Anthony Lamot: No, I definitely recognize that feeling. I don’t do any coding anymore these days.

But in the early days for DESelect, even I did and it, it’s definitely a good medium to enter that state of flow as psychologists would call it. And I do think by the way as it as a, you know, a little side, no, I do think even the best engineers have to do this kind of trial art or something to that works organic. I don’t think anyone can just, you know, A to Z type out a whole piece of functional code that will be probably like a freak of a person. Yes. But what I’ve always found helpful is to, rather than just starting to come up with a good architecture breakdown functionality into modules and then work my way through those modules. And, and when you de, structured like that, then… there’s a lot less frustration, and iterations afterwards.

Aaron Beatty: Yeah, yeah. I would agree with that. I, I’m not always as good at doing that, but I definitely have a smoother workflow, when I have a properly documented plan with requirements and, you know, goals and, you know, pseudo code prior to just kinda hopping in. And then, you know, saving early and often and saving iterations, of codes so that I can go back a step or two if I need to or if I break something, I’m just not always as good at being that organized. And sometimes I just want to hop in and get my hands dirty.

Anthony Lamot: Sure. Well, I still have to stay with this example of this project a little bit long because I’m kinda curious what kind of, what do the data architecture look like? Because I’m trying to sort of imagine what do you need to do a communication like that? So presumably, you need to store per customer their most favorite product categories. So that is based on purchase data, maybe with some sequel updates into a table. But then you want to personalize based on layer product. So do you load a product catalog and does that, is that linked to a content?

Aaron Beatty: Yeah. So I mean, I think you, I think you kinda talked it out there. So it’s largely that. So, the main thing that you need, is a product catalog and, you know, if that catalog has, you know, predefined categories. So like, in their implementation, I think they had like, you know, there’s like patriotic tee shirts, there are like music T-shirts. There are tee shirts by like keywords. So like, you know, if you search for teacher, you might get a, you know, a bunch of T-shirts for what it’s like to be a teacher. And if it’s if you search for engineer, then you might get a bunch of different kinds of funny shirts for engineer. So it was a lot of like keyword based products in a product catalog. And then they always had an image that they would show on their website for all the T-shirts. So I had access as to all of that. So between that and, the transactional data that’s really it. And then, and then the sequel is just looking for each person, their last X purchases and then looking at the categories of T-shirts that they purchased, and then trying to kind of discern their top, you know, however many categories based on whatever it is that we’re trying to do. So sometimes you can just use that… implicit preference based on what they’re purchasing. I have had some situations that are kinda similar to that where we’ve been able to combine the transactional… data along with the preference data from their like preference center, and then combine that with their viewing data based on like the things that they’re looking at on the website and then trying to kind of combine all of that through a series of SQL queries, or what have you to try to kinda guess the top things that somebody might be interested in. And usually, you know, the longer ago that they did that the less impact it’s going to have on your final scores and that kind of stuff. So you can make it as complicated as you want to. But it’s usually some semblance, of those pieces of data that you re-combine into an e-mail…

Anthony Lamot: It’s also interesting that you can combine implicit and explicit preferences. And I would wonder then what actually makes more impact if you would, afterwards, if you would like  AB test it and see if people respond more to their explicit or implicit preferences. I’m gonna guess implicit.

Aaron Beatty: Yeah, probably, yeah, because I think, you know, it’s hard to get really granular when you’re doing explicit because, you know, you can, you only want to show so many options on your page like you wouldn’t want, like if we go back to the Teespring example and let’s say they’ve got a 1,000 categories of T-shirts, you know, you’re not going to have them sift through all 1,000 categories and say, yeah, I like cowboy merchandise and my wife’s a teacher. So I like teacher stuff or, you know, like that’s gonna be really hard to find but you might be like, well, they like T-shirts versus hats. So like you might be able to combine the more broad categories with their implicit… either browsing behavior or abandoned shopping cart type stuff plus the stuff that they actually bought. And, and I think the thing that’s hard with that to really get right, which probably nobody does is like when is somebody gonna buy something one time for a specific use versus that’s the thing that they like to buy? You know, like if I’m on if I’m on the home shopping page and I bought a set of sheets or something like that, that’s probably not going to be something I’m gonna buy all the time. It doesn’t mean I just love buying sheets. It means I needed to set of sheets. I bought them. Now, I’m done with that. I don’t wanna keep getting emails about my next favorite sheets because I just bought them. So like it’s hard to, sometimes it’s hard, to square those edges and figure out… what’s actually gonna make sense in the ultimate e-mail and that’s it’s probably impossible to get right right now, but you can, you know, kinda take a stab at it.

Anthony Lamot: Now for Teespring, they probably constantly have new content available. So how do you make sure that you can actually show the last category winners, to those customers? Is it integrate with some kind of external CMS or is there like a process of constantly updating the latest image for the coolest new shirts? How does that work?

Aaron Beatty: Yeah. I think… this has been, it’s been a hot second. So I don’t remember all the details for them specifically, but generally speaking, you know, if we’re looking at top products, you know, it’s likely not going to include anything that’s brand new because there needs to be data to support that. It’s a hot product meaning enough purchase history or whatever to say. Okay, this is a thing that people like. And so it ends up just being, you know, in Marketing Cloud parlance, it ends up being a data extension that’s getting updated every, you know, probably a day or two with the current most popular items of the different categories. And then you take it from there. So that’s at least how that one worked to my recollection.

Anthony Lamot: All right. Thanks for sharing. We’ll talk about hot products. Let’s shift the conversation a bit to all the stuff we heard at Dreamforce. Salesforce has and continues to make major investments in their data cloud. That’s the name of their CDP, formerly known as genie, formerly known as Salesforce CDP, formerly known as Salesforce 360. But where do you ultimately see Data Cloud going? What do you think about that product category? And what do you see in the market?

Aaron Beatty: Yeah. I mean bringing it to specifically Data Cloud or Data Cloud for marketing. I think it’s definitely where they are wanting to bring the product. I mean, it certainly seems like they’re making a lot of investment into that tool and… I hate to say at the expense of, a, like marketing cloud engagement or something like that, but it does kinda seem like a lot of the energy and, you know, the new AI based tools and, you know, marketing GPT and all that it, you would expect it, I think to be in Marketing Cloud engagement or Marketing Cloud account engagement. And instead it’s in Data Cloud. So definitely they’re making a push to make that platform really attractive. And, you know, announcing that you can now get what amounts to a free trial, of the Data Cloud and kind of taking a look at it basically giving people, you know, their first taste is free for that. You know, the downside of that product right now is that it’s not an inexpensive license and it, it’s at current is really, it seems to me like it’s mostly a way, to solve for how difficult it can be in Marketing Cloud to get segments. And so, it’s a way for Salesforce to sell a product to make that part, of another one of their products a little easier to use. But they are, you know, rapidly advancing, the capability of that tool and I think, the kind of end user… the way that it’s being framed is more, you know, you’ve got your CRM data, but your CRM data is and everything you’ve also got data on the web and you’ve got data based on your social networks, and you’ve got data from all of your advertising. And so you’ve got all these pieces of data in addition to CRM and that, that’s a place where you can bring it all together and create profiles and kind of flatten the data and then activate it through journeys or through… let’s see what it’s called now Marketing Cloud personalization… or back into CRM and so… I don’t know. I’m really torn on it. I think it’s gonna grow into a thing that’s gonna be really attractive for folks, but it’s a little bit of a hard sell at the moment for at least most of the smaller organizations. It’s you know, it’s a pretty big lift.

Anthony Lamot: That’s fair. I think. And for those more well-versed in the Salesforce ecosystem, we all know that Salesforce platform really consist of often different infrastructure, different databases because of the acquisitions it has done over the years. What I do find exciting is that there’s going to be, well, hopefully you’ll become a common ground for all these platforms which will make integration a lot easier if not just completely unnecessary in case of certain capabilities. But I think there’s still some where, you know, there’s still some road to be travel before we get there.

Aaron Beatty: Yeah, yeah, I think so. I think, for those of us that rely on this ecosystem for our livelihood, it’s still an open question for where we should be investing our time and effort and, you know, training and, you know, certifications and all that kind of stuff like where are things headed? It definitely, you know, just based on the things that Salesforce has been saying, it definitely feels like data cloud is one of those areas where, you know, to your point, it’s a common ground for development and, you know, the newer stuff is all coming out of, that same platform. So, I think that’s where things are headed. But man, there’s just so much existing technical structure and depth in all of these other tools that for Salesforce as a company, I don’t know how they begin to bring all that stuff together… but yeah, I think that’s going to be an important part of, their strategy.

Anthony Lamot: Absolutely. Another thing we heard a little about a Dreamforce is AI, it was kinda hard to escape that phrase. You already alluded to the phrase earlier to start a conversation although maybe were you were rather referring to your personal use of AI, I’m not sure.

Aaron Beatty: I mean, it’s definitely right now, it’s definitely the buzz word dire, right? And, you know, my standing joke was walk around and go, who’s AI? Why does everybody keep talking about AI? But yeah, I mean, yeah, it’s inescapable, it’s definitely blown up obviously in the last year with chat GPT and then all the other tools, that have come around as a result. I am of the opinion, that it’s I think it will get us to a place where we can really do that one-to-one marketing engagement. And I think it’s really gonna expedite… that transition like I, you know, I referred earlier is the kind of, the holy grail of art marketing is to be able to have, a marketing segment of one and not have to group people together into these giant groups. And so the scale of data processing, and thinking and I use that term either in the, you know, the sense of a human doing it in the sense of an artificial intelligence doing it. But the, with all the tools and capability we’ve had so far, you just, you don’t have the time and resources, to have a segment of one. So it’s a short hand that you create segments and put people that are similar together under the assumption that they’re going to behave roughly similarly. And yeah, I think we’re just going to be able to get to the point where it’s literally, just this is marketing, for Aaron bad. And it’s based on what we know about Aaron bad and what we know triggers him. And whether he’s on his phone on TikTok or Facebook, or whether he’s on e-mail and we know he’s taken these steps in these actions and what’s gonna be the next action that’s going to be most likely to get him closer to our goal. And so, I think we’re gonna get there pretty quick and that’ll be, really great from a marketing perspective and as a consumer perspective, really scary because… as good as marketers are and as good as algorithms are at pushing our buttons, it’s just gonna get better or worse depending on how you.

So, yeah, that’s gonna be that’s gonna be really interesting, to see that happen. And, and from the marketing perspective be part of making that happen.

Anthony Lamot: Have you also personally doubled a lot with trying out AI tools or tricks with chat?

Aaron Beatty: I mean, I definitely, you know, I alluded earlier to being kind of like, you know, probably imposter syndrome, but just kind of feeling like a junior engineer in, some cases. And, and having a tool that can help me proof read that can help me write code faster than I can on my own or faster than I can with a quick stack exchange, search or Google search. It’s it’s really beneficial from that perspective. And then being, a very small company and then having all the responsibilities of, you know, building a website, having support tools, doing, you know, accounting and all the other things that are really just a function of owning a business. There’s a lot of help that I can get, from tools for copy, writing, for image creation, for writing outlines, for presentations. I mean, on and on, there’s a ton, of benefit to those tools. And then, you know, even on the project management side, we use internally a tool called motion that uses some AI tricks to basically manage and organize the tasks that you have do and putting them on, your calendar based on the time that you have available. So there’s some internal like uses of tools like that are really helpful for time management. And then I’m currently taking a class at MIT remote… for large language models and AI and kind of how to build your own. And, and one of the things that I’d like to do is to start to build some tools based on all of the documentation that I have, all of our kind of internal wiki, best practices and things like that and seeing if we can build something either for internal use or something we could share with, our client partners, and offer them something that can, where they can look at our kind of database of best practices and pull out the information that they need via one of those tools.

Anthony Lamot: Amazing! like some kind of marketing cloud knowledge base with which you can interact through natural language?

Aaron Beatty: Yes. Yeah. So I feel like that’s kind of, the first phase of that. I feel like longer term. I think there is going to be a need to have some tools that will do kind of what I was referring to with the one to one marketing. And, you know, if you take that thought out to its logical conclusion… where I have some kind of database data warehouse data, lake, I get lost in all the terminology, but you have a bunch of data somewhere and then you have some AI tools that sit on top of that to help you to figure out this cookie was when he went to this website. And this, you know, transaction was also Aaron because that’s tied to his credit card number. And, you know, some kind of AI that’s making sense of the data. And then maybe an additional layer of AI that sits on that to kinda analyze it from a, almost a psychological perspective like, a human behavior perspective, plus marketing. This is like they did these things. Now. They’re most likely, you know, kind of getting into generative or predictive AI like this is their next most likely thing that either we want them to do to increase profits or we want them to do to donate to our calls, or we want them to do to go to our university or, you know, whatever the use case is and then to send them a message through whatever channel makes the most sense to try to move them, you know, one step closer to that goal. And if you’ve got a system that can do that, then you don’t really need much in the way of an AI. You don’t really need much in the way of like a preplanned journey or automation. It’s really just the system kind of looking at each person and seeing if there’s any communication we can send to them that will increase the odds of them doing what we want them to do. So.

Anthony Lamot: It actually ties in to some stuff we do as well. We are already integrating a bunch of AI capabilities, you know, into our product. In fact, for this segment, we are already in alpha with people using natural language to create segments and not just that generates the SQL but it’s actually, you know, visible in our EDI. So also non technical users can really understand what’s going on. So it’s kinda cool because we just want to integrate speech to text so soon you’ll be able to just shout at your laptop to build a segment. I can, to do that.

Aaron Beatty: Right. I already do that. Actually, it just doesn’t listen. It’ll so that’ll be great. I’ll be like hang it, do this thing. It’ll be like, okay, it’s done. I’d be like great. That was.

Anthony Lamot: I can see, I can see you go“why don’t’ you listen?” and then it tries to explain very meticulously why it’s not listening.

But one one thing that, that’s also interesting to me is doing more campaign planning and optimization that’s what we’re doing with Engage two people using it now to plan their campaigns, making sure the segments are not being over-engaged with or over-saturated, right? Marketing frequency. But we’re now also introducing not generative but predictive either to allow people to still set rules but to also let the system come up with an ideal send volume per contact. And then even beyond that, thinking about maybe the marketer opens the platform and we can just go hey grade up on all those campaigns. But did you know this specific audience is under served? And then once this, these are the commonalities. And then two, these are recommendations, that we would give to you to do campaigns around like to do a campaign around this product category, probably have this kind of response. So we’re not there yet, but I do think that’s a really exciting vision to work towards. And I think that’s a bit the future of marketing. You wanna offer this… mission control center to marketing operations folks so they can much faster make decisions, and work out your planning. I think that’s really interesting.

Aaron Beatty: Yeah, yeah, no, I think we’re just at, the beginning, of what’s going to be available there. Because if you take, you know, what you just mentioned out as a kind of natural next step, you know, there’s already, you know, chat GPT, and other tools like, you know, Midjourney, and others that can start to create content given a specific tone. And so if you have, a brand tone or a language that your brand uses, but also you wanna make your language unique to the segment that you’re speaking with, you know, it’s almost like earlier when you’re talking about code-switching, for different verticals. It’s kinda the same idea you could do that, for your segments too and still maintain a brand voice, but have it match who you’re trying to talk to. And so you could certainly ask a copywriter to do that. And I’m sure the results would be fabulous. The challenges is the volume of requests and the quantity of segments that you might end up having. And so if you have a tool that can assist with that, even if it’s a first draft, to do that, and kind of go the rest of the way or maybe 80 percent of the way to generating the actual content and then rely on the human being for the last 20 percent to make sure that, you know, there’s not a crazy, you know, third arm or sixth finger… on image and you know, that kind of stuff. And that’ll get better with time too. So, yeah, I think it’s gonna be amazing.

Anthony Lamot: You mentioned that because what I recently did is so, in our team now, I’m still doing a little of enablement and a lot of product knowledge still sits with me. So one of the things, I enjoy doing too is sort of educating our own team. And since we also want to better speak to those codes, to use that same terminology, right? So to have more specific language, I started a little project where I was verticalizing some of our internal documentation. So we have very extensive internal documentation on our products, they use cases and then how that goes into benefits. And I CP, and all that good stuff. But I didn’t have any industry specific knowledge. So what I did was I opened a new session, I pay for the subscription have or significantly better point then I just, yeah, and I just went to our website and I did copy paste all dumped it into like, hey, this is what we do. Read the website and it just reads it. And then I said, okay, now try to think of typical use cases for marketing operations in insurance. And so then, okay, and I do a little bit of fact check like does it make sense? As I understand at I’m, trying to do, okay, now merge what we’re trying to do and explain me in insurance language, what we’re trying to achieve, right? And then I still edited a little bit, but I made that part of our internal documentation. And I’m just going through our, I started with insurance because that’s one of our typical verticals but there’s a number of verticals we serve. So now I’m doing that one by one. So I thought it was a really good use case to use something. Yeah.

Aaron Beatty: Yeah, that’s cool. I mean, and otherwise, it’s overwhelming with the amount of, you know, if you were gonna take that information and create like one pagers, for conferences or take that information and create an e-mail or have a, you know, a subsection or landing page of 

Aaron Beatty: And so if you have a tool that can help kind of, you know, I like the thought of maybe, you know, you put in the first 10 percent, the AI handles the middle 80 percent and then you handle the last 10 percent. Like I like that, it’s like a weird version of the Pareto principle, but, you’ve, got the chat GPT doing like the middle 80 percent of the work and that’s a huge time saver and it’s generally pretty good like, you know, with the proper questions and the right language and the right information the results, are pretty good and you can give it really detailed, you know, requests for what you wanna do.

Anthony Lamot: I did something similar for our page and kind of coming up with target personas and customer avatars and things like that for our business. And, you know, it was telling me that marketing Mike, is looking for help with XYZ and we call the persona “Marketing Mary”. I got Marketing Mary out of chat GPT, and, but like you can dig into, you know, what are, so for this personas, what’s there? What’s their greatest fear? What’s the, what are the things that they’re terrified about? Why? What are the things that they’re that are driving them to make decisions? And so you can start asking it some really interesting questions about what might be driving some of the psychological behavior and like motivations of somebody making a purchase. And, you know, it’s hard to know whether the results are accurate or whether they’re you know, true to somebody’s like psychological needs. But when I read it, I’m like, okay, yeah, I mean, that sounds right? Like I can see myself in that person’s shoes and think, yeah, I’m scared to fall behind in technology or I’m scared with joining with a partner that, that’s gonna just take my money and run or they’re gonna, you know, they’re not gonna give me the service that they promised upfront. And, and so, you know, it’ll lay all that stuff out which you can then turn into marketing and kind of target those behaviors and those thoughts with, you know, we’re gonna solve those problems and not just here’s our product here’s what it does because, you know, most people don’t buy stuff that way. So, yeah, it’s cool. It’s an amazing tool.

Anthony Lamot: You also mentioned Midjourney, I’ve heard several people mentioned I haven’t had looked at it yet, is it significantly better to write content? Because I do find GPT to content it generates can be a little bit generic and dry.

Aaron Beatty: Yeah. So mid journey is more of an image creation tool. It’s and it is, yeah.

Anthony Lamot: Yeah. You referred earlier to some other tool then? My mistake.

Aaron Beatty: Yeah, I can, I did refer to Midjourney but I’m thinking of it more in the context of, you know, if you’re going to build an e-mail or a blog or whatever, and you need both to generate an outline or some content. And then generally speaking, you’re going to have some kind of imagery to go with it. And so Midjourney is a really great tool for that. The weird thing about it though is that it runs through disco. And so it’s a little weird that you have to download like a effectively a chat client and discord, and then install the Midjourney bought. And then you can ask it to do things but, it does a phenomenal job and it, and it’s it craves details. So you can tell it. In fact, you can use ChatGPT to generate a prompt for Midjourney and in get it’s help to get exactly what you’re looking for it to create. So it’s great. But yeah, for me, those are the primary two tools that I use. I know there’s probably, you know, millions of other options out there at this point, but I’m like you, I paid, for, the fancy, your access and, you know, to have API access and I have ChatCPT for and it’s so good. I haven’t really needed, to try any of the other tools yet.

Anthony Lamot: Awesome. Well before we round up, I was wondering Aaron if you have any parting advice you worked with so many customers. You’re taking on new customers? What is something a new Salesforce Marketing Cloud customers should be aware of? Or something like what some good advice you can give to avoid some common pitfalls?

Aaron Beatty: I think, the thing that I see happen all the time… is that… a lot of times a client, a customer will be looking to replace their CRM and build out marketing cloud and do all that kind of all at the same time like they get a nice big budget or grant or whatever to go build this thing out. And what I see happen all the time that drives me a little bit crazy is a ton of time effort and resources will be put into the CRM side of that equation. And then the after thought is, well, what are we gonna do with that information? Once we get it all centralized? And so, you know, inevitably, you know, marketing cloud or some kind of outbound communication thing is one of the systems that will utilize that CRM data. And so as somebody that works in that space, it’s always frustrating when, you know, let’s say a company has a 1,000,000 dollar budget. You know, they’re gonna want to spend as close to a 1,000,000 dollars as they can on the CRM. And then they go, yeah, we also want to be able to send emails and so that ends up being the 10,000 dollar project to the 990,000 dollar CRM project. And it, you know, not to say they should be 50 50, but they’re definitely should be a lot more investment in the, what are you gonna do with the data? And in addition to that, I think it’s important to consider what are you gonna do with the data first? And, and for a lot of these setups it’s last, it’s, the initial investment and time and resources, is spent in what data do we have? Where is it? And how do we structure and organize it so that it’s all in the same place that’s effectively what you’re doing with the CRM, but they’re not thinking about what are we gonna do with it from there? How are we gonna use it? You know, what are our use cases? And, and so I feel like a lot of places get that backwards. They need to, they need to do that first, and then go work on.

Anthony Lamot: I think that’s very sound advice. I would also second that by saying that marketing cloud ultimately is a data driven product. And so you do have to start thinking about the data first. And then all the good stuff can follow. If you start with that, Aaron, it’s been a pleasure. It’s again a great pleasure, to speak with you. I feel there’s so much more we can explore but our time.

Aaron Beatty: Absolutely. Thanks so much for having me. It was great.