Paul Roetzer, founder and CEO of the Marketing AI Institute, and author of “Marketing Artificial Intelligence,” has been educating marketers on the importance of leaving rule-based marketing tactics behind and shifting to “next-gen” approaches based on machine learning, natural language processing, and other forms of artificial intelligence. In this episode of Customer Data Perspectives, we dive into how AI is changing the way marketers think about strategy and execution, some interesting findings on consumer sentiment around AI, why responsible AI matters, and some surprising takes on what innovation may hold for AI in the future.
Watch or listen to the full episode below. You can also tune in on Apple Podcasts, Spotify, YouTube, or wherever you choose to listen to your favorite shows.
Related Reading
- Report: Consumer Sentiment on AI in Marketing and Customer Experience
- 6 Ways AI is Transforming Marketing and Customer Experience
- Top 4 Enterprise AI Trends for 2022 and Beyond
Read the Transcript
Sacolick: Hello everyone, and welcome to this episode of the Customer Data Perspectives. Have I got a treat for you! We’re talking AI today with Paul Roetzer, he’s the founder and CEO of the Marketing Artificial Intelligence Institute. We’re gonna get into that a lot today. How’s it going, Paul?
Roetzer: It’s great. Good to be with you. We were just sharing stories of everybody starting back to school. And we’re all trying to remember what it’s like to be business professionals and not have kids in the house. And so I’m, I’m adjusting so, it’s kind of nice. I feel like I’ve been getting my mind back, my focus back.
Sacolick: I hear ya, I have kids starting school. And it’s also, you know, it’s just that time of year where we’re trying to do a lot of things and learn a lot of things. And, you know, that’s actually my first question for you. You have a background in journalism and content marketing, but now you’re covering AI specifically for marketers. How did you get into AI?
On the Marketing AI Institute
Roetzer: Yeah. Curiosity, I think, is probably the shortest answer to that. So in my past life, I owned a marketing agency. And I started it in 2005. We became Hubspot’s first partner in 2007, we became kind of the origin of their partner program, which today is like 45 percent of the revenue – I think comes into the partner program. So I built an agency. And along the way, in 2011, IBM Watson won on Jeopardy. And I was writing the manuscript for my first book at the time, “The Marketing Agency Blueprint,” and had nothing to do with AI – didn’t even know what AI was. So after I finished that manuscript, I became curious about what is Watson? Like, how did it do that? How did it understand human questions, quote, unquote, understand human questions? And then how did it find the answers and predict? Like I knew it was predicting the answers, it didn’t know, actually know {what} the answer was – probability that it was right. That’s kind of interesting.
Like, it’s what we do in marketing. A lot of people would come to our agency and say, hey, we need to generate 500 new leads a quarter, or we need to increase our audience 20 percent. And so as marketers, we’re trying to predict what action to take to generate an outcome. And so I had come to believe by 2011, 2012, that we were incapable of it, that humans couldn’t process all the choices, we had – the 1000s and 1000s of ways you could spend marketing dollars. And so my interest in AI was actually very focused on, can it help me build strategies? Can it help me allocate my budget? And that sent me down this whole rabbit hole of what is it, and how does it work? And what are the actual use cases? And so that’s really how it started. And then we launched our Marketing AI Institute in 2016, to share what we were learning and tell the story.
So the, the other part of this is, we were journalism people. We were backgrounds in content marketing. But we wanted to tell the story of AI, not build it, we weren’t capable of doing that. So we started telling the story of it. And we eventually spun that institute off into its own company and launched an event and online education and wrote a book. And here we are, 11 years later, still trying to tell the story of where we are and where we’re going.
On How AI is Shaping the Next-Gen Marketer
Sacolick: Yeah, it’s fantastic, Paul, I mean, it is a story. I mean, I started working, and learning about artificial intelligence back when I was in grad school, and that’s over 20 years ago. And I learned how to program neural networks by hand with C code and, you know, writing for loops to simulate an entire layer of neurons. And here we are today. And we’ve got, you know, more library, more technology than we can ever handle, but not necessarily all the skill sets and all the know-how, in terms of applying it. And I always tell people – marketing, you know, lots of data, lots of experimentation. When I speak to technologists, I’m always like, why aren’t you partnering with your marketers? They have all the data problems they have they bring revenue in. Why wouldn’t you want to spend time with them? So, you know, you know, it all came together. When I started reading your book “Marketing Artificial Intelligence,” it’s sort of a good play on words there. It’s just very useful to just see how marketers are actually using it today. But yeah, one of these quotes from near the opening of your book, and you talk about the difference between last-gen marketers and versus a next-gen marketer, I thought maybe you’d contrast that for me, for everybody here. What is the difference between the two?
Roetzer: Yeah, so I mean, it starts with the technology itself. So marketers, we have this, I know if you’ve seen the Chief MarTech Scott Brinker, that landscape it’s like 9,700 marketing technologies, like the little logos on it. And what I always said is the great irony of marketing automation is that it’s manual. Like, I wrote that in my second book in 2014. So the joke of marketing automation as a movement within the industry over the last decade is it’s all human written rules. Like there’s nothing really automated about other than the rules we write, and it doesn’t get smarter, it doesn’t get better.
And so one of my fundamental first pivot points when I start learning about AI is like, oh, this is real automation. This is intelligent automation. This is, the machine helps you automate. And then it gets smarter as more data comes in and learns. And so what I started thinking about is like, the traditional marketers used to using traditional technology, where they have to write all the rules, it’s their brains that figure out everything. And so you mash up your experiences and your best guesses about what’s going to happen and your instinct and you set strategies and allocate budgets. And that’s it, that’s what we do. Where the next-gen sounds {like} – wait a second, there’s smarter technology out there. There are ways to do email, smarter, marketing, spend smarter advertising, creative. Like, the tech can actually help us be better at this. And so the next gen became a way for us to just differentiate someone who embraces how quickly this stuff is changing. And the fact that there is smarter technology out there. And they seek one, the knowledge to understand that technology, and two, they actually go apply it. And so they run pilot projects, and they’re constantly testing and evolving. So it’s more of a mindset of just embracing what AI enables and not hiding from it because it’s abstract, or scary or overwhelming.
On AI in Content Marketing
Sacolick: Yeah, we actually had Scott on the show on Customer Data Perspectives – fascinating guy, and I’ve been following his landscape quadrant for, I don’t know, at least five or six years, because he’d see that, that curve, right. And, you know, now, you know, you pretty much need some kind of machine learning capability built inside these things, and what are marketers doing, right? We’re not hand-picking 100 keywords anymore. We’re not just doing, you know, highly structured marketing automation funnels in our emails anymore, we’re letting the machine help us connect audiences to outcomes. And that’s what we’re doing. And, you know, I think if you’re, if you’re trying to be a next-gen marketer, and you’re trying to figure this out, I really thought “Marketing Artificial Intelligence” is just a great book to get started. And in your book, you really cover a lot of really good examples. You talk about SEO and e-commerce and advertising and sales. But you say one thing that really caught my attention. Why content marketers must embrace AI before it’s too late, like it’s already past mainstream? Can you talk about that?
Roetzer: Yeah. I just feel like when we think about the maturity curve, or like, the adoption curve, advertising was probably first. Like advertising analytics, you could make an argument for, but programmatic advertising and the need, because so much of the money – I mean, that’s where the majority of marketing spend goes, is advertising, digital advertising. So it’s very logical, follow the money is the general rule of investing and in technology. So it is logical that a lot of the VC money got poured into advertising tools early on, building smarter advertising tools. But content marketing, when we think about the main applications of AI in marketing, we think in language, vision and prediction. So language is the understanding and generation of words, whether they’re written, spoken, whatever. And so when you start thinking about content marketing, you have to devise strategy of what am I going to write about and when, then you have to get into the production of that content, and you have early entries in this like Grammarly, which, you know, almost everybody uses Grammarly.
Realistically, Google Docs and Word docs are using some basic forms of natural language processing, and probably some machine learning in there. So like, it’s exposed to everybody, we all create content. And a lot of brands these days are built, in essence, as media companies, where we’re pumping out podcasts and webinars and blog posts and newsletters and like content is at the core of all of our growth. And so it makes a lot of sense that this is an area where like, if you’re not experimenting with language generation tools like copy.ai, Jasper, HyperWrite, these GPT3 based tools, then you’re probably falling behind, because a lot of people that’s become an entry point to them. And I think what’s happening in terms of understanding adoption is we’re now at the point where the average marketer, for either a free trial or $19 a month, can go get a tool and see how AI works.
When we started writing about this in 2016, that was not the case. We would explain it, but I couldn’t show it to you. Like there weren’t any tools. I’d be here like, look, I’ll write a blog post with it right now in front of you. You couldn’t do that eight years ago. And so I think in content marketing, one strategy, I don’t know how you do it without it. And it kind of dovetails into SEO, too. It’s like, how are you building a site now where you’re still guessing at what to write about, or just using keyword volume? Like, that’s like 2015 stuff. So I feel like content strategists and producers and then people who promote that content, you almost have to be using it or you’re, you’re just gonna fall behind.
Sacolick: Yeah, I mean, I’ve been doing content writing my whole career since 2005. And I remember the first time we showed a related article algorithm to our editors at Business Week, and it was outperforming them, right? So they were hand crafting and saying these are the best five articles to link, spending 10 to 15 minutes of their day, every time they wrote an article to figure that out. And then we went back to the editor and showed him the algorithm, it does better. Why? Because the algorithm is using more than intuition, it’s using actual click throughs. It’s actually understanding a little bit about the reader and doing this, it’s adjusting not only what articles but what the order of them are, and when to show them. These are just more things than we can handle. So I just think that you’re right that with 9,700 technologies, and the majority of them having some kind of machine learning, it’s just a matter of rolling up the sleeves and saying, what can I do with this thing? What kind of outcome does it get me?
Roetzer: And another example, you just gave is one of the early ones I used to use, like I wrote an article for Content Marketing Institute, I think is back in like 2017 – it was that exact premise. Like, okay, user A downloaded this eBook, now show them what? Like, so a human would think we could code this and it’s fine, like do that once. What if there’s 10,000 downloads in a month? How in the world do we personalize the journey, and like, know which article to show them next? You can’t. And yet most brands are still hand coding that.
MAICON 2022 and the Future of AI Beyond Marketing
Sacolick: Yep. Now let’s take this a little bit further down the road. You just had MAICON, which is recently just recently,
Roetzer: Yeah, early August.
Sacolick: Early, and you told me a story about EY and Google having a little bit of a discussion about what the future of AI looks like, and their perspectives around it. Can you summarize it for the group because I thought it was really interesting.
Roetzer: Yeah. So part of my thing, so Marketing AI Conference is, MAICON, started in 2019. We just had our second in person after two years, two year hiatus. So when I build the agenda, my general feeling is we have point of view on a lot of important topics related to AI. But I don’t want our audience, our community, to just follow our point of view. What I want to do is to expose them to as many differing opinions as possible, because so much of this is just emerging, like we’re just starting to ask the hard questions as an industry. So we had Domhnaill Hernon from EY, who’s building their global human enterprise, their cognitive, human enterprise. And they take, well, he takes, a very specific point of view on humanity, its role in creativity, and belief that like AI isn’t going to take over this stuff. And so they’re infusing human in the loop into everything they do. And they’re trying to enhance human creativity with AI. They’re not trying to replace it.
So that was one of the keynotes. And it was, it was fantastic. And then we had another from, from a researcher at Google. And his point of view was, he’s working on AGI. So like, they’re coming at it from, they believe. And it’s not just Google, and not not just Vedant – it’s Open AI and Deep Mind and others. Their premise and why they’re working in this space is to build artificial general intelligence, they believe that we can get there. And to them, teaching language and vision to a machine is just steps along the path to AGI. They’re not that interested in GPT3 writing tools, per se, they’re interested in what those writing tools can do to build AGI. And so you have like, if you think about a spectrum of where’s the AI taking us. EY’s stance, or at least Donelle stance is, it’s taking us to enhance human creativity. Vedant’s stance and not shared necessarily by everyone at Google, but his stance is it’s a path to AGI. And so what I was explaining to our audience is like, these aren’t right and wrong. This is, they’re building it for this purpose. They’re building it for this purpose, and which you as the business person, the data scientists, the CIO, the mark-, whatever. I think it’s critical that, one, we have to understand what this stuff is and how it works, so you can ask the next more important question {which} is, why are you building it?
So I’m buying my tool from you, why did you build this tool? Because if you’re buying a GPT3 power tool from OpenAI, it isn’t to generate $10 million a year in marketing revenue. It’s to get to AGI. And like, it’s just, I don’t think outside of the technical people, I really don’t think people even comprehend that that’s something they should be thinking about. And it’s DALL-E-2 like blown up, like image generations blown up. A lot of that image generation is designed to get to AGI – not to like create stock photos, alternatives to stock photography, like it’s not why it exists. So I don’t know, like that’s the stuff that fascinates me when we need to talk with really smart people like that’s the path they want to go down. It’s like oh, okay, I get there’s use cases for marketing. Let’s go talk about the bigger thing though. And it’s an uncharted territory because not many people are talking about it.
On Customer Data and AI in Marketing
Sacolick: Well, we have to talk about the big things because they tend to start on the cusp of disruption, years before they become mainstream. It takes us years to get our organizations ready to think about how to leverage these things, let alone getting the technologies and the skills in place. I mean, we wouldn’t have been talking about AI and VR four or five years ago. But here we are now. It’s feels like it’s on the cusp of mainstream. And now we’re talking Web3 and Metaverse, right?
So this stuff moves pretty quickly, and to know where Google’s going with things and where EY is going with things. What the art of the possible might look like, whether it’s three, five or 10 years, you know, there’s differing opinions around that. I think it’s really important. And we know we can experiment right now, right? We have 9,700 technologies to go experiment with. But I want to bring this backwards, one more step, right. And before I even get to experiment, I have to have goals in mind, I have to have some idea of who our customer or target customer segments are, I can experiment with both of those as well. But I do know, as a technologist and a data scientist, it really is garbage in garbage out when it comes to the data, right? We can throw all the data at, you know, any of the algorithms we want. And it’s going to just tell us a random answer if we don’t feed it, some quality data. So but you know, marketers aren’t data scientists, right? They’re marketers. So what should a marketer do to ensure their data, and really, specifically, their customer data, is really ready for some kind of AI experimentation?
Roetzer: Yeah, I think. So one, I think this, the challenge of data can be an obstacle that prevents people from doing AI or seeking AI because they think they have to solve this big data problem for any AI. And that’s not true. Like if you wanted to go start using AI to write your social media posts, the data is the unstructured data from the blog post. Like you just feed it the blog post. So like, every use case does not require working with the data scientists to build a data lake to get like, no. When you’re looking at the bigger things, the things that are truly going to accelerate growth, specifically personalization, like out of the CRM, yes. Now, what you need to know is that you need to talk to a data scientist. Like, if that’s not you, and most marketers, it is not, then you need to go talk to them.
And to your point earlier, the data scientists need to be going and talking to the marketing. And that’s the key, they don’t know when to talk to each other. But personalization is kind of like the Holy Grail, it’s what a lot of AI promises to do within marketing is this deep level of personalization. I’m not talking like Company Name, First Name, Last Name, personalization. We’re talking like, behavioral and intent, like, truly understanding the difference between your intent and my intent based on our click history, or our opens of the newsletters or whatever it may be, and then taking in all these other factors that you can infuse into the CRM. And so that’s where to really scale up AI, and to achieve what’s possible with it, then you absolutely need the data scientists and the marketers working together on building those strategies. Because the marketers won’t be able to envision all the data they could be using. And they certainly aren’t the people to structure it, and figure out how to then apply it. So yeah, it’s got we got to get it working hand in hand, and they often don’t.
On How Data Scientists and Marketers Can Work Together
Sacolick: Well, and I like where you started from right? Let the marketers start the experimentation with what you have, see where it leads to. And then when you start getting some idea of, you know, what are the performance issues? What data might you may be missing? What are some of the data quality issues? How do we make something really run at scale? Then you start talking to your data scientists and to your technologist and say like, look, we’re onto something here. Let’s figure out how to make this work. Do I have that? Do you think the same thing?
Roetzer: Yeah, I think we’re always there, you want to like, if you’re just getting started, you want some quick win pilot projects. And you don’t necessarily need the data scientists involved for that. You’re trying to find little things that can intelligently automate stuff that’s repetitive, that takes you a while to do it. So like, find somebody who built a tool to do those things. And as you’re kind of stacking up those successes with these smaller projects, you should be talking with the data science team about the bigger initiatives that you can be starting to do this at scale, achieving personalization, improving, enhancing decision making, because you’re building predictive models for churn, and you know, customer conversions and all these things, lifetime value. That’s where the data scientist the marketers need to come into play is like, hey, here’s my big picture goals. Here’s what we’re trying to get to. How should we be working together to achieve that? Meanwhile, I’m over here writing social media posts and blog posts and better email subject lines and all this stuff that I don’t I don’t need the data science team to do.
On Uniting Customer Data and AI Across the Organization
Sacolick: Maybe let’s expand the scope also, right? We’re talking about data science, technologists and marketers, what about sales? What about customer support? You know, how are we thinking about supplying data across the entire organization, so they’re smarter about customers they’re smarter about their own experiments. How do you think about that?
Roetzer: Yeah, so we do think about marketing, sales and service, or experience is sort of under an umbrella. It’s hard because it’s not necessarily under the CMO’s umbrella. Like there isn’t always a single person that’s in charge of that. So certainly, in smaller companies, maybe in the middle market, it might be the CMO that oversees all of those things, or works with the, you know, head of sales. And experience is probably falling under the CMO in that case. But as you get into the bigger enterprises, you have individual leaders in each of these areas, who are in charge of their individual thing. And this is where I think we may see roles develop. Like, I don’t know that it’s the CIO, or underneath the domain of the CIO, that is looking across all areas of marketing, sales, service, product, you could throw in there, and thinking about how do we make it all smarter?
I think AIOps is a logical thing. I don’t again, I don’t know if RevOps, like I don’t – these aren’t people that are trained to do this. Like that’s not their thing. And so I feel like we need to develop talent, that is business strategy, understands the business fundamentals, understands what the AI team, the data team are capable of doing. And just goes, okay, how do we make everything smarter? How do we make it more efficient? How do we unlock creativity? And they’re looking at the different domains within the company to do it, starting with marketing, sales, service. And I mean, honestly, like, maybe even starting with sales before marketing and service, I don’t know if it just depends on, probably your industry and your business model. But yeah, you need to look at all of those together. In our book, we just kind of lumped them all under this umbrella because we create the content for those areas as part of what we’re doing.
On AI, Data Cleansing and Identity
Sacolick: Yeah, I kind of like how you’re putting this all together. And it just makes me think, and I’m wondering how you answer this, you know, maybe you sell the CMO or the CRO on this idea of experimenting with AI, and they start doing their first experimentations, and they find, well, we do need to do a lot of data cleansing work before we get started. Why isn’t there an AI that just fixes all the data from us? Right? IT hasn’t done it…so why isn’t there an AI that just fixes the data? How do you respond to something like that?
Roetzer: I would buy it. Like if we had, we’re going through this now because we just did a migration of our site from, so when I sold my agency we had to decouple the Institute and the agency and that was all the same CRM database, basically. So we had to get it all into a new platform and move everything off.
Well, then you start realizing the issues you have with duplicate, duplicate contacts and things like that. And is this person the same as this person? Like a simple thing – it’s just pattern recognition. Basically, you’re just trying to like, do I think this person is the same as this person? Right, now you have to manually do it. At least within HubSpot, you have to like – it will surface people it thinks is {but} you have to go through and do it every time when you could be talking about 1000s of records like this, and especially if you’re a bigger organization. So yeah, I’m trying to wrack my brain, as I’m talking and thinking, you know, because we’ve profiled almost 200 AI vendors. I’m trying to think if we’ve come across one that this is their thing, like it’s a primary use case.
Sacolick: It’s something that, you know, it’s one of these things, because I often talk to, on the IT and the data side of the equation. And yeah, I mean, there isn’t a magic box that you’re going to throw in artificial intelligence. It’s going to take some automation, right, let’s first make sure we understand where all the data is, right? It’s not all in the CRM, we have some workflow happening in the CRM, we have some sales happening, some marketing happening, but I also have about 60 or 70 other tools that I’m using for marketing, right? You know, I’m doing advertising in one tool, I’m doing my website in another tool, I’m doing my email marketing and a third tool, there’s social media happening outside of that, in platforms that I don’t control. All that data is there, I need to bring it into one place. Then I got to deal with duplicates. I got to deal with joining that data. And in some cases, yes, algorithm can say, you know, Isaac Sacolick,, spelled slightly different, is probably the same person. But you know, somebody else with a more popular name, probably needs somebody in the middle, a human in the middle to make some judgment calls at some point and saying, is this the same person or not, you know?
Roetzer: They’ll see it, where it’s like, okay, the AI went through 30,000 contacts that has a 98 percent confidence level, these are correct. And you can like, review 10,20, like, yeah, fine, whatever, do them. Then you have your 70 to 90 some percent and you’re like, human in the loop for these below 70 percent confidence level. That’s the humans going through all of them. Like, that’s not hard. Like that is a basic function of machine learning, basically. And somebody has to have built it, I assume it exists.
On AI, Personalization and the Customer Experience
Sacolick: So there’s two equations, right? What can we do with algorithm? You know, what do we need people in the loop to be making decisions around it? And where does it make sense to invest your time into? Right? There’s kind of diminishing returns after a certain amount of time. But if it’s, you know, the top, you know, 10 percent of your customers, you’re looking at what their trends are. Maybe you want to go look at that really well, right? You know, and then notice, obviously, a long tail around that.
I want to shift also gears, you talked about personalization. And that intrigued me as a conversation, because we’ve been talking about personalization for two decades now, in some form. And some I think, early days, we probably over promised it in the early days. And, you know, the folks at Customer Data Perspectives, they shared some data with me, I’m just going to read this off. When receiving personalized communications – that could be an email, that could be even an ad, it could be a white paper, but now it’s personalized, how relevant and accurate are the product or service recommendations? And about a third said extremely accurate, which, you know, we all get our inboxes flooded with information. And a third accurate seems pretty high to me. Where do you think it is? And why is it so hard?
Roetzer: Yeah, I mean, I think in some more advanced industries, that it’s possible that that third is real, but I could see that being the high end of accuracy. I mean, we all experience it. Whether it’s retargeting with ads, or emails for a product we just bought, like, I did this with a bonfire product, I won’t name the company, but a company that makes these awesome bonfire things. I’ve gotten more emails with specials and discounts in the last two weeks since I already received the product to my house and have used it, not even like bought it the week prior. So somewhere, there’s a breakdown, I’m getting retargeted, I’m getting email retargeted. That’s a major waste on there. And it’s kind of annoying.
And so I think a lot of these B2C brands that you assume would have the ability to do personalization at a better level, they’re not doing it as well as they should be. And I don’t know where the breakdown is, it could be a data breakdown, it could be third-party like API, like legs, and when that data gets processed and infused in the CRM, like, I don’t know. But I think that a lot of organizations, a lot of brands, especially in the B2B side, like they just still think personalization is name and company. Or maybe like some simple level of I’m going to show them these links in an email newsletter, because they clicked on those links. And the machines helping do that or write some subject lines that might be more apt for they’ll be opened by a cohort of people, like grouping more in.
So I mean, I think it’s hard. It’s hard because we have, what the average enterprise I think, is like 100, and some 110 120, marketing and sales, tech solutions. So getting all those things talking to each other to drive this personalization is challenging. I don’t know that the tech is still what it promises to be, as you said, like 10 years ago, we were talking about this, 20 years ago. And I feel like the tech is just never quite there yet. And so it’s almost like you need to think about, on a spectrum of personalization. And we need to try and set some goals for where we’re going to get to, because going from zero to 100 is just not going to happen in most organizations. But starting down the path of okay, like the newsletter we send every week, let’s do send-time optimization so it goes when Isaac’s most likely to open the email. And then let’s update the links within there based on his past click history, you know, some simple things.
Now, I might not know your intent to buy something, I’m not further down your journey. But maybe I can at least get basic parts of your journey a little more personalized. And so, you know, I don’t know if that’s part of it is not thinking about a spectrum of personalization, we’re just like, go right for the Holy Grail, one-to-one all the time. But even that has downfalls. So, you know, the story of Target targeting a girl whose dad didn’t know she was pregnant, and, you know, it goes haywire, like, and that was a smart algorithm there, it was right. Like, you know, so you have these cases where people are experimenting with that one. But it doesn’t mean you should do it. And like, that’s where you need the empathy probably injected into this decision of personalization. It’s like, okay, at this line of personalization, we’re now approaching a conversation – it’s not a tech conversation anymore, it’s a should we do this conversation, which means we need the ethicist in the room to have this conversation. And I just don’t think many brands are structured to think that way.
On Personalization Strategy and Privacy
Sacolick: Well, I like your thinking around it in terms of a spectrum. And it probably has several flavors. When we say personalized I think people respond to that and say, well, it’s not static. I don’t have a single one size fits all for everybody and therefore, it’s personalized. And obviously, that has several grades of how much data you’re taking in to personalize, and how many different experiments, are you able to run realistically on the outside that start personalizing things? Right? How much intelligence are you putting in front of the customer support rep to know, you know, roughly what my demographic is, roughly that I’m a technologist, and, you know, make some better assumptions about how to respond to my question versus when you ask the same question, right? So now we’re getting into just better ability to serve the customer. And then you start getting into, okay, at what point does it become creepy? And at what point does it become annoying? And then lastly, what point is it potentially biased and even illegal? Right, we’re getting into that spectrum. But we know that regardless of what industry you’re in, B2B, B2C, right? This is the competitive landscape that we’re in right now. How do we use data to better connect to our customers in an ethical way, that’s personalized enough where I might be able to get a better outcome than my competitors, because I’m using more intelligence around it.
Roetzer: And an extreme example would be, let’s say, Alexa was always listening, and you gave it permission to always listen, and I yelled to my wife, hey, we’re out of paper towels. And Alexa hears it, and sends a text message to me and my wife saying, would you like to order more paper towels right now? It’s infinitely doable, like, there’s no technology barrier whatsoever to that happening. It’s a privacy barrier. It’s the creepy factor. And so that’s why I’m saying. The level of personalization that is possible is very advanced, in many cases. It’s just a question of, should we do it? Or Is society ready for it to be done yet? Doesn’t mean it won’t happen. It’s just like, have we broken down the barriers of privacy enough yet, that we’re just gonna allow these devices to always be listening and take actions based on that?
On Bias and Responsible AI
Sacolick: Yes, well, I’m gonna put myself in the shoes of that CMO. And I’m hearing about AGI, and I’m hearing about human assisted innovation. And, you know, my team wants to go experiment, and I want to get more personalized and see some outcomes off of it. And now, we’re talking about creepiness factors, bias, legalities around it, you know, you have a whole section in your book around this bias and responsible AI. What can you share in that, you know, what does the CMO have to know, before they start embarking on this journey? What’s the kind of the one on one around this?
Roetzer: Yeah, the first time we did a comment that was more intelligent, more human. And the whole premise was, you’re going to have these superpowers. Like AI is going to give you powers to do things you’ve never done before, to understand human emotion, to affect behaviors, to drive actions based on emotions. Like, you’re going to be able to do some pretty crazy stuff. But you always have to put the human at the center of it. You have to understand the impact it’s going to have on the end users, on the customers…or whatever it may be. And so the whole idea of responsible AI is, to human-centered AI, is that we aren’t making decisions just because we can do something – doesn’t mean we should do it. And so that that falls into the Intelligent Automation. It’s like, well, I could probably automate this person out of a job, should I? Like, is that the right decision to be making? Or if I’m going to infuse Intelligent Automation that’s going to start affecting people’s jobs or their fear for their job – do I have an upskilling program in place? Like, have I thought through the impact down the line, this is going to have on my customers, my employees, my community, whatever it may be. And so we wanted to do was once you teach people like, through all these use cases, like okay, cool, this is infinitely doable, I can go start on this stuff now. We wanted to before we left them in the book, take them down the path that you have responsibility – you as the marketer have to own the fact that these tools are very powerful. And you need to be asking hard questions of yourself and your company before you just race ahead and do things just because you can.
Sacolick: Yeah, I think there’s a balance, is what you’re getting at is, you know, we want to experiment, we want to do more. But look, I want to do that in application development, I want to do that moving to the cloud, but I’m going to be working with my Chief Security Officer, because they’re the expert. They’re going to tell me, okay, if you want to go do this, here’s the right way to do this. I’m gonna talk to my compliance officer and say, if you know, if you want to go do this kind of analytics and financial services, here’s what you’re bounded by, so that you don’t get into a bias issue, right? So I look at AI, I think probably the big lesson here is that it’s a collaboration, it’s going to require multiple skills. It’s going to require the marketer, it’s going to require the data scientists, we’re not just doing this marketing, we’re doing this for customer support and sales.
We know we need to bring our data in, make it as accurate as we can. We’re gonna use automation to do this. We’re gonna use people in the middle to do this, and we’re going to go down this journey in some kind of velocity, right? And we know if we’re going too slow, we’re gonna get disrupted. If we go too fast, we might make some mistakes along the journey. But that’s the nature of transformation that we’re in today.
What is Your “Easier Button” for Marketing?
Sacolick: I want to leave you the same last question I ask everybody on the Customer Data Perspectives, answer, right? Data fuels, all of this, you know, it takes intelligence, it takes algorithms, it takes technology skills. But what is your wish list, the easier button for gaining a competitive advantage with customer data?
Roetzer: You know, I just, I kind of go back to where we’re talking about earlier, where the AI is helping us figure out where we have noise, friction within the customer database, the CRM. Like, we’re dealing with this as an institute right now, with 35,000 subscribers. Like, I think just the more AI can play an assistive role in identifying patterns, and making recommendations based on data to help us make strategic decisions about what we do – that’s where I want it to be. I just wanted to sit there and as an assistive engine that’s constantly surfacing for me interesting things. And it could be, you know, we had 25 new contacts yesterday that downloaded an ebook or sent for a webinar. And it knows what my ICP looks like, my ideal customer profile. And it’s surfacing, hey, here’s seven people you might want to pay attention to. And here’s why they’re relevant to you like, that’s what I want. I want something that’s truly there to assist me in making better business decisions.
Sacolick: It gets a great answer, you’re saying. AI should maybe be telling us, look over here, you know, pay attention to this? Is this right? What do you want to do about it? Whether it’s a data quality issue or an anomaly in user activity that you want to potentially take action against, but pay attention over here, I think is a great answer. Thanks for being my guest today on Customer Data Perspective. We’re here with Paul Roetzer, he is the founder and CEO of Marketing Artificial Intelligence Institute, and he’s got his new book out “Marketing Artificial Intelligence. Thanks for being here with us.
Roetzer: Thank you, Isaac. It was great. Thanks.
Sacolick: Have a great day, everybody.
