AI and the Future of Marketing: How to Embrace Change Without Losing Control

0:00With AI, if you don't ask the question right, you don't get the answer. I use ChatGPT and Claude every day, personally and professionally. And I still find that they're not always so helpful or accurate as a final draft. For sure. And then when I look out at the landscape, I almost wanted to call my blog, like, show me your templates. Because I've gotten so much advice over the years. And you're like, great. I hear what you're saying. Show me. Show me how that works. You should track the most important metrics. Great. What are they? What does this chart look like? You should have a clear budget allocation model.

0:35Okay, great. What does a budget allocation model really look like? And so the smartest experts get the best results out of AI. So trying to figure out, like, you didn't ask this question and you don't know that this is how you should do it. But here's what you really want would be a cool part of the agent.

0:55Welcome back. Marketing trends. You know how we do it. This is your host, Jeremy Bergeron, as always. And I am super energized about who we have in the virtual studio today. Let me tell you about Carrie Lou Dietrich. Carrie Lou Dietrich is a powerhouse marketing leader, former CMO best known for steering Atlassian through its IPO. She's now a trusted advisor to CEOs and CMOs of hyper growth companies. We're talking several, by the way, anywhere from six to 10 companies she's advising actively. And Carrie Lou really is the go-to expert for operationalizing the chaos of scale and driving next level marketing and business performance.

1:48Carrie Lou, welcome to the show. So happy to be here. You're making me blush. I'm not sure. I'm going to have to deliver. Oh my gosh. Well, I had some time to check out some of your past interviews. I mean, you've done a lot of really cool stuff. I was watching the one you did recently, I think on Transaction was what it was. Transaction was a recent one. Phenomenal. I don't know who those two gentlemen are, but that was a cool, some cool context and convo there. I am, look, we get a lot of people on the show that are awesome. And, you know, you're the first person that's come on the show that is advising as many companies as you're advising.

2:24So I actually want to just get into that for a second, because we've had a few that are advising, you know, two to three and maybe doing some consulting, but like you're advising several businesses and they're all moving and grooving. So I just want to understand how, how are you advising this many? And you also have a family. You also have a life. I just want to double click into the, just tell us how you actually advise that many fast growth businesses. Part of my dream, I guess, is to be the person that I needed when I was at Atlassian. So before Atlassian, I ran awareness advertising for Oracle.

3:01I had a really big budget. I did sports marketing, movie marketing. I wrote ads for the front page of the Wall Street Journal, the back page of The Economist. It was a big job and a high profile. But the company wasn't changing so quickly. You know, my team went from two to five to seven internally. And then we had this massive advertising agency, you know, and 50, 100, 150 people that we were orchestrating and were implementing a lot of the things I was working on. But I became very good at building processes and improving processes in a more static environment. But then when I went to Atlassian, we were an early stage startup, an unknown developer tools company, founders out of Australia.

3:44And there were no processes. The CEOs had never worked at any other company. They were first-time founders who have worked their entire lives at Atlassian now. And everything was constantly changing so quickly. And I was a young leader. I didn't really have a strong network of experienced CMOs. I didn't have all the experience I would need to run all departments, having come from running advertising, having done product marketing, having done demand generation, but never being the leader over all of them at this really rapid scale. And we were innovating in our go-to-market. We were one of the first product-led growth companies.

4:24So when I would go to talk to people, their tactics, their budgets, their approaches were very different from what we were doing. And I just felt very alone and confused. And of course you fake it till you make it. And you invent as you go and you use first principles to come up with the best solutions and you hire the best people you can. I used every business book and strategy I could, But I really wanted someone who could help me on a day-to-day basis with best practices, help me organize my thoughts to have more conviction, and help me make decisions when there's no right answer. And so come full circle, now I partner with CEOs and CMOs of these hyper-scaling companies to help the CEO get the most out of the marketing leader and the marketing leader get conviction and knowledge to do a really exceptional job.

5:14So how do I have seven clients? Almost all of my advisory comes from referrals from people I've worked for before or with before or clients. Actually, some of my CEOs, CEO coaches refer me other clients. So it's an exciting spot to be in right now. So Carrie Lou, I just got to ask out the jump here. Do you have any space for any more clients? Cause you got, you know, you're doing six to 10 right now. I'm figuring there. I mean, you may, are you tapped out? What's going on here? Is there space for anything new? I always take on one or two a year. So yes. And especially in the AI space, there's so much changing and happening.

5:53I'm really excited about it. So for the right exciting opportunity. Yes. Okay. Well, you said my favorite word. I had a second follow-up, but because you said AI, now I have to ask you this one question, maybe more, but this one specifically because AI and B2B marketing are like bread and butter. right now. How do you help companies avoid becoming too reliant on AI while still embracing its full potential? I think it's pretty difficult to become too reliant right this second. It feels like we're in the middle of a major shift. Is it the most major shift we've ever been through? I don't know, but I've been through the shift of the internet.

6:32I had dial up when I I was a kid. I didn't have internet in college until my sophomore year or something. I think they installed it. I had one of the very first phones because I worked for a phone company and I could text on my beeper and that was the best technology ever. So I've been through the internet, I've been through mobile. It was the year of mobile for many years while we were adapting software to mobile phones. I've been through the cloud transition and all of those massive transitions have kind of led us here today to this moment where AI can be all it is. So I think it's changing everything.

7:11In B2B, it's changing all of our products, trying to figure out how to really use AI to make people more productive in the context of an application they're already using. I actually think that's the killer app because getting people to imagine how they could use it has a barrier to entry. At the same time, every single marketing department that I work with and company is trying to figure out how you retool all of your employees to be more productive. And there's probably a future where there's fewer of some types of jobs than we have now, right? There's far fewer saddle makers and elevator operators than there were in the history of the world.

7:54But I also think it's being used to make everyone smarter and more efficient. So I don't think you can over rely on it yet. I do think there's a lot of hallucination. You know, we really need to like double check. Like we're just going to start to see that tip of the iceberg on lawsuits of companies being held accountable for mistakes that their AI chatbot on their website makes. I know there was one where an airline promised, airline's AI chatbot promised someone a bereavement fare that later the airline didn't want to honor because it wasn't truly in their policy. And the lawsuit went in the favor of the customer who was told by an AI chatbot what the policy was.

8:40So, you know, I think there's a ton that's going to happen around privacy, around security, around trustworthiness. But we're starting to see massive gains in the companies that provide that, like Claude from Anthropic, I think is one with the lowest number of hallucinations. And then they say part of that is the way they've trained it around values and social good thinking, but everyone's going to have to retool here. Is there an example that you can think of where AI helped you or, you know, someone that you're working with solve a marketing challenge that you couldn't solve before? Yes. I think it's mostly in the scale and personalization right now.

9:22So marketing teams have always wanted to personalize as much as possible at mass scale, but there's just limitations of time and resources to write a thousand landing pages with accurate, customized information. So a number of companies I know of are doing long tail SEO content around super highly personalized topics and topics that include a keyword with a customized ad and a customized landing page. So, I mean, I think there are some things that we've never done before, but I think even more than that is just moving faster on analyzing. I'm seeing people use AI from product marketing to assimilate customer interviews that would have taken you a really long time to listen to every customer interview, to take notes in customer advisory boards and send updates around strategy options that you could pursue and trade-offs around goals, around translation.

10:27I think every aspect of marketing, people are experimenting with it. But still, I think that the real breakthroughs are going to be the products that we use, making it easy to use in the flow of our normal work. Has anyone made a Carrie Lou Dietrich custom GPT yet? That would be pretty cool. Not yet. I mean, Carrie Lou, I could take every interview you've ever given. I could take every epic blog post, by the way, shout out to Carrie Lou's blog. Like she writes a very consistent blog. I see it on LinkedIn all the time. But I could take every piece that you've ever written. I could take every interview you've ever given.

11:03I could take any kind of commentary that's available online. I can literally train GPT to be the Carrie Lou GPT. And now I've got Carrie Lou Dietrich in my pocket and I can ask her about hyper growth, bootstrap growth, all the things. And so if no one's done that yet, you better hurry because I will. But here's the problem, right? What's the compensation model for that? So my current compensation model is one-on-one work with CEOs and CMOs where I really understand and give them custom feedback based on their context. Some group types of coaching. So a colleague and I are going to start a CMO AI working group where every other month some elite CMOs talk together about what they're trying to approach in AI.

11:47So some group coaching, you know, I'm thinking about having some online courses, potentially, I have this blog, which I request people to pay for, if you can just take all of it and have a custom GPT, and it's free for everyone. How do we compensate people to keep creating content? Because time is money, and it's, it's difficult. So I do think there's a lot of issues there, which, you know, the media landscape has pointed out, right? I think the New York Times has been the most litigious to try to protect some of their content. But it's a real issue if AI can take information from creators and just repopulate it themselves all the time.

12:29Yeah, I agree. I think it's super interesting to consider and someone's got to do it. There's a lot of talk around AI agents, right? a lot. And I think there's a really epic thought leader that I've been connecting with more, this guy, Danny Kitchi, and I think is his last name. And his concept is some of his ideas around taking someone like Carrie Lou and then having them spend some time training an AI agent. And then every time that AI agent gets used, Carrie Lou gets compensated. And so at scale, oh, you've got 100 marketing leaders using Carrie Lou's actual trained AI agent by Carrie Lou. Well, now all of a sudden, Carrie Lou's got this compensation model where she's getting one, she's getting compensated to maintain it and make sure it's good.

13:18And then other people get the benefit of using it. So I think there could be some things swaying that way, but you're right. I mean, now it's like wild, wild west. I have a ton of custom GPTs that I've made because it's the acceleration it gives me, the utility it gives me is, I mean, it's wild. I can create a knowledge base based on one person and I can go incredibly deep and that person is not getting compensated in the way that they could. So I love that you brought that up. It's a tricky world of a lot of transition. I think I also, I use ChatGPT and Claude every day, personally and professionally.

13:54It's part of my job to try to get as deep as I can. And I still find that they're not always so helpful or accurate as a final draft. For sure. For sure. They're a great concepting partner, organizing partner, and editing partner. And then when I look out at the landscape, I mean, part of what drives me in my blog, I almost wanted to call my blog, like, show me your templates. Because I've gotten so much advice over the years. And you're like, great. I hear what you're saying. Show me. Show me how that works. And so a number of my blogs, I try to go and collect people's templates and say like, okay, you should track the most important metrics.

14:33Great. What are they? What does this chart look like? You know, you should have a clear budget allocation model. Okay, great. What does the budget allocation model really look like? And so sometimes with AI, if you don't ask the question right, you don't get the answer. Right. And so I do think there's also this like the smartest experts get the best results out of AI still. And so trying to figure out like, you didn't ask this question and you don't know that this is how you should do it, but here's what you really want would be a cool part of the agent or GPT. So you said you said you use Claude and GPT, which is very similar.

15:11I have like, you know, my daily flow is like email, calendar, GPT, Claude, visit the tabs that are open every day. What are you using Claude for? Are you kind of testing each to kind of see what answers you're getting for more accuracy? Or you're like, hey, claws, more creative flow, GPT, more strategy. How are you kind of pulling those tools out right now? So one person I follow in marketing is Marcel Santilli, who is the CMO of Deepgram, and now has his own ScaleX SEO, AI consulting company of sorts. And part of... I went to one of his trainings where he trains you about the long tail of SEO using AI.

15:49And I got into it and I realized, totally a CMO thing, that I actually don't know enough depth of SEO in today's land to do every step that he's doing. I think, oh, I just want to hire him for all my clients. I don't have to learn everything he knows, but I have to know that he's out there and make sure that we're using it to get the competitive advantage. And he, part of his process is, is, is using the two AI LLMs together, you know, getting different answers out of both of them for different steps of the process and asking each other about them. So, so I, I've used them in concert for a number of things.

16:34I mean, primarily I use it around my blog. So I've used mid journey for images and I've used chat to PT for images. I hope that Claude gets images. That would be nice. Um, cause I don't love what I get from chat to PT on images. And then, then I've started using Canva and I've been like playing around with Adobe's creative suite and their AI. Um, or then I use it to, to concept ideas, to think about the SEO, to edit some of the blogs. I find that the most, um, effective uses for titles coming up with a whole bunch of different titles that would have better click-through rates. And then I use it for different things.

17:11I did a blog a while back about where I wanted to look at venture capitalists get rated in this Forbes Midas list, which shows the most successful venture capitalists of the year. But venture capitalists investment timeline is actually much longer, right? Most funds are seven to 10 years. And so I wanted to look at which venture capitalists had won the most Midas mentions, both as individuals and as firms. And so I wanted to do this big data analysis where I took 13 years of data and I plugged it in and I got some tables and they made some charts. And man, I worked with ChatGPT for days on it and couldn't get it to work and ended up calling a friend who's like an Excel spreadsheet hacker.

17:54And then he ended up using ChatGPT to do some of the data input in some of the unique steps, but generally he drove the process. So I don't know. I also used it to create the perfect variety of meals for my guinea pig to get it enough vitamin C and calories. So I would like to say that I go high and low. That's fantastic. I mean, one of my favorite use cases personally is, you know, we have children, we have little children one in our house. And so for, for story time, and you can steal this one if you want for story time, I'll ask all four of our kids, I'll say, okay, give me a character. So each kid is like turtle, a lion, you know, whatever, ninja.

18:41And so I'll say, give me a character. And then I'll say, what are their superpowers? And they'll say, you know, whatever the, it could be anything, literally. And then I'll say, okay, cool. And then I'll go to chat GPT and I'll say, okay, here I'm, you know, I'm reading a bedtime story to my kids, write a story from these characters. And I also want you to weave in emotional intelligence, leadership, you know, conscious, you know, conscious relationship, all the things I want you to also weave in like telling the truth or whatever thing I want to teach that day. Cause lying is a big thing when you have toddlers, like telling the truth.

19:12And then GPT will tell an amazing story. And at the end of that story, I'll say, okay, now create an image of the characters. And so then it creates this image of the characters and all the kids are like, oh my God. So it's cultivating their imagination while I'm reading it. They're just, you know, they're just listening. And then they look at the picture and it's the highlight. Now they're like, Papa, please listen to the story. Let me tell you. It's the most amazing thing. It's the funnest hack for stories. I love it so much, especially because I had this whole series that I told to my kids.

19:42I'd like my made up series about this seagull named Benny that lived in Manhattan Beach where I grew up and was like at the beach where I grew up. And basically the story is like little vignettes from my childhood that I remember about like snorkeling in Catalina and like running on the beach. And Benny really loves Cheetos because if you go in the water and you're at the beach and you leave your backpack open, Benny will open your backpack and eat the Cheetos, which is true story of my favorite food taken by evil seagulls that have now been memorialized as cute children's stories. But I will say that my creativity in getting the hero's arc could use some help between nine and 1030 at night.

20:24Totally fried. So I don't know. You're about seven years too late to reinvent my life. But I love it so much. I love it. It's really been awesome. We are having a tale of a time at our house for sure. There is an AI benchmark test called the MMLU Pro. Have you heard of this one? It's the Massive Multitask Language Understanding. MMLU was the original one. It was basically a bunch of rigorously curated questions that were asked. And so, ChadGPT, Grah, Claude, all of them were taking this test to see where they are on the benchmark. Then the MMLU Pro came out, which is even more enhanced. And so, it's 12 ,000 questions.

21:08It's across like biology, business, chemistry, commsci, economics, engineering, health, history, law, math, all these things. And so I've been paying attention to this benchmark. And just to put this in your garden to plant the seed, at the top of this test, at the top of the model, which by the way, this, I don't know much about this company because if you search for them, there's not much there. But at the very top, above Claude, above Grock, above GPT-40, is this model called ARX, ARX 0.3. And so what that tells me is, and this is measuring across all these different subject matters, removing hallucination even more.

21:47And so we're getting to the point where it's not at 100%. I think the overall score is, I have it pulled up, it's like 0.7824. That's the top overall score. So we're at 78 % essentially in terms of accuracy. And so it's getting there, but eventually you'll have these models that are gonna be 100 % accurate or really close to it. And then it becomes really interesting of like now you can start to rely on these tools. And I don't know who this ARX 0.3 is because there's not a tool you can use yet, but they're clearly beating the multi-billion Claude and Grok and GPT-4.0. And so that's interesting in and of itself.

22:24And so if you hear of ARCs, let me know, please. But I've been seeing like they've not moved off of the leaderboard for two months. They've been beating all of them for at least two months. I'll have to keep an eye on it. It's interesting coming from Atlassian and a developer marketing background. The consumer side is really important. And ChatGPT seems to have kind of an early lead there, of course. But also the embedded API aspect of the LLM is really important. And Anthropic is beating ChatGPT there. And we'll see. I mean, is beating Cloud there from a revenue perspective, I think. But it'll be interesting to see how this all plays out, right?

23:06Netscape isn't really around anymore. Google took the market. That's right. You know, like Lyft and Uber separated substantially. It does feel like it's still early Wild West. and there's the opportunity for a lot of different technologies, you know, to weave their ways. The difficulty, though, seems in this wave, how much money and compute power you need. You know, the affiliation with Microsoft and Google and AWS being a backbone of the leaders makes it harder to imagine disruption by tiny entrants, but we'll see. Yeah, you know, you're nailing it. This is the stuff that I'm like up thinking about of like, you know, because like you said, it's a lot.

23:52It takes a lot of capacity to move to move these things, to power these systems and models. And so, you know, is someone going to come in with a different way to do that? Is it going to keep just being more money, more more capacity is the winner? OK, I love it. So I mean, I think from a marketing perspective, what's. Terrifying, I don't know, I have this I have this funnel of energies that I just printed out yesterday. it's like the funnel of energy fear fear and and grief and depression are on the bottom and enthusiasm and optimism are on the top and i would say ai really makes you like take the whole journey right because like are we all gonna lose our jobs like what are people gonna do you know i'm in a taxi and i've taken a waymo now and i was like oh gosh i don't know if waymos are everywhere am i gonna be employing lift riders like there's some scary stuff like what are my kids gonna do for their job.

24:48But the flip side, you know, and what I was getting at was for marketing, we used to market on radio and in newspapers. And then there was this small job on a team when you were the digital person. Now an entire marketing team is the digital person, right? And newspapers and radio are still useful as part of a portfolio, but they're certainly not what they were to our parents or grandparents. And you look at this last decade and Atlassian's growth was really on the back of SEO. I mean, 80 % of our leads were inbound SEO. We owned SEO for Agile and for developers. And so if you had a question, you went to Google, you ended up on our website, we educated you.

25:34We had this whole philosophy around no forms, no gated content, just like get as educated as you want. And when you're ready, we're here. Start using the product. Use the product for free for seven days, 30 days. We tried all sorts of different day limits. And then the product was so good that you used it. But the whole marketing model was based around internet searches that went to pages. And in a world of AI, where your agent goes on the internet, and maybe you never see ads, and your agent populates content, and you don't even go to a website to build trust with the vendor who's providing you that information, it kind of comes full circle back to the Carrie Lou GPT.

26:16as a company, you know, so Carta, for instance, does this really great social media data content that's really fascinating, that answers all these questions. And you might say like, oh, that's so interesting. Let me go to Carta's site and learn some more. Oh, Carta isn't just a cat table anymore. Now it's also helping investors manage their entire portfolio. Oh, now I've learned something else. Like it's kind of like the fear. And when I was little, and I had a question, my dad would make me go to an encyclopedia. And you'd find your answer in the encyclopedia, but you'd get to read some stuff that was around it and learn that next to, you know, trees were typhoons and all of a sudden we got typhoons and that you lost that on the internet.

26:59But then we replaced it because now you can go really deep. But from a marketing perspective, it's going to be a whole new wild frontier of how we get to consumers and associate with them through AI. And perhaps AI will start having ads, you know, ads in the answers, right? That would be the easiest, obvious choice. But we'll see how this plays out. I mean, my six-year-old says, she doesn't say, hey, dad, go to Google. She says, hey, dad, pull up GPT, ask GPT. I mean, that's what they're asking. And I know for myself, I'm curious for you as well. Like I, I mean, I spent three years working at Google.