close deals. There's just a lot going on and they want to delegate things. Everybody wants to delegate, especially senior executives. And if I can just delegate to somebody to say, "Hey, go build me an AI SDR." Or just go buy this tool that says that they can do something out of the box and let's roll it out. That's great. Unfortunately, it's just not that easy. When we saw the same issues, a number of years back with outreach and sales often and those kind of tools where let's go grab a list from Zoom info, let's load it in there, let's get marketing to write some copy and then let's start prospecting. It didn't work then. It's not going to work now. It requires a lot more thoughtful energy and effort. And that's why I think Jason's post resonated with me so much because they talked a lot about how much thought and effort and foundational data had to go into making that a success. Yeah. I think I got a little bit ahead of myself, funny starting off the podcast saying, "Oh, I'm so much better podcasting now." And I'm mixing myself up. No, not at all. It's like, I feel bad because you're asking a very fair question. I'm just like, "I don't want to make that assumption." Yeah, yeah. But it's oftentimes something that we see. Like, I don't want to negate what you're saying. That's the default. The default is serenity executive to say, "How can I find a quick solution to this and we just go implement this and see how it works?" The problem in going to market is going to market is really complex and nuanced. And there's really nothing that I can think of that works in going to market without a lot of foundational work, without a lot of good data, good process, good iteration. I can't think of a single thing. And if there is a thing that works, it doesn't work for very long because it's just like you got the first mover advantage. And then very quickly, like all your competitors are doing the same thing. And then now nobody wants to hear it. That alligator email only works for so long. Yeah. And like speaking of those foundations, what are the key fundamental parts of those foundations that you think would make a huge difference for people listening to this or did make a huge difference for Saster? I'll talk about what I read from Saster and what I remember because I don't have this committed memory if somebody is interested and they can read the article for themselves. And I'll talk about what we see working with the companies we work with and sort of what that foundation means. So we obviously have a lot of this outlined in our go to market efficiency pyramid. We've also broken us down into outbound. So we have a lot of us documented. The first thing is obviously going to be the ICP and buyer personas. And then next, of course, is segmentation. And I do remember specifically the Jason talked about hyper segmentation. Right? So what are we talking about here? What we're talking about is there's no AI today, August 14, 2025, that you can just say, here's all my customers act like an expert SDR and just go research all of them and then write the best messaging for our product. You would think that that's what AI could do, but it just can't. Anybody that's spent any time in chat, GPT knows that you need to be much more explicit than that. So if you can be crazy, crazy specific about who your ICP and your buyer personas are and hyper segment them into extremely narrow niches of very similar people that have similar messaging that will work for them, you're not setting the AI up for success or the human. Honestly, it's the same thing to me, building the foundations for an SDR human or AI SDR. Because if you don't point them in the right direction, then it's just going to be a much harder research process. But if you say, okay, we sell to CROs and B2B SaaS companies with like $100 to $500 million in revenue, well, that's a very specific ICP and buyer personas. And then if you even niche it down even further from there, you're now empowering the AI or the human SDR to say, here are the specific things that we need to look for when we go to their website, when we read through their content, etc. to personalize this message such that it will resonate with them. The next piece is what is the process, right? Once we have a segment of folks that we want to go after, we're then need to think about like, what is our research process? Like if I'm training a human SDR, what do we look for? Go to the website, go to LinkedIn, what other sources of information do we have access to that are relevant for what we're doing? What specifically should that SDR look for? Let's imagine that we have a 22 year old SDR that's a smart human being but doesn't have any context on our business. That's usually the folks that are in the SDR role. Kind of similar with an AI, super smart piece of intelligence that doesn't have any context. And then you say, okay, what's the step-by-step process to to research all of our prospects so that you can see. you can then write that messaging. What things should you look for, and then how do you write that personalized messaging? What does a cold email look like? What does a cold LinkedIn message look like, et cetera? Okay, cool. Same thing applies to an AI. If we feed that information, we're gonna be in a lot better spot. The other thing that Jason talked about doing was getting all of their data together. So it's one thing to have the process together. Another thing is to have the data. They talked about how they had to go back in and like clean up opportunities and Salesforce because they hadn't been entered properly. They hadn't followed the process properly. They had to do all of this work to backfill data. So they could then feed the AI and tell it like, this is our ICP. These are our buyer personas. This is the messaging that we have sent out. They loaded in all of their marketing content. They took all of their data to train the AI on like, these are the customers that were able to sell to. This is what's worked as we've sold to them. This is the messaging that's worked. This is the messaging we put into our marketing content. Now help us do that research on the next prospect and tailor our messaging such that it will resonate with them. And then they talked about having a human in the loop. Those are the foundational elements. And this is where I think like we run into a lot of issues with the customers that we work with, especially in the very beginning where they come to us. And it's like, we have an outbound SDR team for example, and it's not working. And you're like, well, do you have these foundational elements in place? Do you own it in your ICP or buyer personas? Do you know what like a qualified or an MQL is? When do you pass this off to sales? What is the process? What is the expectation to follow up? How do you research accounts, et cetera, et cetera? When they don't have that in place, it's like, well, that's why these things aren't working. And it's also difficult to measure what's not working because we just know a bunch of things that aren't being done. But we can't see what is being done. We can't measure what works and doesn't work because we just don't have a reliable process that's being executed consistently. It's going to be the same thing with the AISDR. But if we can point the AISDR at a very specific target and say, here's all the information you need and here's the process that you need to go through in order to research that prospect and write your messaging in combination with having the human in the loop, then we can really set ourselves up for success. And I think that's so important to point out too, is like, whatever is going to make an AI-powered motion unsuccessful is the same thing that's going to make a human-powered motion unsuccessful. So like if your processes aren't right, if your data is all murky and you have a bunch of technical debt, you're not going to have a lot of success with a human SDR motion either because they're looking at the same things that AI would be looking at and they're getting the same false kind of information, researching the wrong things, targeting the wrong people, using the wrong messaging, not really knowing what actually works and what doesn't. So it's very similar, I think. Yeah. And I mean, I was thinking about an example that I was talking to somebody about earlier today, one of our past customers, they were having this issue with their inbound leads. And I was looking at their pipeline and why they were like winning or losing deals. And I went and I saw a sales opportunity for over $100,000. And there's no data there indicating like why they lost the deal. What was going on? Who was the decision maker? What was their decision making criteria? Like what kind of feedback did they provide? The rep hadn't logged their phone calls. Like, and you know, I understand like this is time consuming to log all this information in Salesforce. And there are other ways to get some of this information, including AI tools. But the point is, is like, if I'm looking at this as a human, okay, so this deal was lost, but why? I can't see that. If I then take that data and feed it into the AI, the AI doesn't have any magic power to understand something that's not provided to it. It doesn't know why that deal was lost either. But if you have those solid processes and data in place, you can then feed the AI and you can have the AI help you like refine your ICP and your buyer personas. The refine who you reach out to, refine your messaging, et cetera, by looking at what's working and what's not working. AI is like ultimately pattern recognition. - Yeah. And I know it's easy to like look at something like that example you just explained and see where data is completely missing. But what about when the data's there, but it's garbage or messy or you know, not the correct data? How do you know that it's bad data? - It's a great question. So like, I'll use another example. So this same customer I was talking about, I did a pipeline review or really a review of like their closed deals. And one rep had like a 7% close rate and another rep had an 85% close rate. Now, you have to ask yourself, do you think that rep is so successful that they can close 85% of the deals that they chase? You know that that's probably not the case. You can then look at the sales cycle and you can see, oh wow, this rep is closing deals in 10 days, even six figure deals. And it's pretty easy to assume that what this rep is doing is they're waiting until they get a verbal from their customer and then they're closing that deal really quickly and winning most of them. The reason that data is dirty is like, now we know we can't see all the deals they lost. We can't see what their sales cycle was like. We can't see everything that happened in probably the 80% of deals that they actually did lose. So we know that that data is incomplete and unreliable. The other thing is just like, when you look at data, you know, we might have like an opportunity where we filled out a couple of fields on one and didn't do it on the other one. It's just very inconsistent. A third example would be, and while I'm like really picking on this customer, we looked at their MQL conversions on inbound. And I'm looking at this and I was looking really hard at it and I couldn't understand what was going on. Why were there conversion rates like changing so much from like one period to the next? And I called and I asked them like, what's going on here? They said, oh, we changed our definition of an MQL substantially. At first it was like anybody that filled out a form and it was a score. Then we went back to anybody that filled out a form. And this is over a period of like a year or two. So if we're in that situation, okay, I can't trust into that data. Maybe I could have gone deeper and done like an analysis of the exact period of time that the MQL was defined as this and then when the MQL was defined as that. But we also like lacked a lot of like data on follow up. And so it's really hard to like make heads or tails of that data to understand what was working and what was not working. - It sounds like it takes like a really trained eye to look at this stuff because all that stuff that wouldn't just come to me, you know, looking at some data, I wouldn't be like, oh yeah, the SDR must just be only closing his deals once he actually gets a verbal. Like I wouldn't think of that. - Yeah, I mean, there's a reason like why like our team of VPs of Revops, they're not inexpensive. - Mm-hmm. - Yeah. - You do learn this over time. I do think it'll be really interesting to see AI continue to evolve in that way. I think those things are fairly basic for me to think about. But yeah, I don't think that like ChatGPT is trained on that kind of thing just yet. Again, this is pattern recognition for me as a human where I've just been in this game for so long that you see that kind of stuff and you're like, I've never met anybody that has an 85% close rate. There's gotta be something wrong here. And then the next question is like, well, what do you look at? You look at sales cycle. Like, okay, like why? Because I've seen a shitload of deals like that. I've seen a ton of sales reps that Sandbag and they don't want to put any deals into sales for. So like, that's me operating like an AI just going, I've seen this so many times like I know what to look for. And if you train the AI on that kind of stuff and I have every reason to believe that it could pick that stuff up, but you have to train it. - That's so funny you said, like you operating like an AI. Wouldn't that just be regular eye? - Yeah. Well, what I mean is it's pattern recognition, right? - Yeah, yeah. - Like, we talk so much about like qualitative and quantitative analysis and like another way to call qualitative data is gut feel. Who did I hear say this? It's like gut feel is oftentimes like you recognizing patterns that you just don't necessarily notice like consciously. It's just like you keep seeing something again and again and again. It's like when you meet somebody and you're like, oh, I don't feel good about this person. Why do you feel that way? It's because you're seeing something that you just like not consciously thinking of, but subconsciously or maybe just without noticing it, you're actually seeing like literal things like the way that they move their body, the way that their eyes move, et cetera, that you're like, I've seen this before with people I don't trust, but maybe just consciously, I'm not quite like aware of what's happening. - Yeah, that's so fast. I mean, I can talk for another hour just about like the human psychology of that topic 'cause I love that stuff, but I'll try not to take us off on too much of a tangent. - Yeah, I do too, but we won't go there. - Yeah, so when it comes to process that produces this kind of data, how do we define what good looks like when it comes to stuff like opportunity creation or pipeline stage definitions or outbound activity? - Yeah, so I mean, I think this is literally just sitting down and defining the process. One way to do that is to outline the customer journey and we have a good view of like what it's like to go through that journey as a customer from the initial impression wherever that may be to going to the website to becoming a sales opportunity, going and interacting with sales and then becoming a customer on boarding, renewing, expanding, et cetera. That's one way to look at it. I think is a really valuable exercise, obviously. Obviously, I'm not the first person to come up with the idea of doing a customer journey map and then outlining the process step by step. You know, when we talk about the sales process, like we have things like medic and med pick that give us these guidelines on how to think about our sales process, we need to translate that into a step by step sales process. So, medic or med pick is a good example, like where we say, well, what was the decision making criteria for this deal? Did we win this deal or lose this deal? If we ask that question every single time, say that we require our reps to ask that question and enter it at a stage three, and then we have a large number of deals that we can feed to the AI, then the AI can look at that and say, okay, here's all the deals we lost. What are the patterns that we're recognizing? Here's all the deals we won. Oh, if a customer has this decision making criteria, then we have a good chance at winning this. And if they have this, then we are oftentimes losing it. And then that expands really exponentially with the more data that you provide to it. If we know what competitors were going up against. If we know what stakeholders we've engaged or not engaged, all of these things we start to recognize these patterns. And this is something that humans have been doing in GoToMarket, especially in SaaS for quite some time. This is all standard stuff. This is stuff Salesforce was doing over 10 years ago when I worked there. But now how do we train an AI to do that? And the answer is we have to feed it the right data so it can then do pattern recognition and say, here are common things we're seeing in deals we lose and here's common things we're seeing in deals we win. We were talking before about creating these processes, making sure that the right kind of process is we need to get accurate data. But how do we make sure that the process that's been written down like in a slide deck translates into the CRM actually producing accurate data? How do we like make sure that that written here's how we do things? Actually is adopted and like becomes good data in the CRM. It's a great question because I think this is like one of the primary challenges we face with our customers and the work that we do. And increasingly what I'm realizing is it comes down to enablement and accountability for frontline management. So let's use pipeline as an example. This is exactly what I experienced as an AI at Salesforce. They were very, very specific in saying when you're running a deal, you need these fields filled out in this way. And it wasn't a lot. It was a few fields like they weren't as micromanagy as I thought that they would be before I started working there. But Salesforce like public company even when I worked for them and it was critical that they forecasted accurately. And so they said, well, we need these data points in order to forecast accurately. So we are going to absolutely write our team about keeping this up-to-date real time all the time. So at any point in time, somebody could and would call you out for not having your pipeline clean. What we see with a lot of the customers, especially when we start working with them is that management doesn't have this accountability or rigor. And when I talk about management, I mean specifically in this example, frontline managers, sales managers. They don't have a cadence of regularly reviewing the pipeline and then calling out their reps and saying, hey, like I need you to fill in this information. It sucks. Like nobody in sales wants to spend their time updating the CRM. But it's really critical if A, you want good data so that you can be a data driven organization and do things like forecast accurately. And B, it's really critical if you want your salespeople to fall all the right steps so that they can win the most amount of deals. So you have to strike this balance of not asking for too much. But if you've refined what you're asking for and when you're asking for it and being really thoughtful about saying like, okay, like we really need these one or two or three things at this stage, this is worth the time of a rep to enter this. Then we're going to enforce it and we're going to run reports and we're going to analyze those reports regularly and we're going to call people out. Of course, you also have a lot of tools, including AI, they can populate that data for you. So for example, here at Union Square, we have all of our calendars integrated. We have our emails synced up. And so like I don't have to enter anything in the CRM in order to know like how many times I met with somebody, we do, we are really, really like stringent about tracking whether a company we talked to as a partner or a prospect. And then I can run a report saying like how many partner meetings have I had in this period of time without doing anything except just making sure that that account is clearly marked as a partner. But then somebody has to go in there and run that report and just double check like, hey, like we wanted to have X number of partner meetings every week. Did we hit that? Oh, we didn't. Like we're all those meetings tagged as a partner appropriately. You could also have in a larger organization a group of folks that say, hey, like we're going to, this is what Salesforce did really well. It was like we're owning accounts and we are going to control how those accounts are classified in Salesforce. And we're not going to give like account executives the ability to edit accounts. It couldn't change the address for an account or what industry they're in or this that or the other because Salesforce wanted to have control over that as an organization. These are some ways that you could like really control your data and say, okay, well, we have done a lot of work up front to like build out our account database. And then we're going to automate all the emails and all the meetings that get scheduled. And now we can see every email and every meeting that we have with these accounts and we can slice and dice that data based on like all the data enrichment we've done with tools like zoom info, clay, et cetera without asking eight ease to do anything. Another example from Salesforce. This one's like a little bit of a hack. I don't know if we did this across the organization or was just my team, but like we also really wanted to know how often we had face-to-face meetings. And so my like day one manager was just like, hey, if you have a face-to-face base meeting, just type in F to F and the subject line. Like, okay, cool. Like in the automation, just picks it up. And then now we know how many face-to-face meetings we have. Super simple stuff. And then like, okay, every time I schedule a face-to-face meeting, I'll just type in F to F and like the calendar invite like that's no extra work for me. Easy. I was going to say, I imagine like this, I don't want to say laziness component, but this human component, I guess correct me from wrong. But is that where like new processes regularly break down when a company is trying to enforce K that we need to clean data? Here's how we're going to do it. You're going to do this and this and this. And then the SDRs are just like, man, I just want to sell. I don't want to like spend time typing stuff into sales force or like keeping track of things. No sales rep once to spend their time typing stuff into sales force. And the worst it gets is the more senior they are. That's the R's are probably a lot easier. It's like the enterprise account executives that are just like, I don't want to listen to anything that anybody tells me to do. But not just because they're lazy, right? Like everybody's lazy. Nobody wants to do this stuff. Even I don't. I started with this just as much as anybody. The best salespeople though, it's also that like they want to spend time in front of customers. A common metric we talk about is percentage of time spent in front of customers. So if salespeople are having to do so much admin work in sales force, you're literally like taking away their ability to close deals. And on top of that, like if you think about just having like a min market or S&BA with a million dollar quota, their quota if you break that down on an hourly basis is $500 an hour. So if I spend an hour updating sales force, I've lost $500 of quota to the organization. So management doesn't want salespeople spending all day updating sales force either. We really have to be careful about what we ask for, what data is most important and try ways to use things like data enrichment tools and automation to try to fill that data in so that we're not solely relying on the rep to do it. What I think is like the limit we were talking about this internally the other day is could AI just automatically like update opportunities and move them from this stage to that stage or dead them out or whatever. And like who knows, but the consensus we had on our team was at least not yet. Wouldn't really trust it. But you could have AI go through and listen to the conversations and try to like identify decision makers and do this to help reps. But I still think especially for a big strategic deal, it's important that the rep review that information to make sure it's accurate. If nothing else, it's like how are you as a reps supposed to close some big strategic deal if you're never taking time to think about the strategy behind closing that deal. You're trying to work with 10 or 15 different stakeholders and figure out how to thread the needle to close a deal. Great if the AI can like listen to all your conversations, populate a bunch of information. But if you're not going to take any time to read through that information and validate that it's correct and really think softly about how you're going to approach your next step, probably not going to win that deal. Is this kind of like a way that we could get reps to like actually care about inputting better data, like framing it in this way? You need to know these things so that you can perform better. It's hard. I mean, I remember like talking to colleagues about this at Salesforce, where I kind of came into Salesforce drinking the cool weight already. I love this idea partially because I don't have as good a memory as some other folks. And I talked to a bunch of people every day and then a month goes by and I'm like, I can't remember what I talked to this person about and I don't want to sound stupid and I really do care about that. I had colleagues that didn't feel that way. They're just like first, I remember it and secondly, like, I can finally remember every detail, whatever. But I do remember some of those same colleagues like sort of coming around after a few months at Salesforce going like, wow, I'm starting to see how I can close these bigger, more strategic deals. I don't like doing all the admin work, but what they're asking me to answer, like it forces me to think better about the deal and how I can close it. And that's the goal. Like you shouldn't be asking for stuff in Salesforce just because you want the data. It should be in service of helping the AE or the SDR or whomever to hit their goals. Like we're asking about decision making criteria. It's not just for shits and giggles. It's because we are worried that if you don't ask that question, then you won't tell them what they need to hear and you won't win the deal. Yeah, that makes so much sense. And I just realized this myself. But if anyone is listening longerly closely and knows our content really well, he might have noticed kind of a pattern in way the conversation's been going. First, we talked about foundations. Then we were talking about adoption. Next, we're going to talk about optimization. So like all of this really follows through with the go-to-market efficiency pyramid. I just wanted to point that out because it just it comes in everywhere. Yeah, yeah. I don't know if that's us like just having a hammer looking for a nail, but I think that that's kind of the thing there is let's say that we build out an AI. It's an AISDR or we build something to help AE's or CS or whatever. We are going to need, I mean, Jason talked a lot about having a human in the loop. So whatever that human is responsible for doing, in the case of the AISDR reviewing the content or the messaging before sending it out, showing up to that meeting that is booked, replying to emails. I mean, Jason was saying that he himself was replying like they might have used AI to send the initial message out, but then people would reply to that message and then he was replying to that message himself. Whatever that thing is that we need the human in the loop to do, we need to track it and make sure that they're doing those things. I mean, what is the point of building an AISDR that sends out all these messages? And even if the messaging is perfect, a bunch of people respond to that and then we don't respond to those messages, we don't go to the meetings. We're not prepared for those meetings and then like sure, if we just like spray and pray enough people will close some deals, but meanwhile we're losing all of this like money on the table and you see the same thing with inbound. You have a really great successful in deals because you have this really broken process to follow up with those leads. Yeah, whatever the humans are responsible for doing, we need to be really clear about what that process is and how we track that and hold them accountable to do that. Moving on into optimization and acceleration here, what is the right time to introduce AI into your GoToMarket stack? Is there an optimal time to do it? I like to think about what is the thing that we need to be doing and then can or can AI not help us do that better than a human could. Let's start at the base of the pyramid and let's start with ICP and buyer personas. On the one hand, I'm like, well, if I feel really strongly that like, actually, I do feel really strongly that we know our ICP and buyer personas, so I haven't used AI to try to solve for that problem. On the other hand, I don't think there's any reason to argue that you couldn't use AI to improve your definition of ICP and buyer personas. But either way, you have to do it. If any company tells me they want to do anything, go to market and they don't have a clear understanding of their ICP and buyer personas, it's like, well, that's step one. Whatever you do is not going to work as well if you're not clear on who you're targeting. Then we move to like the next step and the next step. So AI can be used in every one of those steps. But what I think is a failure is when we just say, hey, we're going to just buy chat GPT licenses and give it to every rep and then like, go use chat GPT and then like use it for what? Okay, I'll go do some account research. Now I've got chat GPT and I'm going to go do some account research, but I don't know how to use chat GPT to do account research. What I think would be a lot more valuable is to say, okay, as a team, we're going to be laser focused on implementing some AI to do account research. How do we go about doing that? How do we make that a success? Instead of just pitching licenses at our AES and being like, hey guys, go figure it out. I don't think that's sufficient. So what would you suggest for that then? Like that particular process? Yeah, let's dive into that. Let's use that as a concrete example. So let's start with what is the process to research and account? Is this an enterprise account? Is this an SMB account? What are the specific things that we should look for? The reason it's important to ask that is, like someone in the organization probably knows the answer to that. Does the 22 year old SDR that started their job three weeks ago, fresh out of college know the answer to that? Probably not. Does the chief revenue officer or the top sales rep or the head of marketing, they very well may know those answers. So we talk to them, we get those answers, we then sit down and say like, okay, like what's the step-by-step process? We go to the website, we go to LinkedIn, we look for this, we look at that, then we form our messaging. Now we can take that exact same concept and bring that to the AI and we could go to custom GPT for example and say here's your instructions. Go to the website, I know that we can't just like tap into LinkedIn and how to chat GPT, but imagine we could go look at all of this information and then form your research and then what does a good email look like? Here are examples of good emails. That's another thing, right? And now we say, okay, we've thought through this process, this is what good looks like, this is what you need to do to research it and now we train our reps. Hey guys, like here's what you're going to do. You're going to go into this custom GPT, you're going to give it instructions, it's going to go to the website, it's going to go here, it's going to go there, maybe you can't access LinkedIn. So what you're going to do is you're going to go to LinkedIn yourself and you're going to copy paste the page and paste it into chat GPT, I'm making this stuff up by the way because we haven't gone deep down the rabbit hole in this particular use case. And then, okay, we've got that now, we've got our research, now it's going to write the email. Now you're going to take the email and you're going to review it before you send it. And then once it's good, you then send it out. And now we have used AI to automate our research process and our process for writing a customized email. We then need to take a step back from that and like, start to see, is this working? Is this not working? Are we happy with these emails? Are we not? Like, how do we need to like tweak the process to make sure that that is working? And Jason pointed out in his article, we wanted to make sure that we would send an email that we would send ourselves. If the AI is writing an email and you look at it and you're like, well, this sucks. Why would you want to automate that? Yeah, you don't just automate garbage, which is like something we've all experienced unfortunately firsthand in our own inboxes the last year or so. Let me turn that question back on you. You and I have been working really hard on trying to use AI to improve our ability to produce content. We have not in any way whatsoever, like just jumped into chat GPT and written an article and then just published it. No, God, no. What have you done? And I know it's been interesting because we've been working a parallel pass on this. What have you done to try to build out this workflow for writing content? Yeah, and I mean, so much of sharpening the acts. I have trained my own chat GPT project very, very carefully on who our ICP is, content we've written the past, the beliefs that we have surrounding all these different code of market topics, the way we want to come across people, all that stuff, how we value quality in our content. And I think that is helped tremendously. I don't think I could go into just a brand new blank slate GPT and try to get the same quality of outlines, for example, that I would get from what I've built right now. But also, we spend a lot of time finding the topics and stuff first and knowing what we want to talk about. And then I can input that into chat GPT and be like, "Hey, this is what we want to talk about. This is the ideas that we have and the issues we're seeing and the things we want to teach people now give me an outline section by section with bullet points for like a narrative piece that we could write on this. And then it'll give me something. And a lot of the time I like massage the prompt if I look through everything manually and see like, "Okay, is this actually what I want? If it's not, then I'll reprompt it again to try to like hone it into what I'm wanting. And once we have a good outline down, I ask it to write the newsletter based on the guidelines that we've already set out for it. I never, pretty much never use like anything on the first try, the first draft that it gives me. Usually, I need to like take it section by section and massage stuff and sometimes manually changing a couple things. But it makes the whole process so much faster than if I was just sitting there writing it all myself. And by the end of it, I feel like I've had enough manual tweaking and oversight of everything that it's done. It's good content. Like, it's still unique thoughts that we have lessons that we're trying to teach written in the way that we would write it. And like very carefully quality controlled that I wouldn't say just, oh yeah, this is just AI content. I don't think you could just put a prompt into chat GPT and come out with content that we have. Yeah. And I mean, like on my end, I'm trying to think here like, I can't remember publishing any newsletters that I didn't first go into chat GP and say, give me some ideas for for newsletters I want to write. I come up with my own ideas too, but chat GPT or whatever Gemini, whatever we've used like has given us a bunch of ideas. And then I filter those ideas and say, okay, of the 50 ideas that it gave me, I like these 10. So now, I'm manually like human and loop approving those ideas. Then I kind of hand it over to you to like kind of create the first draft. Do you go through that entire process? And one of the reasons why I haven't micr-managed you in that way and kind of just let you do your own thing is because you already know like what a good newsletter looks like. You've been helping me for a couple of years now write newsletters. So you already have that foundational knowledge. You're not like a new marketer that's new to marketing or new to this organization, new writing newsletters. And I'm just like go figure this out. We've already figured out the manual way to do this. So now I'm like, okay, you know what good looks like. So then go figure out how to use the AI to do that. You've done a bunch of iterations for that. You produce the newsletter and then I review it. And sometimes like depending on how I'm feeling, sometimes I rewrite the entire thing. Sometimes I'm like, hey, this is good. I'd make these small tweaks. Sometimes I rewrite certain sections. But either way, like it's going from the AI to you through multiple iterations. Then it's coming to me. Then it may go through multiple iterations. And then finally, like we're signing off and saying, this is content that we genuinely believe in. And also, we fed it all of our previous content so it can build upon it. So if we take something like the Go to Market Efficiency pyramid, we can like say, oh, well, like we want to take that and we want to go like deep into ICP. Okay, cool. That provides this foundation for the AI that's very different from just like, hey, write a bunch of Rob Ops newsletters for us. Yeah, absolutely. And I mean, it's two things, right? It's one, you know, having, like you said, that person, a human who already knows what good looks like and is already a master at the craft itself. And two, unique insights from a human subject matter expert, because we'll get top of ideas sometimes from chat GPT, but we're not getting our actual insights and the things we teach people from chat GPT. Like that's all coming from you. We might get like ideas, but yeah, I mean, there's a heavy filter there. But I will say like one of the things that I like about AI is like a lot of times it, I mean, it's really good at just recognizing patterns across the world and saying like, here's the common pattern. So let's say I'm like, hey, give me like a seven step process to define ICP. It's going to give me something that's better than like 90% of people that work in Rob Ops. Just because like none of those folks are experts at defining ICP. So I get that and then I look at it and I go, okay, how do I feel about that? Do I agree with all these things? Is there something missing? Would I reorder that? But it's easier than starting from scratch. And I think the same thing is true of like writing a sales email if you know what you're looking for, writing a CS email, etc. You have to know what good looks like. But a lot of times if good is just like, here are the basic things that you might not know because maybe you're reading the newsletter and you're just also not an expert in ICP or you're a customer and you're not an expert in whatever the thing is that like is being communicated to you. And it's like here's the basics that can be really valuable. Yeah. And we produce a ton of content, right? Like we're producing stuff every single week. So yeah, using AI has been so fundamentally helpful in accelerating the speed of which we can do this and the amount that we can put out to our audience and our prospects and customers and the value we can bring to the industry without burning ourselves out or preventing us from working on other parts of the business that needs to be worked on. Yeah. You're against it at first. I was against it at first. I was. Yeah. Well, I had a lot of misconceptions about it. I was like, "A.I. written slap, not on my watch." Yeah. Because I care a lot about the quality of what we put out. And yeah, I'm like one of those people. If it's not good and valuable and useful, I don't want to even create it. There's no point. Yeah. And I appreciate that. And I think like for me, I was looking at it through the lens and I was like, "Hey, Rachel, I'm not asking you to go into chat UPD and write a newsletter and then publish it based solely on what chat UPD says." But let's break it down to its component parts. And let's see if and how it can help us with those component parts. And I love like A.I. is a ability to help you with idea generation. I mean, I don't know if I told you this. I named my daughter with A.I. Oh, really? That's crazy. Yeah. I mean, I spent a lot of time on it. Like this is like a big decision that I didn't want to screw up. But I was like, "Give me all these names." And like we went through it and then I shared with my wife and like we eliminated a bunch of the names. Like one of us didn't like and then we kept going back and forth, back and forth until we finally landed on something. Yeah. And I love it. I don't know if you care about saying. I don't want to add to my daughter's name on a podcast. That's weird. I love it though. I think it's a great name. Well, thanks. I appreciate it. Yeah, I haven't like posted photos of my family on like LinkedIn. I don't know. Like I'm not a big social media guy despite all the time I spend on LinkedIn. Even like on Instagram, like my wife posts everything and I just get these little messages. And it's like, "Do you want to post this to your story?" And I don't even know what that means. I'm like, "I still can't figure out the difference to me." Like a story on Instagram or whatever the other thing is. I just basically just close the buttons that are like sent to me by my life. And then everybody's like, "Oh, I got a lot of pictures of you posting." Like I'm not posting them. I don't know what I'm doing. That's funny. Coming from someone who's like so savvy with tech and stuff. It's not that I couldn't figure it out. I just don't care. Yeah, that's fair. All right. So moving on to how people can do this for themselves. A lot of this sounds like it could be a pretty heavy lift, especially if you're looking at your foundations and realizing, "Oh crap, I need to change a lot of my processes here." Maybe we're missing processes that we didn't even have. And now I'm realizing that I got to do this. Where do we want to tell teams to start? I mean, I think the whole point of this is to say, like, start at the foundation. But like, I think specifically with AI, as we are rolling AI out in our own organization and helping our clients, the first thing is like, let's look at a very particular use case. That's going to make things a lot easier versus like, let's just buy a chat GPT and try to figure out all the different ways that we can use it. It's like, "Okay, that's going to give a mediocre result." What I loved about Jason's post was like, "We have a very, very clear end goal. We want to build an AI SDR that will write messaging for us and send that messaging out and help us book meetings." And then let's think about exactly what that looks like. Jason talked about how they leveraged their existing database of contacts instead of going cold. And then the process to go through that. There's a million different use cases that you could have in AI. And we wrote a newsletter and did a podcast on this recently unlike the common use cases that we think are most interesting. You pick one of those. And then you think about what are the foundational elements we need for that thing? We need to know who our ICP and buyer personas are, obviously. We need to know what the step-by-step process is to do that thing. And then we need to think about each of those steps and how might we leverage AI to do that. And I think this is where that's the hardest part. And then there's a ton of people out there. Our team included that you could hire and say, "Go help me implement this. Go find this tool and that tool and integrate it here and pass this data from here." And like, to me, that's the easy part. Maybe it's the easy part because I don't know how to do all that stuff. And I just hire people to do. And maybe it's really hard for them. But it's easy for me in the sense that there's people out there that know how to do this stuff. But if you go and hire some AI dev to go and build your AI SDR and Jason said this in his article as well, and then you just press a button and hit go, it's not going to work. There's absolutely nothing that makes me believe that I couldn't find 100 people out there that know how to like program an AI SDR to say, "Go to their website, research it, write a message that like pulls from the website and then send that message through our email server out to this list of people. But if you do the foundational elements, just like you would if you're trying to build a human SDR team, or hopefully you have built a human SDR team and now you want to take to the next level, then you have all the writing ingredients. And then not to plug us too much. But like, the whole reason we're doing this content is because while I can't say we've done this a hundred times, I don't know who has successfully, we have helped to build SDR teams and AE process and CS process and we can help build these foundational elements and work with you to build the AI to do this stuff. Even for like completely new, go to my commotions and products. I mean, yeah, like that's going to require more iteration. It's a lot easier if like you already have it successful, but yeah. But it's doable and we've done it. Yeah. But Jason's team with Saster, they got their results in two weeks. How realistic is that for other people? Well, he said they got the results in two weeks by doing nothing else but that. I think that's a great question. Like people will always ask us this stuff. Like, well, how long does it take us to do this? Like, if you're really focused, you can do stuff really fast. The problem organizations run into is especially with senior stakeholders. They don't want to commit the time and the energy. I mean, Jason is an extremely successful tech leader and he's talking about spending all of his personal time on this for two weeks straight outside of meetings and like the daily stuff. That's a significant investment for somebody like Jason to make. So if there's like a CRO listening to this thing and like, I'm going to delegate this to my like Salesforce admin to go figure this out. And in two weeks, they're going to do this by themselves and we're going to launch it. And that's going to be a success. It's not. But if you sit down and say, Hey, like over the next two weeks, we are going to devote ourselves to this. It's all hands on deck. We're all going to work on this thing. I mean, I can't guarantee that your organization can do this in two weeks. Your data might be messier than Jason's. You might have less resources, et cetera, et cetera. But I think within a very short period of time, it's very realistic to get something that could actually work. I mean, how long did it take us to work on the way that we generate marketing content? I mean, there's a really concerted effort like here and there. I think overall, it probably took us maybe a week or two. If that doesn't mean we don't continue to iterate on it. But yeah, it's not like we spent a year working on it. And then after a year, we're like, oh, we still haven't figured out how to do this. Oh, yeah. I think I talked to you about this a while ago too. Like when we're working with a client, for example, getting through like the fundamental part of the pyramid can only take three weeks, three to six weeks. It really depends on the size of the team, but yeah, it can happen really fast. I mean, you could do it in a day. Yeah. I don't know if this analogy makes sense to people. But like, I remember during the financial crisis, the New Yorker had this great article. I think it was like 60 pages long. And it talked about all the stuff that went down over the weekend, where the government and all these major banking executives all met and tried to figure out how to save the world. And when they detailed this article, for what happened over the course of I can't remember if it was three or four days, whatever, when the major bailout package, et cetera. And you're like, typically that kind of stuff takes like 15 years. But like if it's urgent enough, you just get all the people in the right room and it's all hands on deck. And I remember like, as a funny accolade, there's a scene in this story where Jamie Dynan is in his limo or whatever it is. And he's there with like somebody else from JP Morgan. And he goes like, oh my god, this is the scariest thing I've ever encountered or something like that. And he's like, dude, we're right in the limo to go visit the Fed. You're not getting out of a U-boat storming the beaches of Normandy. But anyway, like my whole point is like if we make it all hands on deck, like which is what it sounds like Saster did, there's no reason we can't get a lot done really fast. And where we see organizations fail, and this is what I've seen for 15 years of being in this space, is we dip a toe in the water. CRO comes in and says, we need to like roll off this new initiative and then they disappear and you can't get a hold of them. They don't want to review anything. They delegate it to somebody else. Then they delegate it to somebody else. And before you know it, like the only person that cares about this is somebody that's super junior and doesn't have the power or the context in the organization to affect change. And then it just dies on the vine. That is the big death rattle of any of this kind of stuff. It's when the only person who cares about it has no authority over actually doing anything about it. Yeah, like we were talking to a prospect like the other day and this guy was asking me like, okay, like how do I know that this can work? We've tried this before and I just saw them as like, look, and this is a, I think the company we were talking to is a really small company though. And I'm like, look, if you commit yourself to this, maybe we completely screw up and you fire us. But I guarantee you're not going to fail at this. I worked with 300 customers at Salesforce over the course of three years that worked with multiple different implementation partners, internal external resources, et cetera. Half of them were failing at like getting Salesforce to support like a really efficient go-to-market engine. And the other half were quite successful. And every single one of my talk to it is just like, there's not a single one of these companies that were failing where like the VP of sales or CRO or CEO gets on the phone with me is like, I, I've tried everything. I've been committed to this and we fail that and not a single one. There's not one example I could think of. What was more often happening is like their Salesforce admin would be like, we've been trying this so hard and we're failing miserably. I'm like, cool. Can we talk to like the CRO or the CEO about this? Well, no, no, they don't want to be involved in this. Okay. That's why you keep failing. Yep. And so what would you say to like the skeptics, CRO or revenue leader who thinks that all this foundational work sounds like an excuse to just delay AI adoption and just shipping it? I hope they have a good investment for a fully, because I don't think they're going to have a job in 10 years. 10 years. Maybe not five. Probably one. I don't know. What's the average 10 year? Oh, I mean, I don't think they're going to be employable in 10 years is what I mean. Oh, employable. I see. Yeah. I don't see how you survive in this environment if you are not willing to. I think that like we have gone over the last 10 or 20 years from an environment where you had a lot of salespeople selling in a very old school way and it's still worked. It might not have been that efficient, but it And the reason it worked is because if you can hire the right salesperson and they go to the website and they manually Research everything and then they write a really thoughtful email or log a thoughtful phone call and they do it with enough like hustle and tenacity They're gonna close deals if you give them literally nothing more than like a laptop and a phone. I've personally done this Even in today's day and age like you can do that But that's changing. There's just absolutely no way that you're gonna be able to compete with this level of technology in like five or ten years If you're like we're gonna continue to do everything manually and so I think that you have these leaders that have gotten This far with that mentality because it's like look I don't really give a shit about Salesforce like log your opportunities in Salesforce. Log your calls. Whatever. Oh, we've got outreach Okay, cool like go delegate that to somebody and if it doesn't do xyz y'all get super upset But like whatever I don't care This becomes like significantly more complex and I just think like the organizations that invest in this stuff and getting this right are going to accelerate so much faster and I just don't think that that old school mentality is going to survive the next five or ten years I think that that's gonna be like my Rachel have you ever been to a farm? Oh, yeah, okay I don't know a lot about farming but my grandfather was a farmer. We had a farm We sold it like I don't know ten twenty years ago. It was like a hundred acres But you know the old expression forty acres in a mule. Yeah, okay So last time I went to a farm was a few years ago. I was in this exact head program We went out to visit these farmers to like learn about the agricultural industry the first farm was 17,000 acres and it was one farmer Oh my god one farmer 17,000 acres the next farm was 34,000 acres with cattle and it was three farmers now granted like they had some like farm hands and some like temporary seasonal labor etc but like the point was I'm like how do you farm 17,000 acres with one person when my grandfather could like barely do a hundred acres by himself It's like well, we have a shitload of advanced like machinery and technology like these tractors like have satellite connections and basically drive themselves You know these massive machines a million dollars of equipment and it's like you can't farm your forty acres in a mule without starving to death I think that go to market is headed in that direction I think we've been headed in that direction for a long time, but the problem is is like a lot of the technology that has come out has had like marginal impact and like Salesforce helps us organize things But like really good sales reps were doing that in a notepad and they were doing that successfully maybe not quite as efficiently but successfully Then outreach came out and it's like I still don't really use these tools as much because I don't know like they just don't help my personal workflow all that much But I think like as AI really takes off you're just not going to be able to keep up Yeah, yeah, that's totally fair So how would you suggest people like maintain their momentum and they go to market team then while they're doing this like behind the scenes foundational work Because you still have to like sell into your day-to-day stuff, right? Well, that's the whole idea of like having a revops team and having a revops roadmap I think they'll like the CRO the CMO the sales marketing team CS etc Like should be focused on like their daily activities, right? Then you go to your revops team and you're like hey, like let's go build an AISDR or let's go build a tool to like do account research And you figure out how to make it work And then you release it to the team and you say hey guys, we have a very specific thing that's gonna save you time Let's just use account resources example Here's exactly how we're gonna like use AI to do account research and make you better and faster Now you have an AE. It's just like focused on closing deals closing deals closing deals and all of a sudden they're like Show up to an hour of training and like here. Here's how you can do better account research Okay, cool. I do that now. I've got another tool in my tool belt great That's a lot more tangible thing than trying to boil the ocean and like just like oh, which tools should be implemented Let's buy this one. Let's buy this one. Let's try this no like take one thing and work on it until you make it work And then you can move on to the next thing and pick the thing that is gonna have the greatest impact in your organization Think about like if we get improve one thing in our go-to-market would it be generating new business or would it be improving that revenue retention Would it be closing more of the deals in our pipeline or generating more pipeline? Do we have like a good pipeline management process No, okay. Well, is there a way to fix that? Do we need AI to do that? Should we do that manually? Like I think that we need to look at What's the thing that if we fix it whether it's with AI or not? It's kind of the greatest impact in our business And then if we're gonna use AI how do we use AI to make that a thing work and then let's just commit ourselves to doing that until we get it right Well, however many times it takes whatever it takes we just keep iterating until we've got it right Instead of running around trying to play whack a mold doing 15 different things at once And I think that's kind of what you and I did with the marketing content I mean, I we're still tweaking it is definitely not its final version But I think we committed to say like we are going to find a way come hell or high water to use AI to help us generate content Yeah And we for anyone listening we did a very recent podcast on exactly that the go-to-market ops decision tree like how do you Prioritize what to do first how do you figure out where you need to start and you went through all of that all the different layers like Is it this or this and you go down to the next hours at this or this like one of those Everyone seen them pull little Boxes with the arrows that did just yeah decision trees Yeah, is that what they're formally called decision trees or are you like talking about like a flow chart? I don't know yeah, like little flow chart tree. I don't know the name of this stuff you build in my mirror or Whatever it's called. Yeah, yeah Like choose your own venture. Yeah. Well, anything more. So we wrap here Yeah, no, I thought was it for me. I was just about to say we'll have links to our frameworks Where that has a ton of information on all of this stuff and you can find that go-to-market ops decision tree Podcast on our podcast as well. Maybe I'll link that as well in the show notes if anyone's interested there Well, and also the newsletter itself that we're talking about right where do you publish that didn't we? Yeah, we um we have already published the newsletter based on this topic So I'll link to that too if you want to read the written version cool. Thanks Rachel sweet. Well, thank you so much Eddie Thanks for listening to the show if this resonated and or you'd like help with anything we talked about in the show Please reach out to us you can find us at unionsquareconsolting.com and the info will be in our show notes