Finding Your ACP (Actual Customer Profile)

Foreign.

Welcome everybody to another event of the HubSpot Admins Hug. I'm Kyle. It's great to be here with you. It's been a long time. Thanks to Deanna for holding down the fort for however many weeks it's been since I've been here. The little baby you saw is the reason I was gone and I am now back. This is our homepage. If you're not familiar with this page, please check it out. Keep an eye on it. This is where you'll get information about our upcoming events. We're here every Tuesday. You are here. We are going to learn about ACP Actual as opposed to ICP Ideal customer profile. Next week we are not here. It's a holiday here in the States so we're taking that week off. But we got some good stuff throughout December. We are starting to build out our January. Keep an eye on this. More more events coming soon. Love having you here looking retrospectively. All of our events get recorded and put on this YouTube channel. So if you ever miss one, want to rewatch one, share it with a friend, whatever, this is your place to do that. We just had show and tell. I'm very sad I missed it. Heard it was great. Show and tells are maybe my favorite thing we do here where you all get to learn from each other. Some of you get to show off stuff you've been working on. We would love to do that on a fairly regular basis. So please, anytime you built anything you're you're proud of, whether it's a tiny simple thing that you think everyone should do or wildly complex thing that you are the only person on earth that could do, we would love to see it. We would love to see a diversity of solutions and things being done in HubSpot. And once we have a collection of enough willing presenters we will schedule another event and make that happen. We have a space on community.HubSpot.com it's great. We have a group on LinkedIn which is also nice. We have this. I'm going to go out of order here a little bit. If you go to adminhug.info you can find links to all the things I'm showing here. And this is. This is a pretty handy resource so unless you want to just grab all the links in the chat as Deanna is sharing them, this is your place to go to find all the different resources we're sharing here. The last one is we have this form because we are here every Tuesday. We want to make these events as relevant to you as possible. So please, anytime you have a suggestion or an idea or feedback for us, let us know. Especially here at the end of the year, we are, we are setting our goals and vision for 2026. We would love your input on how we can make the admin hub ever more relevant to you and the work you do. That's enough for me. I'm super excited to hand it over to Nico, a person who I encounter on the Internet on a basically daily basis, I think, but apparently we have never done anything together until now. History being made. Very excited. Take it away.

All right. Thank you so much, Kyle. Amazing and super crazy. Honored to be introduced by you. So this is certainly one of the most humbling things that's ever happened. So, yeah, today, acp, right? Hate to say it, the ICP that you guys have been talking about forever, it's actually not really your icp. Probably more than likely, definitely. So we're going to talk about why ACP is so important and why that trumps ICP so much. The theory is basically when you talk about icp, it's somebody who thinks they know your business. They come in, they sit down, they talk with you. You guys sit there for two, three, four hours, sometimes six hours, and they swear they know everything about you. And they come back with this profile and it's supposed to fit everything. You know. Unfortunately, what typically happens is at that point you've thrown out a giant net and you think you're in the right ocean and hopefully it's going to catch the right fish too. But that's the biggest problem, right? Don't really know. It's just a theory. What you're expecting is the dream client, right? You hope that you've shaped this thing so well that anybody that walks through the door is going to fit this complete mold, because obviously they must, right? We've set it up, we took all the time, we sat there for hours. So clearly everything that walks in is going to fit the mold and we should be able to close all of these deals and should be effortless, right? Unfortunately, yeah. Doesn't end up being the case. And so that equates to a lot of spend that goes in the wrong place to get the wrong people who waste a lot of time of yours. So what we want to do is shift. Right. Unfortunately, right now we're still working with the old stats, doing things, the old ways stuff I really hate. If you guys know me, you know that I'm like extremely data driven since the last like year and a half. So if you're not doing things by the numbers. It's not necessarily that, oh, the calculations are telling you everything you need to know. It's what's in between the calculations. It's the inference that you can make now with AI that allows you to see what you couldn't see before. So now you can actually find out like where all the money is going to waste. You can find out which sales reps not necessarily closing the right types of customers. And the best part about it, you can plug the gaps. So when we talk about icp, again, we're talking about ideal customer profile, who you wish you could have. When you talk about acp, we're talking about the actual customer profile, who is actually in the CRM, what has actually happened, not just from like the time that you started the deal, but from the time that they didn't even actually talk to you, right from when they just went to visit your site all the way through to when they actually became a customer of yours. And even afterwards when they start talking to support, the whole journey is more important now than it's ever been. So the ACP is one thing that's found in like the initial definition of it is basically a customer where you have a closed one deal, right? That's a customer, it's an actual customer of yours. That's where it begins. Right? So in terms of like where do we start? What's the data that we're going to look at? It's going to be based on close one deals. I knew there was going to be one that was going to do this. So, so how do we define these ACPs, right? What we're looking at is again that inference of their behavior. We're not so much looking at this data driven isometric, okay. If they match these stats, this is definitely the case. Those stats are going to fall in line anyway. But what we want to find out is like what's the behavior? What do these people typically like to do outside of just what are the stats telling us, Right? So we want to look at like what's the engagement, how often they're buying, what's the expansion potential based on what they've bought before, what you know, how often they might buy in the future if they're going to come back? Support efficiency, do they need a lot of help? Don't they need a lot of help? All the people that have, you know, how, how many people are tracking the stats of your support requests that happen post sales and then going back and doing the monitoring to actually put a Measure in and see, oh, this customer and this type of customer needs this amount of support. Post close. Who knows that data? Raise your hand. Yeah, I didn't think so. Most of you guys don't even call your customers like six months after you're done to find out, hey, if what we did was a lift, right Oaks come, you would, you would. So relationship strength, right? This is a really, really big one because again, if you're following all of these things, you'll understand that this isn't necessarily just about like finding a particular person and finding a stat and putting an ROI on it. It's much more about finding that person and developing the relationship with them. So the good news, the data you have is just fine. It's okay. We can do this without having to do a ton of enrichment. It's not going to be the end of the world. Would having more data be better? Obviously the more data is, the better. Always. That's, that's not going to not be the case ever. So. But the, the nice thing to know is that you can get into it now without having to worry about like, oh, we need to bring somebody in and clean it up and I'm not going to know if the deals were closed. We're. That's fine, right? That's cement. We can take a look at it. If there's some, you know, dust on it, we can brush that off. It's no big deal. Really easy steps. Just take a look at the data, analyze the data, find out our cohorts and then build off of that. It really like all of these concepts are going to be very simple. The toughest part about it is changing your mentality, understanding that you have to go from I think this is the case to based on what we found, this is actually what is the case. Right. You have to change those meetings where people want to actually drive like that marketing idea based off of, well, I feel like this is really going to have an impact. Change that attitude. It's got to be. No, this is actually what is showing the impact. This is actually what's showing us the direction we need to go in. Right. So let's find out who we're looking for. Right. The assumption target. You know, we're going to look at VPs of marketing, but when we look at the data inside your CRM, it actually shows us that you make more closed deals with directors Insight, same budget authority, but they have a completely different buying process. The action for that would be that you redirect targeting and restructure your sales Approach taking a look at like how you discover your cohorts. Right. Want to identify distinct customer cohorts based on actual behavior. Right. So we're taking a look at the actual data and we're breaking it down to do exactly what you would otherwise. Right. With your icp. Only instead of having to sit there and waste hours with somebody else hoping that they understand your business, you can just train a model to understand your business ahead of time, which takes very little time. If you need a hand, I've got a 10 minute course, no big deal. And after you've got that model in place, now you've got the perfect hunter for your cohort. Right. Model understands your business, you're going to give it this prompt and the prompt is essentially telling it to take a look in the CRM data based on everything that it finds. Id like three to four cohorts that match based on all the closed one deals and just giving it some like again surface level data to look at initially so that we can at least just start to sift a little bit. We're not necessarily trying to get too granular yet, but we can always add on to this. And that's what I really like about this methodology is that it's a nice basic workflow that you can branch so much more detail off of. So looking at this, essentially what we're trying to do is extract those three to four groups of high value customers and we're defining value here as people who actually genuinely contributed to revenue in a meaningful way. In other words, excuse me, it wasn't just somebody who walked in the door. And your sales cycle is typically two, three weeks, but then six months later they closed the deal. Right. And depending on what that deal size was, that's going to determine what the value is. Right. So that's where we're, we're really trying to separate things is some people might say offhand looking at that information. Oh well, you know, wasn't really that great a deal. We wasted so much time on it. We should try to close people faster. Yeah, but if that deal ended up being worth the squeeze, maybe we need to take a look at that. Maybe we need to turn that into a segment. Maybe we can actually find more people like that in the CRM who have that, that typical thing, the what buyers want white papers and PDFs, didn't want nothing else. Don't worry about it. We want to give them too much content. Turns out customers that you had in your CRM converted after watching product demos and reading case studies so it turns out that your best buyers are solution focused. And that means that you need to shift your strategy to be solutions focused. If you're not looking at what the behavior of your buyers is leading up to the purchase and after the purchase, you won't be able to determine how they got there in the first place. And most important, most importantly, why they got there. Everybody talks about like, oh, well, it was SEO. It's aeo. Well, it was this landing page that I made. It was this email that I drove. Yeah, but did it actually have the impact? If it did, where is, where's the measure for that? The measure would be that they matched, they came through, took the measured amount of time for the close. Right. I think most people say, well, we have a typical close between 30 and 60 days or between 30 and 90 days. Okay, well now we can actually know what that number is almost finite. And we can say, oh, this customer is supposed to close in 12 days. So did it actually do that? Even though they might have come through the same channel, hit the same landing page, did all the same steps? We didn't pay attention to the type of person it was. We only paid attention to the stats. So we assumed that the stats equaled the person. And then when it didn't close, we wondered why that campaign didn't work. Right. Does that sound familiar to you guys? So let's get in there. Right. After we've found the cohorts that we were looking at, what we want to start doing is actually dissecting them in almost like a psychological way. Right. So we want to see what the characteristics are between the companies. We want to take a look at what the compositions of their buying committees are, want to take a look at content that gets consumed before closing, communication channels that get used, events that triggered conversation conversion, common objections in notes or call logs, sales tactics that worked. All of this at once, Right. Using that GPT connector. That's literally the case. You just deep research, connect to GPT slap in this prompt. I mean, go one and then two, and then you're sitting back and you're watching basically that initial net start to come down even more. Right. Because now we're starting to really identify, like, okay, of that pile that we started out with, which ones were really successful. So even though we have like the most successful pile now, it's even more successful than that. And it's more defined than that. Right? Yeah, yeah, you could, you could definitely burn a lot. Yeah. Research, if you're using API burns tokens like no tomorrow. So Be very careful about that. But I think even on a free level you can get deep research searches. I don't know how many. I'm sure it's quite limited, but definitely worth checking out. And I would suggest it for Claude, except that the token window size is still too limited for right now. But I mean depending most the case, the problem is token size and token context size is going to determine based on like how large your CRM is. If you've got 100,000, 200,000 contacts, probably going to have to use GPT, not really going to be able to use Claude. If you've got, you know, maybe 30k or less or 50k or less, maybe Claude would be able to do it. But you want to work a little bit more granular to refine the prompt and refine the parameters so that it's not going wild and looking at just everything in the CRM. Right. So after we've actually honed it down a little bit more, then we're looking at the. When taking it, your. We assume, like I said, that we have like a 30, 60 or 90 day sales cycle. Turns out we've got some fast, fast decision makers that we didn't really understand because even though they fit the mold, maybe we just closed them faster. And it was, it was what it was. And we have notes and sales guys have notes. But how often is your sales team actually sharing stuff between each other? Raise your hand if your sales team gets together every morning and they're like awesome guys. All right, what did you do yesterday? How'd that work out? Like what works best? Why don't you tell me what's the secrets? Anybody sales team doing that, like seriously working together in that way? Because now they can, because you can take all of their individual notes, smash it all together and now you have this map for. Oh, okay. This is what works for this closing group. This is what works for this cohort. This is the type of messaging that works here. What that does at the end of the day is something that I really love doing. It takes the playbooks away. How does it do that? Well, if we found the right customer based on who we've already done business with and then we take a look at our sales reps and how they work so we understand who they close the best with. Then when it comes to assigning reps, assigning their leads, well, now we're matching the lead to the rep to the assignment. So the likelihood of the rep not closing now becomes extremely low to the point where if they don't might almost be on them as opposed to being on the system. Right. Because now you have this needlepoint definition of who it is that they're going after. And the playbook kind of starts to go away. Because the whole matchup is based on the way in which they do sales in the first place. Right. Instead of just like round robin. Okay, well, they've got availability. Okay, well they should know how to do it. Oh, well, they haven't finished onboarding yet. Yeah, but maybe they have a particular way of closing and they understand it. Within two weeks, you'd be able to use this to match up closed deal. I mean, say two weeks if you have B2C. If you've got B2B, probably two, three months, whatever the sales cycle might be. But you'd only need five to six deals from each sales rep that had a decent amount of nodes closed one to be able to make this determination. It doesn't take as much data as you would think not to start with. And then as you keep going, it just keeps getting more refined, so the decisions just keep getting easier and easier to make. So, like I said, usually have that sales cycle, but unfortunately we figure out that's not the case. Been trying to use that one size fits all solution for everything using those playbooks. That's what happens is we get lead in. We don't even necessarily define. I don't know how many CRMs I've been in. You guys, you can talk about it all day. I don't know how many CRMs I've been in where there's Personas and there's no workflow to assign them. What's. What is the point of that? So why not just have the filtering set up properly out the gate and not necessarily even have to worry about Personas? Because everyone that walks in the door is a valid buying customer, or at least way closer to being one than they would otherwise. Right. Sorry, Josh. I want to be your best friend, though. So, yeah, we want to take action by velocity on the cohort itself. We don't want to take action based on the segmentation. So when it comes to the opportunities that we might have now, we're going to look into the CRM and we're going to say, all right, pass on this on the past 12 months. Score each open opportunity from 1 to 100 based on how closely its behavioral patterns match the successful cohort. That's right. So now we can take a look at pipeline, and we can already make the assumption in pipeline of like, way more accurately, of hey, this deal is actually, you can adjust those percentages that actually mean something because now you will know how far that deal is likely to travel before it either goes close one or close lost. Because the behavioral patterns, the reason that this is so powerful is because behavioral patterns are so powerful. They follow rhythms, they follow stances. Like, sure, you might get one outlier out of a hundred, but that's about it. That's way better than 50, 50. It's way better than 60, 40, 70, 30. So, yes, it can. Cameron, put it on the right setting. So, yeah, this is definitely the way to close the gap on the missing space between your lead generation and what you're actually able to close. It's closing the gap on who your customer is versus who you're actually marketing to. And it absolutely closes the gap on skill level needed between customer acquisition to end of service and reacquisition or renewal impact. Again, targeting focuses on customers who actually convert. The conversion is done by behavioral pattern. So it's not even a matter of like, oh, if we can get them to, to do this, it's, oh, well, we know they're going to because they want this content. So we just have to put it out there. Speed and retention. Because when you put it all together, what ends up happening is this beautifully flawless buyer's journey. And the only thing that makes me want to continue to use a brand is how easy it is for me to get in and out. If I can get to your site, buy my product and be done. Awesome. If I can meet with you, proposal kickoff, Start the project. Awesome. If we can sit down, you can give me the details and I understand it by the end of discussion. Awesome. It's all about the value that you're able to provide during the amount of time that you have to take. So find the people that really matter. Find out what it was that got them to your door in the first place, and then start talking to those same people in their channels where they live online so that you can start garnering more, more of them and make more advocates of them. So try it for yourselves, right? If you got plenty, if you got 20 or more deals and close one stage, then you can definitely start with this and you can take a poke at it, see how it works for yourself, change your outlook, stop going based off of what you think is going to happen and start going based on what you know has happened already. Isn't hindsight 20 20, right? Like, that's, that's how that's supposed to work. So there is this this theory of like, oh, looking backward doesn't really help you. It keeps you blind from going forward. But yeah, that was kind of the old way when we were just looking at stats and we weren't looking at behavior and those stats weren't able to tell us the behavior in the same way that we're able to know now, because before we had, you know, quantitative easing and we had again, we were just applying formulas. We didn't have the ability to take a psychologist and layer them on top of the quantitative data to then actually tell us what the behavior pattern is. But now you do. It's pretty much in your pocket too. So ask yourselves one question. What did the customers do? What did they actually do before they bought? Before they signed the deal? Right? That's what you're looking to find. And once you've found, will begin to tell you everything else that you need to know about not only that customer segment, but how the rest of them act in terms of when they would buy from you, why they would buy from you, why they wouldn't buy from you. Because you can take a look now at all the streams that went away, which people typically don't pay attention to for whatever reason. I don't know. But the stuff that you lost is way more important than the stuff that you won, because the stuff that you lost is telling you where not to go, it's telling you what not to do. And if you're not using that and recycling it so that you understand, like, oh, we need to change our methodology so that we start, so we disclude these things that we keep from doing this stuff because it's actually hurting our sales, it's hurting our ability to convert. How would you know unless you're actually data mining it? Otherwise the deal was lost. The rep tells you, yeah, well, you know, they just weren't interested. You know, it was too expensive or whatever. Really, something must have happened in between there. Like a few actions must have taken place. There must be some ripple effect. Let's go take a look at that line and find out what really happen under there. Then we can say whether or not they would have bought the product. And maybe you just need to approach it differently because if that, if that person lines up with a certain cohort, we know exactly what you should have said to be able to get there. So data is there, tools are there, all you have to do is make the right decision. And the only thing that's stopping you from understanding your customers better than your competition is your willingness to actually buckle down take a look at the data and actually and make actions actionable decisions. I can't talk now. Take a look at the data and make actionable decisions based on the data, not based on theory. So that's, that's what I've got. I do have a one pager that I will put on the post as well and be able to give to you guys so that you have just a reference. It's got a little bit more to it than just these basic prompts that I went over. So it has a little bit more meat on the bones and will likely give you some, some nicer output. So can give that to you. And then unfortunately Cloudflare kind of put a really bad kink and things. So definitely going to be doing some follow up content with Deanna and put a little video together for you guys to show you what that looks like as well. But yeah, if anybody has any, any other questions.

Thank you. So one question. Oh, we have a few questions in the Q and A everybody if you have questions. I'm looking at the Q and A now, so put them there. But Jeanette asked earlier on and it looks like moved it over here too and it's got a few upvotes. She says you can only do this by connecting your LLMs to HubSpot, not just using Breeze Assistant. Do you want to talk a little diff, a little bit about how you see these tools and strengths and weaknesses and how you're doing?

Yeah, it's definitely a method, in my opinion that is more meant for open AI or using Chat GBT as opposed to using Breeze. It's not as if you shouldn't do one over the other. But I'll put it to you this way. Breathe. And I can understand where that's coming from because it's like, oh well, why would I take time to teach a GPT model about my business when Breeze already knows about it? And then why do I need to go get the data when Breeze is, you know, supposed to be hooked up up to it? There are some. I don't know why and I'm not entirely sure what you'd have to ask the team, but there are some like security level permissions issues with how much data exactly. You can see out of every single record with Breeze with the assistant currently, but I'm sure that's gonna eventually like move and go away. I don't know like exactly how much, but the problem is it's a bit of a black box. So the reason that I don't do this inside of HubSpot is because if you give this to your assistant, you can't really watch the progress to see if it's looking at exactly the right data. You can't force it to tell you to bring back all the deals that it looked at in order to make the assessment. So it makes the oversight of it very difficult. Even if the data were to come back, I'd still want to see like how you track the data. I'd still want to be able to see, okay, how did you make that determination? How did you make that decision? I'm so glad you brought that up because I am like I and I should have had it ready for today, but I am going to be building a ACP agent through agent AI. So nice. I'll be posting that as well. I am not going to say that that tool will be definitive. One thing that I definitely want to throw out there is and what makes this like something that's not open season for everybody, right? Because I'm sure another question is like, well, if everybody has this then nobody loses in the marketing game. It's not so much that as much as it is, once you start using it, you're going to understand how you need to tweak and put different weights on the important importance that you give to certain things. So maybe your sales notes are really strong so you give a higher amount of value to your sales notes or maybe there's some other aspect, maybe tickets are strong or something else that gets weighted into your decision making process for deals closing and you're going to want to adjust those and you're going to want to apply some, a little bit of math to it. Not that you have to do it, but you're going to want to start to apply some point levels to allow the model to create again, even a sharper spear than you started out to begin with. So that's one reason. And then outside of that, I mean you have the connection to everything else, right? So I've got my connections to notion, I've got my connections to drive, I've got my connections to all of my other third party sources. So I can use a lot of external resource on the information that I collect as well. So it's, you know, it pulls all that information into one place where now I can do whatever I need to with it as opposed to having it not. And I'm not trying to say not that like don't use assistant for stuff. I just mean like this is a heavy handed thing that you want to use a specialized tool to do basically the same. Best way I could answer that.

That makes sense to me. Zach asks the evidence step seems to be the secret sauce. Do you have any tips on how to get AI to recognize behavioral patterns in meaningful properties? Building in our assumptions Previous attempts have led to AI asking which properties to look into, which is what I'm hoping it can help me identify.

So the the other prompts that I'm going to post do have some and I'll actually let's see I'll post an example let's say like an example output would be something that shows so let's get all this text looks something like this no copy paste doesn't even want to work today. That's awesome. I swear that's not blocked by Cloudflare Flair. I will post the prompts and you'll definitely see it's geared for HubSpot so it's got HubSpot properties that you can use by default. What you're touching on is the secret sauce for for sure. It's kind of like what I was talking about earlier about creating custom weights for the properties that you want to use. That is what makes it specific to your business and specific to the way in which you do your business. In terms of properties obviously like the standard issue stuff right? In terms of like deals, companies and contacts. But yes there is definitely value in creating some custom properties that take multiple values into account that you can can use formulas with to get some data that you would otherwise not be able to get. There are even and so when I say that this is a basics framework that can build in a lot of different ways that's one another is that you could again create some additional properties that you would need and then use your GPT or CLAUDE connector inside of HubSpot within workflows to enhance that information and and do some external searching, bring that back, put the pull it together with the information you've had, put it through a custom script and generate some data that you wouldn't normally be able to get. A lot of different things that you can do to essentially come up with new formulas I would say new ways of being able to measure your customers in a way that makes sense to you. Because I know a lot of the problems when it comes to scoring people is okay we need to score them based off of the same metrics that we've always had and somehow we need to figure out everybody else's differences based on the same metrics. And so this is really a great way of you being able to define what your metrics actually are and what they actually mean and how much they mean. Because once you start dissecting, okay, well, here's what the patterns are in the deals. Here's what the patterns are in closed one versus closed loss. Okay, now we understand this is, this is the way that we need to approach the system. Here's the people that we need to do the handoffs to. Not only that, but based on that, now we know what properties they're going to be looking for and because of what they've done in the past and what matters to them. So now we can go back and retroactively apply that to our deals and we can take a look at how those deals now match to what those reps have been doing. Or it can be the same way for support. Right. Support tickets that close with a high success rate, that are short. We can take a look at what those types of support tickets are. Obviously we can take a look at the data within them. And normally, like especially within HubSpot now you're able to automatically build out knowledge based stuff, which is fine. And that's knowledge based data. And it's awesome. It's externally serving. What would be internally serving is for you to actually understand those problems at a higher level. Why? Because even though your knowledge base is filling in gaps, are you paying attention to them? What are those gaps? Are they, are they customer gaps or are they product gaps? Are they service gaps? Right. Like it's worth it to understand this information and pull it all together to be able to figure all that out. That's exactly what you're doing. It's not so much a matter of being able to, you know, again with, with the assistant, you know, I don't know what the power of it is. Like, I just know that you definitely can't grab seven objects and cross reference them at the same time. So. But yes, I will get some detail. Not a ton, but I will give you some really good starter detail for in addition to these prompts. Great.

Michael posted a question and as soon as he posted he said, oh, as soon as I posted it, it got answered, but it's gotten a couple of upvotes. So I think it's worth revisiting. How many closed one deals do you need to get started to identify reproducible patterns?

I've seen really good results once you get towards 20. So like between I, you know, I would say 20 to, you know, 18, try not to go under 20. I've had really good results with 20, so 20 or more seems to be like a really good stable point. And that begins to start telling you at least a small breakdown of like, hey, here's, here's the small channels that we know so far. Right. And again that. So like 20 deals. Oh my gosh, like what is that? Could be a startup. Startups want to know where to go to. Right. And they're not. Again, you're, because you're a startup, you're focused on doing business, you're not necessarily focused on stats, you're definitely not focused on reporting. You're too busy doing deals, you're too busy putting stuff together. I know what it's like. So having that reporting able to be done for you, especially automatically. Right. So of course you start with, you know, a GPT experience distraction. But if you take it further down the road, you can also automate this process so that it can be done on a regular basis so that you are constantly keeping up to date with who your actual customer is. That way you're able to track over a long period of time, like over a year, you can now begin to see not only who they are, but what is it that they are going towards and what are they going after. Right. So maybe they're buying through your business a journey that you didn't even necessarily know they bought a product in the beginning. Maybe they bought a product from you six months later, maybe they bought a service from you a year later. And it turns out this story, when you bring it back to you matching all these other people, it's like, oh, that's how mid, that's how mid level businesses shop with us. Oh, okay. Well now we know exactly where they're going to go and what it is they're looking for at each purchasing point. So if we get a mid level business that comes in and they're looking for this, we know this is going to be the next thing. Yeah. And I don't know what to tell you. These might seem like really easy concepts. It might sound really easy. I can't tell you how many businesses, high level businesses, millions of contacts, don't have this information, don't look at it, they don't think about it. They just go off of what they think they know. They go off of what they think has been working. They watch their reporting week to week. How many of you guys are still making marketing decisions on a week to week or month to month basis, but not looking at like the last year to try to find out the last year's trend. It's time to really, it's Time to like dig into all that, bring it all together. Really start to understand what's going on in your CRM. Really start to understand the value that's in there so you can get rid of all the crap. Right? If you just focused on the value and you got rid of everybody that you lost deals with, then all of a sudden you can put your, your sales team in the right place. And what your ops team should be doing is looking at all the lost deals your ops team needs to be digging through and finding out like, okay, what happened here? Why did we lose these? How could we turn these around by comparison to the people that we won? Because maybe it was a lost deal. Again, there was like a little hitch decision that happened, or maybe it was, maybe it was bad timing, but there could have been just a week period or something. You know, one thing that GPT is going to be able to do again, it's a simple framework so you can expand from there is it's going to be able to take real life world events into account. So maybe we can look at stock market dips and decisions made off that. Right? It's, it's a aspect of being able to take a look at shopping habits. Maybe overall, maybe there's market decisions, maybe there are political decisions, maybe there are even ecological decisions that are getting made. Something related to climate. Right. Customers shift all the time, the behavior changes all the time, and it's due to a lot of different factors. So it really is like super important to be able to look at the same data from as many different vantage points as you possibly could.

So you, you touched on marketing a little bit in there. KK asks how ACP and AI can help with campaigns.

Well, it can help with campaigns in terms of knowing exactly why they bought, what they bought, for what reason they bought. It tells you what you need to market, why you need to market it, where you need to market it and who you need to market it to. I know, I like, I hate boiling, like making it all crazy simple, simple. But guess what? That does. Okay, so like you're, you know, you're probably sitting there like, well, there goes all my stats time and my reporting time. That's what I spend all my time doing. Okay, well now focus on the design. Focus on making it awesome. Focus on the channel that it's in, maybe the design of the page, like all the stuff you couldn't get to before because of the minutia getting in your way. Do it, get it done right. Like, I'm sure there's plenty of redesigns and changes and all sorts of stuff that should have gotten done. But there's daily stuff that gets in between. And the daily stuff is a matter of making decisions. And those decisions should be getting made based off of traditional behavior based data. And if you just did that now, all of a sudden you can get back to actually moving forward, expanding your business, doing the scaling that you should have been doing already.

Love that. So I believe that is all the questions we've received. Unless anyone wants to throw one in last minute. Do you have any closing thoughts you'd like to leave us with?

Closing thoughts? Just, you know, I don't know how many people that I've worked with so far with AI, teaching them, training them, showing them what's possible. And all I can tell you is that in every single instance, as soon as somebody tried it once, that's all it took. Most of the people that I've trained, I sat down with them for 30 minutes and that's it. I've never seen them again. And that's the best feeling in the world, honestly. Like, if I'm training you, I don't want to see you again because if I do, that means I failed, right? But if, if I never see you again and all I hear from you is like, hey, this is this cool thing I did, or look at this thing. All right, awesome. Now you know, mission complete. And so I can tell you, it's just, it's been the same way in every single case. Whether it's learning it from the beginning, whether it's going to the next step. You know, I only have four tenants when it comes to working with AI. The first one is treat it like it's a person. Because it basically is. You're talking to a brain. It's a knowledgeable brain. It knows everything, right? So being that it's like a person, every context window that you open with it, unless you've used personalization and some little details, it's like talking to a brand new person. It's like onboarding a new employee. That's the way I like to look at it. And in the same way that you would onboard an employee to what it is that you're doing, falls into tenant number two, which is context is king. So if you were talking to a new person that you needed to help you do your job, you're not just going to tell them, hey, do this thing. They have no idea what you're doing, why you're doing it, what reasons, what else might happen. You know what you don't want. There's that aspect too. They might know what you do want, but not what you don't want. So now all of a sudden you spend all of your time trying to sift through what they did to correct them and get to what it is that you wanted. Context really, really matters. Third, they know all of the things, so stop acting like they don't. And if they, if it doesn't have the right answer, just kick on web search. You'll be alright. Okay. Each tool has its own purpose as well. So there really is to me, no one size fits all solution. I am very agnostic to it and I find each one has its strength in different places. So it kind of depends on what it is that you're trying to do, what type of task you're trying to accomplish. And number four is always take it one step further. So when you get to the end of what it is you've done, maybe you're, you're prompting and you're trying to come up with a good prompt for something spits out the result and it's close to what you wanted, you know, maybe you wanted a little bit more. Have you ever thought to ask, hey, is this the best this prompt could be? Or can it be more efficient? Can it be better without as much, you know, as, without as much wording? And the best word to use is it is the ambiguity low enough? Ambiguity is like the biggest enemy to language models. So one of the things that you really want to do is, especially when it comes to prompting, hey, put this together with as little ambiguity as possible so that it is understood as straightforwardly as possible. That eliminates all those follow up questions that are like, but what about this? But what about this? And what about that little detail like yeah, just make sure that the instructions in and of themselves are impossible to screw up and that way we're good from the get go.

Great. Thank you sir. If anyone here wants to find you, connect with you on the Internet, where's your preferred preferred place for them to do that?

My preferred place, I suppose is LinkedIn, but you can also find me through Value first, which is a platform that is championing other HubSpot admins and also anyone that's in the ecosphere for CRM that wants to better themselves, learn more, do more. And yeah, I think those are two main spots. I'm on LinkedIn almost every morning that I can be at 8am Eastern, 7 Central. Me and Chris, Carolyn, we do a AI podcast, so check that out when you can.

Amazing. Well, thank you so much and thanks everyone for being here. We'll. We'll call it a day there and hope to see you all back next week.

Thanks. Bye, everybody.

Welcome to the admin hook. No flashes, facts and a whole lot of love we get it done accidental or all in by choice admin and flies and lift their voice fixing fields and hacking time making magic with no headline Tiana brings fire in a crew that's here to lift you higher we swap the whims and share the pain no one's solo in this game it's the action hug where legends walk in from chaos to clarity that's how we win dirty got a challenge accepted portal perfection fully connected. Quick fix queens workaround kings custom coded everything tabs open coffee strong you've been solving all along we've got the know how we've got the map even HubSpot product pulls up to chat you're not just heard you help shape the tools admins rewriting all the rules thanks for building with the hug thanks for powering the hub the hind is a campaign for tonight much there's a whole crew hold inside it and it's together powered by insight.