Your Marketing Maturity Model: What AI to Use When | Dreamforce 2025

Hi everyone. Good morning. Thank you so much for joining us here all the way in the content campground. Today we're going to be talking about your marketing maturity model and when to use what AI. It's all about determining what the right AI is for the right task. So before we get started, everyone's favorite slide, a forward looking statement slide just to remind you to make your purchasing decisions off of current commercially available products and services. Thank you so much for joining us here this morning. We're really thrilled to get started. My name is Brett Bothy and I am on the product marketing side of the house and I'm joined by my colleague Heather Stables on the product management side of the house and we support the marketing AI group within the marketing cloud team at Salesforce. So we're extra thrilled to be here to talk through the marketing maturity model. So today customer expectations are changing and that's being driven by AI. Now we have all been using search engines for years. We've been tailoring our exact right SEO, we've been prompting in the correct way, we've been writing our asks and then we're browsing and determining which link we want to call up. But today's landscape, you can basically put half of a thought into an LLM. It'll finish that thought for you and then pull all the recommendations for that for that finished thought and what you should purchase. And why does that matter is because today that means that all customers are expecting completely instant and personal responses for every action that they take. Now as marketers, that's our charter to solve, that's our charter to meet. Those are expectations that we are charged with solving. So marketers today are really in a pickle because we're, we're being held back. We are expected to deliver these super instant experiences, but we still are operating with totally manual processes, with campaign planning that's happening in email chains and one way marketing that absolutely does not adapt to real time customer engagement. And lastly to silo data because data lives in a million different systems and it's impossible to action for your marketing campaigns. But our customers are living completely in the future. They're one in, or excuse me, a third of customers today are prompting every single day and so they're using these AI tools. They're expecting total customization for their experience and we are needing to deliver that. But it's pretty challenging. So for us we have to lean into AI and we have to determine not what this shiniest, most exciting AI feature is, but we actually have to understand what's practical and when can we apply that practical tool for the exact right use case? So that's our job as marketers, is to really understand how and when we can do that. And the way in which we do that is through the campaign life cycle. So marketing is living and breathing through the campaign life cycle. These are all of our sort of activities, activities and actions. And it starts with campaign creation. It starts with campaign planning, where you're determining things with your teams. It starts with composing different content blocks and generation. And then it goes to journey orchestration where you are actually delivering segmentation and journey design and actually making those journey decisions. And then lastly we have campaign activation, where your personalization is coming to life and you are nurturing those leads and you are building two way conversations. Optimization is something that should be happening across the entire campaign life cycle. So in today's landscape, campaigns take about seven and a half weeks to launch. But what if you could reduce that time by about 80%? What would you do with that time? You would be able to infuse creativity and strategy at every component of this campaign lifecycle. You could think about all the tasks that could be taken off of your hands so that you can start to iterate faster and you can start to apply more strategic thinking to different components and optimize everything. So that's really where we're taking things today. And when you think about what AI is going to do and how it's going to transform your campaigns, you're going to be going from completely manual processes and static journeys that are preset and don't adapt, and ultimately broadcast messaging that is completely one way that a customer can't engage back with you. And ultimately campaign analytics that are static and at the end of a campaign after the fact where you can't action any of that on an existing campaign. Now, when you apply AI across the campaign lifecycle, what you're going to see is that you can have now instant campaigns. You can have adaptive journeys that respond to how a customer is engaging with you. You can have scale personalization where you have component blocks that are customized to each individual subscriber. And then you can act with insights with impact so that you're making optimizations throughout your campaign process rather than sort of at the end or one time. So how do we do that? You know, we think about that through this AI maturity model. Now I want you to think about the X and Y axis here, just for context setting. But here we're talking about customization and use case complexity. Now with AI, there are a lot of tools that you can use that have outsized impact for one small action, like applying something to an entire segment that it can decision off of all of that data. But then the more complex your use case gets, we're really talking about specificity. So you want to be able to customize for that very specific use case and deploy an agent or a model that's really built for that task. So that's really what we think about when we're thinking about maturity. We're thinking about one size fits most and letting the model do a lot of the thinking and a lot of out of the box features that are great for that purpose, all the way to a little bit of customization where you can really start to play with different actions and iterate and build agents based on your use case. But each of these components has a place in your campaign lifecycle. Each of these building blocks is something that we can all be deploying and using today. It's not about getting to the third step and forgetting the other two. It's about using all of them and knowing when to do that. So when we think about out of the box models, these are really best for broad use cases that have relatively predictable outcomes. So that's something like segmentation or a particular part of a journey or send time, which we'll get into later. But thinking about an action that you can apply to a broader set and let the model make the decision based off of likelihood too. Now the next step is out of the box agents. What these are best for are specific tasks that don't need a ton of supervision. So when again you think about that campaign lifecycle, where are those tasks that can be sort of taken off of your hands and then actioned by something that you don't need to sort of watch every single second of the day, but you can check back in and make sure that that's happening in the way that you anticipated. And lastly, the customization component here. Now these are jobs that require context and tailored actions. So that's where it's a little bit more custom built for your own use case based on the complexity of what you're looking to achieve. So that's the path to AI maturity. And to really dive a little bit deeper into that, we're going to go into an actual live code demo and I'm going to hand things over to Heather to take us through that.

Yeah, thanks, Brett. So for my demo today, I am going to be playing the role of a marketer at Cumulus bank, which is a fictitious institution. So My team and I are planning a highly anticipated campaign for the new year to target new home buyers and promote some of our financial products that are suited to their needs. And we have really big goals for that campaign. So I want to make sure that I'm really set up for success. And one part of marketing campaign success, of course, is the health and engagement of my audience. In this case, I'm focused on the email channel and I'm starting out by looking at this engagement Persona report which is driven by AI predictions that take into consideration my marketing department's send patterns and my specific subscribers open and click patterns. So the loyalists are the people who are opening and clicking my emails a lot. Selective shoppers open from time to time, but when they do, they tend to click. Window shoppers are those who open but don't really ever click. And those in the win back and dormant Persona, they're not really opening or clicking. And when I look at this report, I see that a large percentage of my audience is in that win back and dormant Persona. I want to see if I can do something to shore up the engagement of my audience before we run that big campaign in the new year so that I have a good place to start from. So I'm thinking I'm going to run a win back campaign and I've got one started here. This win back campaign is going to send some of my most historically high performing email content. This is content that promotes popular benefits of my bank and they have clear call to action and customers tend to click on them. We have high click rates. The content is pretty evergreen, but even though it's evergreen content, I don't want to just like send it to every single customer in this segment. So once I've sent the first email in this in this series, I'm going to target this to my goals of trying to engage that win back and dormant Persona. This is where the Einstein decision element comes into play. So I'm going to select that this is my I'm going to target the win back segment and use that engagement score to create paths that I can send different members of my audience down. I can target them based on things like open or click likelihood, but I'm focused on Persona. Based on that report we saw earlier, I can craft a journey for each of these Personas. But today I'm focused on that win back dormant Persona. So I'm going to remove the other ones from my journey here. I'm going to select one of those emails that I talked about. Let's pick this one. Where customers can locate ATMs. This email is just going to go to those people in the winback segment. But I can use AI to personalize this journey even more by using send time optimization. So send time optimization is making predictions about what the best time during a given week is to send an email to somebody. I think I want for everybody who lands on this step to get the email within 48 hours. So I give it that parameter. But Einstein is going to be making that decision about when to send this at the right time for each of my subscribers in that Persona through this campaign. Now I have a third email that I want to send, but I want to make sure that I am driving engagement with my emails and not unsubscribes. So I'm going to use another Einstein decision element to help me to do that. In this case, I'm going to target my under saturated segment. So in this case I'm using the frequency prediction. The frequency score uses my again my marketing department score send patterns and then my subscribers open and click patterns to determine how many emails every subscriber wants to receive in any given week and making sure that we're not oversaturating them with email. I don't have to worry about what that count is. I can just add in this decision split. And again today I'm focused on those who are in the under saturated, in the undersaturated bucket. And so I'm going to narrow down to that to that path. I'm once again going to select an email message, another one of my evergreen emails to set up this campaign. I'm going to flip over to one that I have finished where I've built out everything with the splits and the weights and all. And this is of course layered on top of any personalization that is done with personalization that's done within the content of my emails and any segmentation that went on before they even enter into this campaign. So I'm going to let this run for a couple of months and then I'm going to check back in on that email Persona report and see how we've done. And looking back on this, we've kind of gotten in a time machine and we've gone forward to look at this engagement report and I can see that I have a higher percentage of my subscribers in more engaged categories than win back end dormant. And so that puts me in a good place to start building that home buying campaign. Okay, so speaking of that home buying campaign, I'm going to turn to the agent force with the out of the box campaign agent to help me get started here. Even though this is a highly anticipated campaign, I am a little bit stumped on getting started and overwhelmed by all of the steps that it's going to take. And so I'm going to start drafting this campaign with the help of AI. So I'm going to just paste in a prompt that's describing the campaign that I want to run and I ask Agent Forest to draft this brief for me. The brief is like an enhanced prompt that's going to help to create the rest of the content that is going into my campaign. It takes what I wrote and it expands it into a key message and a target audience. Now, if there's anything here that agentforce came up with that I'm not totally happy with, I can edit and iterate with it right here within the flow of creating my campaign without having to go elsewhere to another tool or a chatbot. So once I'm happy with that, I'm going to ask AgentForce to get started on creating those assets. And AgentForce is going to create for me several different assets across multiple channels, email and sms. It's available to me in the campaign canvas for me to continue to iterate on, to create, to edit, to ask for more help from the campaign creation asset. And then once I'm happy with how everything is put together, I can activate my campaign. So what have we seen? We saw first of all some of our out of the box models and how they can be used to customize a flow. Very simply, I really love these features because there's powerful machine learning behind each one to help you determine how to target your subscribers based on their likelihood to engage with engagement scoring or the right number of messages with engagement frequency, or the right time to send with send time optimization. But all I had to do to get started with these models was just turn it on with a single click. The models began to be trained with my company's data. And then you saw how easy it was to add those splits into the flow and to target based on these predictions. We also saw the campaign creation agent, which is an out of the box agent that can help to create your campaigns and get to those goals in a matter of minutes, as opposed to the weeks that it can take to manually create all of this content. So where would I go from here if I'm ready to tackle more complex use cases? Well, I can customize either with predictive or with agentic. So in the case of predictive, maybe I would want to go to Einstein Studio. And build a model for who's most likely to purchase or to create a model for what the next best action is for my subscribers. And then I can bring those into my flow to customize the settings segment or the path and target everybody exactly where they're at. But I can also build custom agents that will help advance my customers journeys. Like a personalization agent that's going to insert the right content block into their email. Or one that helps me to adhere to brand guidelines when I'm working with the campaign agent. Or one that can respond to customer responses to email. So, so once I send out those home buying campaigns, that agent could reply to questions that customers send back through the email channel, putting an end to that no reply email address that we're all really used to sending from. Right. So AI can be employed across so many different parts of the journey. It's not just one monolithic tool. And to talk about how you might get started on that journey, I'm going to hand it back to Brett.

Thanks. Okay, so how do you map your AI journey with all of those learnings in mind? You know, first we want to think about your data readiness. AI is only as good as the data that fuels it, right? So we have to really have good data. Next is the AI roadmap and use cases. So you want to just make sure you have a really clear vision for what you're building towards. And last but not least is change management because AI is new for a lot of people, or at least new, new in application. And so we want to make sure we're building practices that support learning and behavior within our organizations too. So to dig a little bit deeper into each of those, you know, we know that customer data is captured almost at every component of the life cycle, but it might live in different places. And that makes it really hard for you as marketers to do your job effectively. So we want to make sure that that is in one place that is actionable for you and your AI to act if. And lastly, we want to make sure that that data refresh is not, you know, it's still living in a 24 hour cycle. It's living in at least an hour or less than that minute to minute cycle because that's how customers are engaging. And once you have that level of data freshness, then AI can really start to become effective with some of those decisioning steps. Second is again those AI roadmap and use cases. You want to define exactly how AI can help you meet your goals. So if you think back to that Campaign lifecycle slide. You know, you definitely are going to have goals or KPIs that you're trying to target at each of those steps. Think about where AI can slot in to those steps so that you can really understand how you can start to play with that KPI and what your exact target improvement is, you know. Next is identifying areas where AI can be most effective. So you might have opportunity for AI to fit in every part of the campaign lifecycle, but you want to prioritize, you know, one to three, use cases to, to get really to get started with and get comfortable with. And lastly, more important, I would say that even other components of marketing technology, with marketing AI, you want to test, learn and iterate. It is so iterative. You know, we've always said that with like an email send or a journey, but with AI, you want to test, learn, iterate for every component that you're trying because it gets better with every time you iterate, every tweak that you make. And lastly, we talked about change management. So this is a mindset shift sometimes, right? Where it's this concept of new skills and really new workflows like we're thinking about optimizing now at every phase, even including building a campaign, which isn't something as marketers I don't think we've always thought about. So you just want to make sure you're sort of fostering that mindset shift that we want to be sort of twiddling with things and tweaking with AI earlier than we ever have been before. Second, there is identifying champions within your team because you want to make sure that you have subject matter experts and not everyone can be an expert on every feature. That's a lot. So just try and pick areas where one team member can own one thing and apply that to your life cycle. And lastly, it's of course about fostering a culture of celebrating early wins and understanding that this is very much a progress over perfection kind of task. So you think about that maturity model. You're not starting at the third step, you're starting at the first one, but you're continually building upon that. So that's what we have for you today. If you're interested in learning a little bit more. We have a new AgentForce marketing trail available today through that QR code, so feel free to scan that. This session will also be available on Salesforce plus, if you're interested in watching when you get back home and you want access to these slides, it's available in your app now to move tomorrow we have the marketing keynote and it's all about marketing that sparks a conversation at 10am where we're going to be diving into some really exciting agentic use cases for agent force marketing. So don't miss that. And thank you for being here.