Mandy Hornaday: The world of marketing is changing faster than ever. And at the center of that transformation is AI. But let's be honest, for most B2B marketing leaders, it still feels a little overwhelming. We're scrolling past LinkedIn posts about agents, workflows, and automations, while just trying to keep our heads above water. Get the next campaign out the door, and maybe write a halfway decent prompt in ChatGPT.
Mandy Hornaday: That's exactly why I invited Nicole Leffer to join us today. Nicole is one of the most practical, insightful, and down to earth experts on AI for marketers. She's not just talking about the future. She's helping marketing teams actually use AI right now to save time, get better results, and work more strategically. Her journey? Wildly inspiring. Nicole went from running an e-commerce chocolate company, yes, really, to leading marketing at a B2B SaaS org to now teaching thousands of marketers how to work smarter with generative AI. She's trained CMOs, led enterprise workshops, and built automations that would make your head spin, all while keeping it incredibly relatable and actionable. In this episode, we dive into the real state of AI adoption and marketing today beyond the hype, what the most advanced marketing teams are doing, and how they got there, how to start using AI as a strategic leader, not just a content machine, and why the best way to learn AI is to embed it in your daily workflow. So whether you're curious, skeptical, or already experimenting, she brings the clarity and confidence every marketing leader needs in this new era. Let's dive in.
Mandy Hornaday: Hey, Nicole, welcome to the Growth Activated podcast. We're so excited to have you here today.
Nicole Leffer: I'm so excited to be here too. This is going to be a fun conversation.
Mandy Hornaday: Super fun. I have been looking forward to this all week because of course AI is the topic. And I'm certainly an avid user of ChatGPT and it's sort of like my assistant and my brainstorming partner day in and day out. But when I scroll LinkedIn, I know there is so much more I should be doing that I'm not doing. And there's this growing anxiety of how do we bridge the gap between what we can realistically do, you know, today as one person and how do we keep up with our competition? So I could not be more excited to learn from you today.
Nicole Leffer: And you are not alone. So I'm excited to be able to share all of the things that I am learning as I go as well.
Mandy Hornaday: Awesome. Well, hey, Nicole, let's start with a little bit about your background. Share with us what your experience is in and sort of what has led you to the AI world.
Nicole Leffer: Yeah, so I have a very non-traditional background for somebody who is doing what I am doing. I actually started out when I was in college. I started a chocolate ingredients company. And this was in the early 2000s. And I decided very early on this should be an e-commerce company, which was like unheard of back then. So I started selling chocolate on the internet. B2B and B2C.
Nicole Leffer: And it actually started as B2B and then we bridged into both and really ended up growing an e-commerce business, which even though we were selling chocolate back then was more of having a tech company because we had to figure out like how the heck to have a platform to sell on the internet, bringing in developers to build out the functionality and all the things that we wanted. We wanted people to be able to see recipes and all kinds of stuff on our website. So our website almost felt more like a SaaS platform back in those days.
Nicole Leffer: And I did that for 15 years. Ran all the marketing and e-commerce and everything for that business. In 2020, went separate ways from that part of my life. It was time to move on. My accidental college business that lasted a long time. And I ended up becoming the head of marketing at a B2B SaaS company in 2021. When I joined up with them, I stumbled upon a generative AI writing tool very early days, because this is 2021. But I was like, oh my gosh, that's so cool. We are getting this for my team. So obviously, I'm an early adopter. Chocolate was early 2000s, AI 2021. Brought it on for myself and my entire team. We started using this all day, every day. We're trying to really hack around and figure out how you could use it, even in ways it wasn't intentionally designed for, and saw huge results for our marketing team.
Nicole Leffer: And about a year later, we went separate ways from that company and decided to learn as much during the time of transition as I possibly could about really, what is this technology? What else is out there? You know, really trying to understand that. And of course stumbled upon OpenAI and got involved in their world. They're the makers of ChatGPT. So I was on their email list when ChatGPT came out, got the invite, and was on there the first couple of days.
Nicole Leffer: And you know, I've been sharing this in a community of tech CMOs that I am a part of. And I was just going on about this whole time, and I had been crickets. And then they started hearing about ChatGPT elsewhere. And somebody said, I think that Nicole knows something about what this AI thing is. Nicole, would you mind getting on a Zoom call with a few of us?
Nicole Leffer: And I ended up getting on a Zoom call with, I think it was about a hundred CMOs that ended up on that few of us call. And suddenly they were like, well, you turned my team on this right after the call. And so I was like, yeah, that sounds fun. So I started doing that. That was over two years ago now that I started training marketing teams at mostly B2B SaaS companies on how to use this tech.
Nicole Leffer: I've expanded beyond just SaaS. So now it's a lot of different B2B. And sometimes in tech, it's B2C companies because I do have both backgrounds. But I've been doing that for a couple of years. I have a course that I ended up creating last year and keep updated that's Foundations of Generative AI for B2B Marketing, but very deep into working with all these marketers that are hands-on adopting this tech and helping them understand how to use it and also seeing what is and is not working for them. That was way too long-winded of an answer. I'm sorry I talked too much.
Mandy Hornaday: It's fascinating. So many parts of your journey are fascinating. I love a chocolate shop. That's amazing.
Nicole Leffer: I used to live in Brooklyn. I don't talk about that often, so nobody knows I have this background in chocolate.
Mandy Hornaday: It's incredible. And it's also incredible, my ears perked up when you went from that straight into B2B SaaS as a head of marketing. I mean, that's an impressive career pivot as well, because I do talk to marketers often who are in B2B services and they want to shift into SaaS, and it's really hard to break in if you don't have SaaS experience. So that's impressive as well. And I just love how your business has organically grown over the last year. It's amazing.
Nicole Leffer: Yeah, it's been really fun.
Mandy Hornaday: Yeah. So I would love for you to give it to us straight. I am so curious, what is the current state of AI with B2B marketing teams right now? You must see so many different variations across the board. Tell us, as someone who works with a lot of B2B marketing teams. I see varying levels, and I would love, with the volume you're seeing right now, for you to walk us through what's actually happening out there.
Nicole Leffer: We are seeing a tale of multiple different paths that companies are on. I've been saying there's a V in the road or a Y. People are going down one of two paths so far and they're getting further and further apart. I would say maybe it's like a W, there's a third path at this point that's emerging. We have teams that are still burying their heads very deeply in the sand and doing nothing, right? I'm not talking to nearly as many of those, but I definitely know from communities I'm in that they exist, where it's like, we are not even allowed to touch AI.
Nicole Leffer: Now, a lot of those teams, they don't necessarily know it, but they do have members of their team that are secretly using ChatGPT accounts that are personal accounts and doing stuff. I talk to marketers all the time that are like, I just do it on my own devices. And then I sit there and type it in like I did it myself. So I highly recommend not falling into the camp of we think it's scary, therefore we ban it, because they're using it anyway. They're just not using it in a regulated way with tools that you've security vetted, any of that. But that side does exist.
Nicole Leffer: Then in the middle, and where I would say I think most teams are right now, is you have the teams that have gotten to the point where it's like, we have a set tool that our team is going to use or a couple of tools that are approved by the company. For the most part, what I am seeing is one of three tools. It's like a core foundational tool. It's either a ChatGPT approval. The recommendation I have is either Teams or Enterprise. And that's usually what most of them have if they're at the point that they're approving the tool.
Nicole Leffer: So they're using a business account for ChatGPT or they're using Google Gemini for Workspace if they're a Google shop. And it's like, oh, it's right there. We already have it. It pretty much essentially comes with our subscription. So as soon as Google did that, a lot more companies embraced Gemini, or they're using Microsoft Copilot if they're a Microsoft shop. I would say the vast majority, that's what I am seeing fall into that camp.
Nicole Leffer: Now, most of those are doing some interesting things. They're usually a little bit more beginner level. Some of them are pretty deep into only having that one tool, but they are a little bit more on the intro level. And then we have on the far side these teams that are all in on adoption.
Nicole Leffer: They've probably had some of those core tools. They didn't wait until 2025 to approve a tool, or even 2024, well, probably 2024. But they were early adopters, and we're all still adopters, let's be honest. But they got it, and they're already well-versed in the core tools. They're building out really in-depth custom workflows, they're starting to explore automations, they're starting to build in way more complex AI workflows. And those teams, I mean, if you look on the two ends of that and the gulf between the not using it at all and the ones that are like 100% all in, and I'm hearing stories from B2B companies, they're like, in two days.
Nicole Leffer: We did with ChatGPT, and I'm not even talking fancy, well, I mean, ChatGPT is a pretty fancy, sophisticated tool, let's be honest. But with ChatGPT in two days, we were able to do the equivalent of a $100,000 six-month consulting project that we would pay an external agency or consultant for. It would take six months, it would take 20 people to do. And we did it with one person internally with ChatGPT in two days. And it was just as good, if not better, than we would have gotten from an agency or a consultant.
Nicole Leffer: So that's the gulf I'm seeing. You could have a project that would take that kind of time done in a couple of days with a tool that costs you 30 bucks a month or even the version that costs 200 bucks a month. It's wild, the gap that we're seeing. Companies building out automations that are doing BDR type work for them, SDR type work for them, competitive intelligence research. That's not even agent type stuff, but doing competitive intelligence and persona research. They're getting a lot deeper than just, I'm writing a blog post with ChatGPT.
Mandy Hornaday: And that's so great. Because I was going to ask you, when I think about it, I'm probably somewhere in the middle of the second and the third. I'm definitely not to the point where I'm doing automated workflows and sort of deep, deep research and things like that, that I'd love to be doing, but I'm certainly using it more than just social posts and blog posts and just polishing emails and things like that. But yeah, it's so fascinating. Even as a fractional CMO and on the consultant side, it has been fascinating to me. When I think back, when you're putting together a really highly strategic marketing plan and roadmap, it used to take us at least a month, a solid month, to do full competitive research, market research, you know, really understand what's going on with your customers, make sure that you're going after the right personas and the ICP. And it's like, oh, it feels like it could be done within days now.
Nicole Leffer: Yeah, you can definitely do that in an afternoon if you want to. Now you're going to need some time for human review and polishing and stuff, so maybe you need a day or two. But it's crazy, if you really do want to lean into it and take advantage of what is already in our pocket. I mean, most of us don't have experts unless you are such a huge company with unlimited budget. Most of us don't have access to experts in every conceivable topic under the sun.
Nicole Leffer: There is definitely something to be said that when you are the expert, you understand what is good or is not good about what is coming out of something. However, if you're not going to consult an expert to begin with, you're just going off of your own ideas, you're going to get a heck of a lot of a step up from the level you're getting out of a good, highly trained, really smart AI model versus just pulling it out of your own mind. Because it has something it is basing its stuff on, and you're probably, in a lot of topics, you're just guessing. And it's not just guessing.
Nicole Leffer: And so it's a really interesting thing. There is definitely something to be said for the experts using the AI are going to be even more accelerated. But for a small business or somebody that doesn't have access to an unlimited budget to bring in gazillions of people who know everything about everything, it's like the most playing field leveling technology I think we've ever had. And if anything, I think right now smaller businesses are at a greater advantage than bigger businesses with this tech. Which has been the most eye opening part of all of this for me.
Mandy Hornaday: Yeah, no, for sure. I mean, even as a solopreneur, I feel like I can compete with bigger teams in a lot of ways. And I would be curious, Nicole, for the teams that are in on the highest level of adoption, doing the most impressive things, how much time are they investing in up-leveling, in up-skilling within AI and focusing on, I guess, because I almost feel like you'd have to just submerse yourself in figuring out, okay, how do we best automate different processes? What kind of agents are going to be helpful for us? What's out there? There's so much to learn, which is obviously why you are in high demand. But how much time, I guess, are people diving in and leaning into learning about AI in order to achieve that level of adoption?
Nicole Leffer: It depends on when they started, to be honest. So the ones that started earlier have had to invest far less time. And this is actually a reason to really get off the sidelines now versus waiting so much longer to do it.
Nicole Leffer: When you only have to learn, if you started when I started, it was very simple, right? There wasn't much to it. It was very simple. And when ChatGPT came out, it used to just be like, there was a place you talked. It was literally just that chat interface. There were no extra buttons. It was literally just like a text message. And now there are like millions of buttons and different models and different things you have to select. Well, that's a lot more learning. And they added one or two of those things at a time.
Nicole Leffer: So if you started at the beginning, it's not hard to kind of add on and get up to speed. I mean, it is because it moves so fast, but not like it is if you just started from scratch day one today. The best time to start was three, four years ago. The second best time to start is now, if you haven't already. I would say you could, in a matter of, I don't know, my course time is about eight hours of content to go through. Now, people watch some of the lessons more than once and stuff like that. But people who have gone through my course are doing mind blowing things.
Nicole Leffer: So it's not like they spent six months of full time study trying to understand it and do some really incredible things. If you build some foundational skills, you invest 10, 12 hours in really deep education. But even I have people who have done just my two hour trainings and they're doing amazing things. So there are different levels of how deep you want to get into it. Now, it's not going to be the only education you're ever going to do. But once you have a really solid foundation, then it really starts to become the AI gives you your time back for the time you put into it. Right? So at some point it just becomes, and it should, the earlier the better, it becomes: what do I have to do today? How can I infuse AI into this?
Nicole Leffer: And then once you start getting in this habit of everything you do, you're starting to use AI. Even if you think it might take longer because you do it with AI, like you have to practice it. So you start out, you go, I'm going to risk spending 10 extra minutes because I'm learning from that. And then most of the time you're going to save 15, 20, 30 minutes. And sometimes you're going to save five hours, and you're going to save five weeks, five months.
Nicole Leffer: Well, when you start getting to the most of the time you're saving 20, 30 minutes, and then you start saving multiple hours on things, that compounds. So the time that you're putting into doing the projects, which is really your learning, the best way you're going to learn, once you have a basic understanding of how it all works, the best way you're going to learn is by just doing. And the good thing is the doing pays itself back.
Nicole Leffer: Nothing else that we've ever learned has really paid, except for maybe automations prior to AI. Maybe that has paid itself back. So then once you're opening up all that extra time, then you can start learning how do we start automating some of these things. But actually the work you did in your workflows before automations, that just becomes a foundation that you transfer to the automation. So it's kind of this interesting snowball effect, where it is a big time investment, but not in the way that people think, that I just have to carve out time on the side to do AI.
Nicole Leffer: Now, if you're in a public corporation or even a very, very large private business, the bigger issue on time is going to be the meetings upon meetings upon meetings to deal with figuring out how are you going to adopt AI and how are you going to learn AI and how are you going to implement AI. That's where the intense time I'm seeing for teams to get up-skilled and moving is. Not as much in the actual doing. If you can actually, instead of doing that, just be learning it, it's not nearly as much of a time commitment.
Mandy Hornaday: Yeah. And when I think about it, that's super helpful, by the way. I love the idea of just get in and use it every day for all of your tasks and really learn that way. Because one of my questions for you was going to be, are there foundational things that need to be done first as a part of your roadmap? Like making sure you've got really strong data hygiene, if you will, or making sure you've got strong automations in place. I was just curious. But talk to those out there that have got maybe a marketing team of five, maybe 10, and they want to go from that middle part of the W that you talked about to one that's high adoption. What would some of the steps be? How would you outline very specific steps for them to drive that level of implementation and adoption?
Nicole Leffer: First, I want to answer the question about, do you have to have data hygiene and that kind of stuff? Yes and no, right? And the answer I feel like I'm supposed to give as an AI consultant is, yes, you have to get your data in order first. The reality is a lot of people aren't starting because they're so worried about the overwhelm of getting their data in order. And these tools are getting better and better and better at being able to sift through some of the stuff. So to a degree, you do have to have your data in order. You have to have stuff to be able to give it.
Nicole Leffer: But also some of the stuff you need to be able to give it, for a marketer, is stuff the AI can help you create. So it depends on what you're trying to do. If you want to do genuine data analytics and predictive AI analytics and stuff, then yes, that kind of stuff, you have to have your data in order. But to be using it for half of what you're trying, 90% of what marketers are trying to adopt it for, yes and no is the answer to that.
Nicole Leffer: It's more, you have to have the inputs to give it rather than some huge data organization process. So that's the first thing. But if you want to start really taking it to the next step, the first thing I would recommend doing is think through, what do you do? And work wise, what's your job? And if you're on a team of five to 10 people, which, by the way, a team of five to 10 people are going to get more out of AI. That's the group where it's like, oh my god, the crazy amazing stuff that it can do for a team of that size. Because if you start strategically using it as that group, just the volume of work that you can do and meaningful work that really moves the needle is just incredible.
Nicole Leffer: But everybody needs to understand what their actual jobs are. Honestly, on a team of five to 10, most people don't have any idea. That's okay. I can't tell you what I do every day. You just do what needs to be done each day to get the things done. But if you really sit down and break it down, you're going to start to realize what you do. You've got to start going, okay, the easiest way to do this to me is either take a voice memo on your phone and start talking to yourself through, this is what I do.
Nicole Leffer: And I'm not saying super basic. It's like you're about to create a blog post. You know you do this pretty frequently. And I'm just using a blog post because it's simple to describe. But you can do this with literally anything you do in your job. When you go and you do it, just tell your phone exactly what you did step by step. So, okay, so I opened the webinar transcript, because I'm writing a blog post based off of a webinar we did. And I opened the webinar transcript and I started looking through it to pull out all of the key ideas that were in the webinar. And I did that by highlighting each of the things and creating a separate document that had all the key ideas I wanted to make sure to incorporate.
Nicole Leffer: And then I'm literally just telling my phone, or I'm telling a Zoom and getting a recording, whatever. And then I took that bulleted list of all the ideas from the webinar transcript, and I turned it into an outline. And then from the outline, I looked at it and I went, what other stuff am I going to have to research or learn, or who am I going to have to talk to and get approvals from? What are the things I have to do to make this actually happen? And then I went and did those things to bring it in. And then I went and I wrote the blog post. But most people say, I write blog posts.
Nicole Leffer: If you say, what do you do? I write blog posts. No, no, you do every one of those things. You need to be clear on what are every one of those things that is required to do each thing that you do. I'm saying document it like a voice memo, then get a transcript, then have ChatGPT put that into a process. Just write it into a process document. Because the first thing you need is to know those processes.
Nicole Leffer: Then from there, if you have basic knowledge of AI, you can start highlighting that and going, wait, now where in these steps can AI do this? And then from there, you can start actually implementing. Either you have pre-written prompts that are ready, that have mad libs, you fill in the blanks, and you just get it to be able to do it. Or you can start building those automations. Because you can start automating these steps.
Nicole Leffer: You'll get the base of knowledge of, this is how a human would do it. But with an AI you might need to do a step that does this because of how their outputs are. You'll start to understand it. And then also, we're not really there for the level of reliability with the agents, which are more autonomous in doing it. But that's coming.
Nicole Leffer: So if you want to be prepared, having all of this documentation of how these processes work is going to help you as soon as that arrives. That is more the data I would recommend getting in order than get our data in order. Understanding your workflows. Number one, because once you understand your workflows, then you're also going to go, this is the data I would need to get in order to start actually automating this. But I think right now people are like, I need to clean up my HubSpot data.
Nicole Leffer: And it's like, but do you? Do you actually, as a human, really do the vast majority of your work by going into your HubSpot and looking at customer level data? Because my guess, if you're a B2B marketer, is very rarely is that the case. For the vast majority of roles, there are going to be the data people. But for the vast majority of roles, it doesn't matter that it's truly cleaned up already.
Mandy Hornaday: No, fair enough. Fair enough. I love your point about the processes. It's funny because I'm sort of a tale of two minds on that. I would say as a VP of marketing in my past life, yeah, I probably wasn't thinking about it, I just know what my job is. I'm probably not thinking about the day to day. And so I can totally see why team members are there. And then I think about where I am right now as a solopreneur, and I do think about those processes a lot, mostly because I'm thinking about outsourcing some of it. But I've had this ongoing question of, I'm sure AI agents could be helping me and then being more specific or cost effective with what I actually choose to outsource. Even within podcasting, right? I take the podcast transcript and then I need to create an episode description and an episode title and then a transcript and chapter markers and all these things. And I use ChatGPT to help me with most of that today, but it feels very disjointed. So it saves me a bunch of time because it's better than me writing it. But what I'm really interested in is the autonomous agents, or having just an agent.
Nicole Leffer: You don't even need it to actually be an agent that is like an autonomous agent to do what you're doing. Over a year ago, I did exactly what you're talking about when I built my course. The first draft of it has evolved since then. But when I built it, I built a Zapier automation using Zapier and the OpenAI API. And what this automation does is, when I edit a course video, I use Descript to edit the course videos. And then as soon as it is edited, I save it to a Dropbox. That Dropbox triggers an automation.
Nicole Leffer: And that automation tells Zapier, it goes through a process where it creates all these different things. It creates a folder and then it transcribes the video, puts the transcript in one place. It creates course notes for the participants. So they have really solid course notes based off of the transcript. It gives that, which is a deliverable for them. It formats that in a way so it's pretty, and it only takes me a minute to clean up the design when I download it.
Nicole Leffer: It gives me a lesson description and learning objectives for that lesson in a separate file formatted in HTML so I can just paste it into my course platform. It gives me, in that automation, an AI voiceover, which I thought, when I built it, that people would maybe want to listen to on the go, the course notes as a voiceover to review. I don't think anybody's ever actually listened to the voiceover, so I don't know if there's really any value.
Nicole Leffer: I thought it made sense at the time. But it makes all of these things. I feel like there's something else that it's making in that same automation, but all I'm doing is saving it to a Dropbox. And then two to three minutes later, in my Google Drive, I have the transcript, the notes, the lesson description, learning objectives. Oh, it makes me a file that tells me all of the follow-up that I need. If I said, I will also provide you with this prompt, or I will also provide you with this homework exercise, it tells me that. If there was a homework exercise, it makes the directions for them based off of what I said in the video. It does all of this on Excel. You have to look at it, you have to review it.
Nicole Leffer: But literally, I did eight hours of course content. If I was to do that manually, as the starting point of the course, I would never in a million years have finished this course. It just wouldn't have happened. It happened within five minutes of every course being edited. Every lesson being edited, I had all of this stuff. And honestly, the course notes were better than anything I would have done as a human. It did so much better at documenting what I was talking about as an educator. I was like, oh my God, forget me ever doing anything like this again.
Nicole Leffer: But it was pretty wild. You can build, and that was not autonomous agents. It was me basically having it run through an automation where the AI, like Zapier, called on the OpenAI API, sent it the prompt, sent it the inputs. And then that feeds, like one thing feeds the next thing in the automation, but the AI is not making any autonomous decisions.
Nicole Leffer: So does it make autonomous decisions about how to accomplish the task? It did not need to do that. I knew my processes, or what a human would do to do this. And I just broke it down so that the AI could do it for me step by step.
Mandy Hornaday: Okay, so probably a novice question here. I love that. I need to go look into that immediately. But how did, because I'm also thinking, there's like Instagram posts that come out of it. There's LinkedIn posts that come out of it. There's a lot of different deliverables where you may have different rules, of course, for each of those different outputs. So where are you feeding those rules?
Nicole Leffer: So this is not novice. This is actually advanced AI stuff, right? I wouldn't feel...
Mandy Hornaday: Okay, well then I feel better about it.
Nicole Leffer: You could probably do it if you're using ChatGPT every day. But in my course, that's the last lesson, is now how do you start to intro to automating that stuff, not the first lesson. If you're just getting started with AI, look at this as aspirational for the future. You've got to learn how to just use ChatGPT and have a conversation first. When you're building that, you're doing it step-by-step. You're telling Zapier, first you do this and then you're going to take the output from this and you're going to put it here.
Nicole Leffer: And then you're going to put it here and you're going to put it here. Well, you could just build into that. Like, now we have the description and all of this stuff. And I need you to turn this into the HTML for the page that it goes on. Or, and then it could be another step. It's like, take that description and write the Instagram post for it, the caption for it. And you could do another one that's a step. So they're individual prompts you're giving. And it can call back, can you like mail my...
Nicole Leffer: Right? Do they still use the term mail merge? I don't even know. That's from like the 90s.
Mandy Hornaday: I've heard of it, but I honestly couldn't even tell you what mail merging is.
Nicole Leffer: Okay. Let's change that example so I don't age myself like a million years. So I'm going to let myself say that. You know when you do the first name that pulls in on an email blast? Where it's just like, when you put that little tag there and tag it as first name and select that as the thing that it pulls from the database. Okay, so the automation works the same way, except the thing that's getting dropped in instead of first name could be webinar transcript, or the thing that's getting dropped into the prompt could be the output from the AI from earlier in the automation, right? So it kind of just gets merged into it. So it's just running your prompt based off of the stuff that's getting fed into it.
Nicole Leffer: Everything that's getting fed into it that's adjusting those variable parameters all came from that initial recording. But some of it got created along the way and gets fed in. So it sounds complicated. It is, in a way, complicated. But it's not really any different than doing it step by step in ChatGPT yourself. You're just having Zapier run it. So however many steps of that process, as long as it started from the core of that transcript of your podcast to begin with, eventually, if you wouldn't have needed anything besides that podcast, neither would the AI. Eventually, you could sort it out.
Nicole Leffer: So it's going, okay, now that I have the description, we create all of these different assets. You could even, and I haven't done this, and now I'm thinking this as we go. You said Instagram was one of them, right?
Mandy Hornaday: Yeah.
Nicole Leffer: Okay. So you have it, you've got the podcast, the description. You have it write the Instagram caption. I would actually, in this context, maybe have it come up with the concept for the Instagram post and what the graphic would be and what the caption would be. Then you take that input, have it feed to the next step. And it's a brand new prompt. And this one is just to write the caption. But it needs to know the context of the image to write the caption. And then you have it feed that caption with the original description of the entire post. And you have another one that's going to write your prompt for the image for the GPT-4o image generator. And then you take that.
Nicole Leffer: Now, as of while we're recording this, I don't think you can use the 4o image generator in Zapier yet. I don't think it's in the API yet. I could be wrong on that. It's hard to keep up. Let's assume you can't. That could just be one of the things that it sends you, is that prompt, and you could copy and paste that prompt into ChatGPT and have your graphic. It can get to that point where you're now not even having to do the graphic design or anything. It's just about starting to understand your processes.
Nicole Leffer: We've gone off the rails with this example. However, let's bring it back to where we started. That all starts with, let me just tell my phone my workflow. Right? Because you can't do any of what I just said if you don't know the process. And the way you know the process is talking through, how do I do this? And then the easiest way, get the transcript, have ChatGPT write you the documentation.
Mandy Hornaday: Wow. Well, and even as you were talking, I was thinking, I guess if you're just treating it like individual prompts that are all strung together, during those individual prompts you could also at that point provide more information, like your Instagram strategy that you wanted to reference, or target personas that you want it to reference as it's doing certain tasks.
Nicole Leffer: Exactly. And if you know how to prompt well and you know how to build automations, which are two skills you're going to need for doing this, neither of them is as hard as you think it is to learn. I would say probably a day or two of time to build it out. But once it's built, you're saving all that time forever. So you don't want to be building out, I mean, that's a complex workflow we're talking about. I'm saying a day or two to build out. But once you've built it out, it's done.
Nicole Leffer: You may occasionally need to go make tweaks to it and stuff. And the reason I'm saying it takes that long to build out is because you've got to keep testing it and go, okay, now let me refine my prompt. If you already have really phenomenally good working prompts that consistently always give you the same results in ChatGPT, and it's already strung together as a workflow, then it's not going to take you anywhere near as long to build, because those prompts work with those same models.
Nicole Leffer: But that's why I say, let's back this out for people who are new. You start using ChatGPT every day to be able to communicate with it. You start understanding what model is the one I want to use for this or for that. And then you also start to put together these consistent workflows. So the prompts that you can use over and over in ChatGPT.
Nicole Leffer: If you have prompts and flows, I put in this prompt, then I put in this prompt, then I put in this prompt, right? Once you have that and it works consistently, that is very easy to automate. That's your canvas, that's your sandbox. Then you just go dump that into an automation. You're going to need to make a few tweaks to exactly how it works, but overall it's the exact same thing. It's very, very easy to do.
Nicole Leffer: But you've got to be playing with ChatGPT, right? If you're not even playing with the tools, you're definitely not going to get there.
Mandy Hornaday: Yeah. Well, and what's interesting too is, I played with custom GPTs a couple months ago, so I'm sure they've come far from even then, but I felt like I had uploaded a lot of the rules and boundaries I wanted and yet the output was never consistent. As an example, I would say, okay, make sure that this blog post has between 1600 and 2000 words so that it's optimized for SEO. That was a rule that I had uploaded. Then anytime I'd get the output, I'd be like, what's the word count? And it would be like, oh, sorry, Mandy, this is 2400 words. We will update this to match the 1600 to 2000. It was frustrating. I abandoned, probably wrongly so, I abandoned custom GPTs at that time and was like, heck, ChatGPT as a conversation is just easier for me. Because at least then I can picture it.
Nicole Leffer: So a couple of things on this one. I actually use some custom GPTs for certain things, but they're really simple, rinse and repeat kind of process, very specific, exact outputs that are very replicable. Using custom GPTs for anything remotely complex, I think that that is a huge mistake that people make, personally. It is effective in stuff. If you have a team and nobody's going to learn how to use this stuff at all and you just have to give them something to use that's pre-prepared, that's better. So that's my first thing, is I don't think that custom GPTs are always the best solution.
Nicole Leffer: There are certain things they're a good solution for, but I think a lot of times people say build a GPT for that, when that should not be a GPT, just realistically. They're not always the best at direction following. Also your prompt really matters for how it's going to get, how you structure it. So that's a big thing, and they can't count. So that's probably part of why you weren't getting, it can't tell you how many words. It's just going to estimate, guess, make a random call. It's not going to say it's five words if it's 5,000, right? It's going to have a general idea, but you can't ask. These tools can't do it.
Nicole Leffer: And maybe if they are opening Python code and running scripts to actually go through and count the characters and make adjustments and stuff, but you're probably not prompting it to do that. You would need to be explaining that all in your prompts. It just gets real complicated. You might have a better time with projects. I still don't know that you're going to get word count, but something like that. Sometimes projects are a better solution for more complex rule following, having the info in there, but still being able to have that interaction with it.
Nicole Leffer: And I'm not anti-GPTs. I use GPTs for a lot of certain things, but I think sometimes they're looked at as, I could do everything with the custom GPT, and it's not really what it's for. The other thing I will say is, more complex workflows, automations will always be far more effective than GPTs. If you want a simple input that gives you a great output, because you have very specific control at every single step of the process to get it to go where you want it to go. And so the consistency is going to be a lot higher on automation.
Mandy Hornaday: Okay. Totally. Yeah. I wish I would have known that, because I think the way I understood custom GPTs was that it could really, I think I probably thought of it more as an agent, that it was going to be this role, the strategic role, take a lot of different roles, and they're doing a lot.
Nicole Leffer: They're definitely getting better. I'm not saying they're not, they're definitely getting better. It's just, there's a lot of work to get to the point where they're phenomenally consistent at more complex stuff. I have seen, I have friends who have built them, they are not something that any lay person is going to build. Some of the stuff I'm like, oh my gosh, how did you figure that out? How did you think of that? And they're putting in JSON files and code. The way they're doing the direction is not something I'm going to explain to somebody.
Nicole Leffer: So it's not that it's impossible to get a GPT to do some of those more complex rule following things. It's just that you're going to, that's where you're starting to get to, you need an expert that really understands this tech inside out to build those kinds of GPTs. You better be really sure there's a return on the investment of building a GPT versus, oh, maybe I'll just do it with a normal ChatGPT conversation. There are definitely plenty of things I do in my workflow that I go to GPTs for, so I don't want this to come as a knock on GPTs.
Nicole Leffer: I would be more inclined to be like, if you're going to do the SEO kind of blog posts, you have GPTs for different steps of that process. So it's not a one go, give me an SEO optimized blog post. It's just going to give you a better result.
Mandy Hornaday: Makes sense. Yeah. Okay, so I want to pivot a little bit and talk about tools for beginner level. So I hear you loud and clear in terms of determine what your processes are, start by dictating it or documenting it. I love that. I think it's a great place to start. There's so many tools out there. Would you say ChatGPT is probably going to be able to do 90% of what we need to be able to do at this point, or how would you be encouraging? What's a tool stack, if you will, that you think every B2B marketing leader should be thinking about right now?
Nicole Leffer: For the vast majority of marketers, you need one tool for your team. Again, this is not people who are already advanced in AI, because the rules start to change. But for most marketing teams as a starting point, you need what I call a foundational core tool. And you need your team to learn how to use and maximize and understand that foundational core tool inside out before you start adding lots and lots of different tools on top of it.
Nicole Leffer: The tools I would consider a foundational core tool that I would put into that category: ChatGPT, obviously a paid account, not free. Be clear, anytime I say ChatGPT, I'm talking about the money kind, not the free version, especially in a business context. So ChatGPT and Google Gemini, without question, can count as that core tool. Claude can count as that core tool. They don't have quite the same level of functionality built into Claude, but if that's your choice, it's a very good tool, and it can do a lot. And then Microsoft Copilot, if it's the only thing that you're allowed to use, can also count as a core tool.
Nicole Leffer: So those are really what exists in my mind that can be that core foundational tool that your team needs to understand. They're all going to be similar in how you communicate with them effectively. That's going to be the same no matter which of them you're using. So if you're learning how to use that type of a tool, it's very translatable, right? So if you start with Microsoft Copilot and you're learning how to communicate and do everything you can in Copilot as your core tool, you can start translating that and take all that knowledge and then go into ChatGPT, or Copilot is going to evolve. And the only reason I say if that's the only one, is just because they haven't brought in all the same functionality yet that some of these tools have.
Nicole Leffer: It also has advantages. You can use it in the sidebar on a Word document or a PowerPoint. So it's not like there's a negative to Copilot. And I mean, I'm an OpenAI girl. There's a place in my heart for it. So to me, my number one choice would be OpenAI and ChatGPT. But I would rank them probably in that order: ChatGPT, then Gemini, Claude, Copilot. But any of them is going to be equivalently a great core tool.
Nicole Leffer: Once your team understands inside out what is actually possible with that tool, which means that you can't make the decision that you can't do anything with that tool. You don't make assumptions about it. You always try to figure out, is it possible? Where is it? All of those things, then, and only then, where everybody's really good with it, should you be considering bringing in anything on top. With the understanding that there's a very good chance that at some point in the very near future, ChatGPT or Gemini or whatever is also going to be able to do that. And you're going to maybe want to get rid of that extra spend.
Nicole Leffer: Because would you continuously build your tech stack. The number one mistake I see marketing teams making is they're trying to bring in different purpose-built tools for all different kinds of things they want to do. Then nobody knows what to use for what. They don't have any idea how to use any of the tools really other than at a very surface level. And they're getting mediocre results out of every single thing they're using, versus just actually learning how to get good results out of one thing.
Mandy Hornaday: Interesting.
Nicole Leffer: Or they're not even using it because it's too confusing to even know what tool they're using.
Mandy Hornaday: Yeah.
Nicole Leffer: Right. So that's where I would start, is build real foundational skills for the core. And then once you know how to use ChatGPT inside out, that's super translatable. So you want to add Gemini to your stack too? You're just going to know how to use it. You might have to figure out where they keep the button. You know how to use Copilot inside out? You're going to be able to translate that to the other tools. But then also, most of the other stuff that you start adding, that AI understanding you've developed is going to start making it much easier to adopt the other things too.
Nicole Leffer: And then maybe one marketer on your team needs a tool that can make AI generated songs. So you get them Suno, but it's not like you have to give your entire team that tool. Or one person needs to make video. So you get them Descript, but you don't have to get that for your entire team. There may be certain tools that from the get go, you know, video stuff, you may need Descript from day one. But most stuff, just learn your AI tool inside out.
Mandy Hornaday: I was just going to ask you, what are some of the most fun tools that you've seen for marketing teams specifically that you think would be really cool to explore, once you've got that core work down really good?
Nicole Leffer: The biggest thing I would say is, what do you need the most in your workflow versus what's the most fun? But my fun tools? I do like playing with the video generation tools like Sora and Runway. They're really fun. You can put in a text prompt or a picture, it can turn it into video. I mean, Runway can do a bunch of stuff beyond that. It's fun. It's just for a fun and understanding and experimenting thing.
Nicole Leffer: ChatGPT's image generator has gotten so good, I don't necessarily know that you need external image generators, but Midjourney, I think, is kind of considered the gold standard for image generation still. So if you want more image generation capabilities besides ChatGPT, you could explore that. I find Suno to be a really fun tool. You could write a song in ChatGPT, or you can even have Suno write the song, and then it can turn it into music that sounds like recording production music, and the style, and oh, it's just, that's a fun one.
Nicole Leffer: And there are definitely uses to that, right? If you're a marketer and you need a background song or whatever, there's some definite uses to that. But honestly, I feel like ChatGPT and Gemini just have such fun features baked in. The image generation stuff in ChatGPT is so fun to use. And Sora comes with ChatGPT accounts, so you can play with Sora with your ChatGPT account anyway.
Nicole Leffer: Operator, which is OpenAI's agent, comes with the pro ChatGPT account. It goes out and operates on the internet for you. So it can actually take actions. It's fun to play with. I would be very careful with business processes with Operator, but there's so much other. And then I love Zapier for being able to build the automations. Practically, Zapier is a huge one. And API accounts for the same core models.
Mandy Hornaday: Amazing. Oh, I'm so excited. I am going to try and set up at least one automated workflow this month. I think that'll be...
Nicole Leffer: Do it. It's so cool.
Mandy Hornaday: Well, Nicole...
Nicole Leffer: Keep it simple. Don't try to do what we said as the first one.
Mandy Hornaday: That's the one I want to tackle. I think I do, eh?
Nicole Leffer: Be ambitious. Who am I? Don't listen to anybody who says don't do it. You might want to just start with, take the transcript and put it into bullet points for me. Just to see how it works. And then you can take that as your base and start the more complex ones.
Mandy Hornaday: Fair enough. Fair enough. Okay. I love it. Well, Nicole, I could talk to you all day. This was amazing. I've learned so much. Thank you for joining us today.
Nicole Leffer: Thanks for having me.
Mandy Hornaday: Yeah. Just to wrap us up with one final question, what is your biggest piece of advice you'd give to B2B marketing leaders right now in this AI world, just to put a bow on everything?
Nicole Leffer: That's a really good one. I would say lead by example. If you are a leader, if you want your team doing this stuff, you better be doing this stuff.
Nicole Leffer: You can tell them all day something is important to learn for their careers, but if you are not doing it, and telling them you're doing it, and telling them how you're doing it, they're not going to necessarily actually start adopting it. You have to prove to them you actually see the value. And that it's not just something else you're throwing on their plate and saying it's important. And it's going to make just as big of a difference in your role and your job and your career and your future as it is in theirs. So you need to be doing it too. And if you want them to do it, you've got to do it yourself.
Mandy Hornaday: Yeah. Sound advice. Amazing. Nicole, thank you so much for joining us today. I know you're super active on LinkedIn, so I would encourage everyone to check you out there. But if people want to learn from you, what is the best way to do that besides following you on LinkedIn?
Nicole Leffer: Yeah, I do post a ton of free content on LinkedIn. And then, just hands down, the best thing is going to be to take my Foundations of Generative AI for B2B Marketing course. If you go to my LinkedIn, you can get to my website, or just go to NicoleLeffer.com and there's a courses tab where you can get to it. But if you do that, you're going to understand ChatGPT and how to talk to a tool like ChatGPT very well, not just talk to, but use the features and everything that it can do.
Mandy Hornaday: Amazing. Amazing. Awesome, Nicole. Well, thank you so much again. It's a pleasure having you.
Nicole Leffer: Thank you so much for having me. I hope you have a great rest of your day.
Mandy Hornaday: You too. Bye bye.
Mandy Hornaday: Thanks so much for tuning into this episode of Growth Activated. I hope this conversation sparked new ideas, challenged your thinking, and gave you practical tools to help elevate your impact as a marketing leader. If it did, I would love for you to pass it along to a friend or a colleague in B2B marketing. The more we grow together, the more we raise the bar for what marketing leadership can look like. And as always, in the meantime, keep activating growth for yourself and your company. See you next time.
