Jeannie Walters: 0:01

Hi everybody. It's my favorite episode of the month. It's time for CX Pulse Check, where we check in on what's happening in the real world around customer experience. Of course, I also bring in a special co-host, and today I am delighted to bring in my friend, Andy Crestodina.

Andy Crestodina: 0:26

Jeannie it's great to be here. I'm great, I'm great, I'm glad to join you. This will be fun.

Jeannie Walters: 0:31

It will be fun. And for those of you who somehow don't know who you are, Andy, because you are everywhere and have been everywhere for a while now, why don't you introduce yourself to our audience?

Andy Crestodina: 0:44

I'm the co-founder of a digital agency called Orbit Media Studios. Orbit is 100% focused on websites. We build and optimize websites for both search and conversion lots of B2B lead gen sites and I'm also just a longtime content marketer, SEO, social media marketer, email influencers, teacher. I'm just active and love CX, analytics, conversion. I love the things you love, so I'm glad to join you here today.

Jeannie Walters: 1:16

Yeah, and for those of you who are not following Andy on LinkedIn and following your blog, it is always full of just incredible information. I've learned so much from you over the years, so thank you for that and it's just a joy to have you on. So thank you, and I picked out a few articles here that I thought would be compelling for both of us, because you have all this incredible background in content marketing and digital marketing and just the way the world is going in that way. And, Andy, it might surprise you, but we're going to talk about something called AI.

Jeannie Walters: 1:51

Have you ever heard of that?

Andy Crestodina: 1:51

Two-thirds of my conversations are about AI. I'm actually not tired of it. I love it. I find it to be useful and happy to share my points of view here as we go through these.

Jeannie Walters: 2:02

Well, I am excited to hear that because I feel the same way. I'm excited about the potential. I think we are in a really interesting part of history where we are at the cusp of something and we get to kind of figure out what feels right and what the rules are, which is pretty cool, and I know there are things that aren't going to go well, but we're going to do our best to use it for good. That's my attitude at least. So let's talk about I mean, we need to, of course, make a run for the border. So we are going to talk about Taco Bell here.

Jeannie Walters: 2:39

This is an article from the Wall Street Journal and the headline is Taco Bell and KFC's owner says AI driven marketing is boosting purchases. And I thought this was so interesting because we've talked about AI in the fast food space before, around kiosk choices and drive throughs and things like that. This went into a little bit more of a broader perspective about personalized communication and how that links up with some of the choices in the store as well. So what do you think about this? What did you take away from this?

Andy Crestodina: 3:15

Well, this is an example of a case study that it's not surprising if they're seeing a lift and they have such scale that a small lift can make a big difference to a company like that.

Andy Crestodina: 3:26

I think there are lots of businesses like mine and lots of my clients that wouldn't see a significant enough lift for the investment here at this point in 2024 from that level of personalization. So when you have I mean you're selling chicken, you're selling tacos what is this person like based on their activity at this kiosk or wherever in their experience, like can I give them something that they're slightly more likely to engage with? Yes, and is that worth $10 million to us in technology investment? Yes, because the lift is so high. So I think that it's my. I love it, I believe it, I think it's the future. My caution is that for companies for which the value is gained by, you know, getting a lower volume of higher value leads, let's say, like a B2B company most many of us in digital here not that relevant. This is not something that anybody could just emulate. Yum brand or whatever they're called now, has different opportunities than many of us, but it's definitely a space to watch.

Jeannie Walters: 4:27

Yeah Well, and I think you bring up a really important point about we have to make sure that the investment pays off, and there are lots of situations where it might not right now, today, and what I see is that a lot of the customer data is just not organized the right way, it's not in the right place, it's not centralized, so people kind of plug in AI in just the marketing space and it's not connected to the rest of the customer journey. So it actually leads to really high expectations based on the personalized marketing and then if we can't deliver that through further in the journey, we might be setting up for actual more disappointment for our customers. So it really comes back to understanding what is it? What's the goal here, what are we trying to do? And then I'm seeing a lot of my clients and different organizations that we work with who are really large and complicated. They're trying little pilots to see.

Jeannie Walters: 5:25

Is it worth scaling? And I think that's something that you know. You read an article like this, you hear about that success and you say, oh, it must have worked right, like right away. And I think it's really important to stress that we need to test things. We need to make sure that our information security and our customer data and all that is aligned in the right way. But I think one of the things that I'm sure we've been talking about for a decade plus now is this whole idea of personalized marketing.

Jeannie Walters: 5:57

And one of the things that I'm seeing changing and I'm curious what you're seeing is that used to mean certain things. It meant, like using somebody's name and giving some suggestions, but now in this article they talked about how their Pizza Hut consumers some of them are very frequent, some of them only order on the Super Bowl right, like that's their one thing. And so they are able to, based on those data points, really understand who they're talking to and then provide the right offer, the right coupon. All of that based on actual behavior, which I think is where we're going, instead of just saying to customers what do you want and they tell us and they don't necessarily behave that way. We're using behavior to really understand customers. So are you seeing that as well?

Andy Crestodina: 6:45

I am. I have that same feeling that you did. It's like this isn't an opportunity that everybody would have. You're seeing like an output or an outcome at like phase nine maturity. Like they've got, they know that you only bought once a year because they've been tracking for years. Not everyone's got years of data or their data is like structured properly or able to get this to achieve that.

Andy Crestodina: 7:06

But no, I think it's one thing I think I'm seeing now is that the personalization technologies and tools and the testing, AB testing, technology and tools are sort of the same tools now, because you're, if you want to make so, let's just translate into something more likely, useful to people like me. You want a landing page to perform from a traffic source. You want to know what versions of this page will perform better. The tools now aren't creating an A and a B and you're going to get a lot of data and then giving you an insight. AI personalization and testing tools are creating 15 variants of that and showing it, continuously changing slightly who's seeing it? It might say, oh, mobile visitors are 8% more likely to engage with this word or this position of the button. So personalization is sort of happening in the testing environment, just as kind of the same thing, and the tools are sort of spinning up and creating lots of variants.

Andy Crestodina: 8:09

It's really interesting just how AI is going to blow up the A and the B approach to testing and the three variants versions of personalization. I'm watching those technologies converge and create ongoing iterative, data-driven variations with all kinds of options. In some cases the tool is even writing the variants using AI.

Jeannie Walters: 8:39

Wow, that's interesting to think about, and I think part of what I think is so exciting about this moment in time is that some of what was inaccessible to smaller organizations, to entrepreneurs, is becoming accessible, because we don't need a developer all the time, right, we don't need those very specialized skills. If we can lean into asking generative AI to help us with those skills. I'm not saying it's going to replace that completely, but I think that there is a lot around that with even testing. Like what should we test? Help me understand. What should we test? Is it the language? Is it the design? How would we do that? And I think that's where I personally am also using it in that way. I'm starting to lean into if I don't know something, my kind of go-to right now is saying well, how can I learn this quickly and how can I learn this in an actionable way for me? And so I'll say, like this is the scenario, help me understand. And sometimes what that means is it says you need somebody smarter than you, right, like you need somebody to come in, but at least I know that. So I think it's really interesting to think about those different pieces and those different tools that will be merging together and that we will have available to us in new ways. So, yeah, I'll be watching Yum Brands and then I'll be, you know, on step one and when they're on step ten. But that's okay. That's okay.

Jeannie Walters: 10:19

So this next article is from CX Dive, and this is where the this is again about generative AI, because it is everywhere, but this one is the headline, retailers confront trust issues as generative AI becomes commonplace, and they talked about a few different surveys one from Bain and Company and essentially a lot of consumers didn't always understand that they were interacting with AI. And we've been talking about transparency. We've been talking about providing trust through transparency with customers forever, it feels like, and yet we're still not quite getting there. So do you think that's the problem? Do you think we're not being transparent in the communication and the expectation setting? Or is it just that this is the moment in time we're in, or what do you think?

Andy Crestodina: 11:17

It is such an accessible and inexpensive and fast tool that probably later today you and I will read copy that was generated by AI, that someone used to quickly make it. It might even be an email that we got from someone on our team who knows. So it's a weird time and I think that there will be a big trend toward disclosure. I think that you're going to see it. A lot in images. AI images are currently sort of easy to identify anyway. If there's regulation, I think you'll first see it in Europe.

Andy Crestodina: 11:48

I don't think Americans really are that likely to push businesses to disclose these things. Maybe California. It reminds me of privacy laws. I don't think that we can realistically expect to always see disclosure for when we're interacting with an AI or we're reading something or watching something that was generated by an AI. The labor market impact you'll see probably first in entertainment, where they're unionized and they're going to say you know, hey, the video game industry is up in arms because you know, there was all the concept designers are becoming obsolete, you know, but it's really the genie's out of the bottle and I don't think that there's any going back.

Andy Crestodina: 12:31

And personally, as a brand, I think that we have an opportunity to build trust in this moment by disclosing this and I think that, just as an addendum, sort of at the bottom of things, if it was made by AI or a watermark, if it was an image, I think that people will appreciate that and if your audience cares that, this could be good for you. But then again, I don't use AI to write anything. I use it to do research, I use gap analysis, I use it to analyze GA4 reports and look for correlations in survey data. But at least for me, my experience, content that performs well is super collaborative, like we're making right now. It's very visual, like we're making right now. It has strong points of view and takes a stand, which I'm doing right now. AI doesn't do any of those things well, so easy to differentiate content from AI. For many of us, this won't be an issue as a brand, but for all of us it's an issue as a consumer.

Jeannie Walters: 13:34

Yeah, I think it's a really again, it's an interesting question because, for example, one of the things that they're struggling with in education is where is the line of what's acceptable use and what's not? And is it? You know, AI is great at ideation, it's great at finding patterns, it's great at kind of distilling information that might be overwhelming for a regular person. So that can all be really useful if you're writing a paper, right. But if you are the one writing the paper but you're using those tools, do we have to disclose that? And that's where I think, to your point, the genie's out of the bottle.

Jeannie Walters: 14:17

This is the life we're living now, and I think what I thought was so interesting about those survey results and I just saw it again and it's 70% of people who basically said that they didn't know that they were dealing with AI. And I kind of wonder what that means, because we were talking a little bit before we started recording and you said, yeah, but we've all had these suggested next purchases, that's AI, right. And so it's kind of like I'm curious. I want to kind of go back and see how did they phrase that question? Who did they ask? When did they ask, because it could be that they were, you know, dealing in a chat and didn't realize they were dealing with AI, and that can be very frustrating because that can cause these loops that you can't get out of and and you think you're dealing with a person, but at the same time, if we can use AI to provide more service and more empathy in the moment, I think that's okay, right?

Andy Crestodina: 15:23

And it may be preferred in some contexts. I mean, there are, you know, would you want an AI therapist? That sounds absurd, until you realize like, wow, I wouldn't be squeamish about disclosing anything. Like I don't have to be polite, I can just tell you know, there are probably categories where people will suddenly embrace it and say, well, yeah, actually that's a I don't want. You know, humans, I gotta be. You know, you're a little more sensitive. You know, is this person? You know, can I get right to the point. Customer service, some even healthcare things. I mean, there it we may actually find that AI is a great way to use the internet or to get advice or to get a service. So that's all TBD. Politics very bad. It's going to be very good at manipulating us and triggering us, and so there's a big sociological impact. It's all kinds of things to be concerned about, but generally, yes, this is a great moment and humanity can decide. As a country, as people, we can decide that this should always be disclosed. I think that would be a positive.

Jeannie Walters: 16:28

Yep, and I think when it comes to disclosure and you mentioned, you think there are some disclosure regulations coming. I also would think about even like early search right, like when we think about how they would say like these results might not be right, like let's just say it, like we're doing our best, but these results might not be right. Google originally had that. I'm feeling lucky. Remember that button.

Andy Crestodina: 16:55

Yes, I do.

Jeannie Walters: 16:56

And it gave you like three instead of all of them, and so I think that that type of disclosure I think we just need to embrace as part of this and knowing that you know what machines make mistakes too, and especially when they're trained by humans, which all of them are, and we have to just accept that. And going back to your point about sometimes it's preferred in service interactions and things like that, I like to remind the teams that I work with, like, empathy in humans is a finite resource. We don't have as much empathy at the end of the day as we do in the beginning, and so if we are put into that situation where we have to provide that same empathy throughout the day, it's not going to be the same, and so the consistency of the empathy might be really valuable.

Andy Crestodina: 17:47

Oh, yeah, it crushes emotional intelligence tests. It's extremely good at identifying someone's mood and adapting itself to that person's mood. It's kind of a myth, you know, that people that you're not going to feel a close connection or it can't really understand you. It's going to be better at that than people. I saw a study. People like if you take Tylenol, you're slightly less empathetic. People like strange, like, wait a minute. There's all kinds of things that affect empathy. Like am I hungry or tired?

Andy Crestodina: 18:17

You know, I know for a fact that I'm not always empathetic. AI is inexhaustible. It does not have limited resources to apply to analyzing your voice and face. So I think that we will all come to appreciate it, but I think it's very important for us all to know when we're working with it.

Jeannie Walters: 18:37

So yeah, and that's where the trust comes in and that's where the transparency comes in. And when we talk to clients about chat especially, people really want to know if they're talking to a person or not in chat, and if you don't reveal that, that can lead to a lot of mistrust for the entire customer relationship. So I do think erring on the side of more transparent is where we need to go and at the same time, we as a society are learning this isn't so bad, and I actually this is okay for us to interact this way, but yeah.

Andy Crestodina: 19:09

Hey, Jeannie, I've got a question.

Andy Crestodina: 19:11

Suppose you're a college admissions officer and you're reading an essay and at the end of the essay there's a disclosure that AI was used to help write that essay.

Jeannie Walters: 19:19

Yeah.

Andy Crestodina: 19:20

Does that increase trust. What do you think?

Jeannie Walters: 19:22

Yeah, we are in the thick of that right now. We have a senior right now who is applying to colleges. So this is my life. Um, yeah, I think that all of education is struggling with this and struggling with where is the line. If I were that admissions counselor and saw that, I would not think of it as that student's work today, because we're assuming that to your point it was written by AI. But my experience, like yours, with writing is it's very, it follows a formula, it feels very formulaic when it spits out something, and I think that the more that we can really empower students and people in the market too, that the human voice really does matter, like we are wired differently by design, and having that true voice come out, I think, is just going to be more and more important as we move forward.

Jeannie Walters: 20:29

What would you do?

Andy Crestodina: 20:31

I could well in that example, I think I would as a college admissions advisor or officer. I would assume all essays were at least partly written by AI, and if I saw a student disclose that on their essay I think I'd be impressed.

Jeannie Walters: 20:45

Interesting.

Andy Crestodina: 20:46

I think it'd be different. I think it'd be like this is different. This is the only one who said this and if you're trying to stand out, I mean that checked at least one box. I teach a class. I teach a class at Northwestern it's for masters of communication students and I require AI. They must use it. My assignments require them to use AI to audit their homework before turning it in. I teach them the right prompts. I teach them to edit and improve prompts and evaluate the output of prompts or of AI responses. I give them, I think AI is an amazing teaching tool and the work of my students is better now than it was years ago. But if a student decides to use AI to do all their homework for them, they are wasting their money on education.

Andy Crestodina: 21:30

So if your goal as a student is to learn, it's incumbent on you, as the student, to push yourself to improve your skills. And if you're abdicating your job at communication, which is the main benefit of higher ed I think, by not writing anything yourself, I think you are going to be at a competitive disadvantage in the long run. You're not going to be a good public speaker if you can't put words together on the spot. I mean there's a million disadvantages for the student that does not learn to write.

Jeannie Walters: 22:05

Yes and I think that's a really good point too is that there are ways to incorporate it that are really meaningful and helpful and they're going to be the way of the future.

Jeannie Walters: 22:17

So I love that you're incorporating that. I'm noticing I have a college student and a high school student right now and I'm noticing how there are some things like everything old is new again, like blue books, right. Like instead of having them type things out, some professors are asking them to write in blue books in the classroom to make sure they're not really overusing AI. So there are some techniques. There's some math stuff that my senior here is doing and he has to take pictures of his handwritten work to hand in, so it's, and I sympathize because the teachers want the students to learn this and so they're trying these different techniques to figure this all out. But I think again, this all comes down to where we are in the evolution of all this and we are figuring this out as we go and we need to collaborate on that and we need to say, like I bet your students really appreciate that you teach them how to use it, because otherwise they're they're going to be it's all trial and error, right. Like we're all figuring this out.

Andy Crestodina: 23:23

Yeah, my job is to prepare them for the market, which is very competitive.

Jeannie Walters: 23:28

Yep.

Andy Crestodina: 23:28

And I don't think. I think it's. I think all teachers at least in college I mean should teach their students how to use AI. Why would you not? It's like, you know, in the 80s or let's say the 70s, should accountants use calculators. In the 80s? Should lawyers use computers? I mean it's, there is no. I think it's absurd. You're trying to prepare this person for the rest of their life, like to go compete well in the marketplace. But oh, don't use that super powerful, you know highly advantageous technology that might make your output better.

Jeannie Walters: 24:09

Yep.

Andy Crestodina: 24:10

Yep, so yeah, I mean, but then again I'm not a high school teacher, I'm not sure what I would do to help them learn the foundations.

Jeannie Walters: 24:18

Yeah. Well, and I think it's a good point because there are things that, um, you need to know even to use that calculator, right, and so it's like that type of foundational education is so important. And I keep asking, like, how are people even teaching business and marketing, because it's so fast paced right now, and I'm sure you've had this experience too, where sometimes I maybe interview or talk or network with people who have been in kind of a steady corporate job for a long time and it takes five minutes for me to realize they are not keeping up at all because they don't need to in their mind. But in today's world you need to, like everybody needs to be a constant student with this and experiment and try things and figure out what works, because otherwise there is no doubt you will be left behind, in my humble opinion.

Andy Crestodina: 25:11

Yeah, I think managers need to try to learn from their teams. How are you guys leveraging this? I think there needs to be, you know, all the best practices which you hear everywhere, like cross-functional, you know knowledge sharing, shared libraries of prompts, but literally there are tasks now that can be done in a week that used to take a week, that are now done in a day or less, and this has and we're this is 2024.

Andy Crestodina: 25:37

So most people aren't yet using agents which will combine prompts and tools and do actually do things. But no, I think that the anybody that doesn't still apply their own experience and knowledge and background and points of view is at a disadvantage. Really, when this becomes even more widely used, it tilts the scales toward people that are more, that are better at the fundamental human things, like you know strong points of view and making people feel special. You know offline interactions, you know events, all of the networking it's. It's those people will have a strong advantage. And look at how you, Jeannie, we've known each other for 15 years, maybe, yeah, probably more.

Jeannie Walters: 26:26

Yeah.

Andy Crestodina: 26:26

Yeah, that wouldn't have happened if we didn't get offline. So I think that this conversation is an example of people who just stay behind. You know, LLM their whole lives are never going to get the fun that we're having now.

Jeannie Walters: 26:41

Yeah, it's true, and I think that, as things evolve, the other thing that I'm seeing that I think is exciting is that we have these large language models, these LLMs, but now these organizations are figuring out their SLMs, their small language models, because you can do that just for your space, just for your organization, and sometimes that's a lot more valuable than you know, opening up to everything. So I think there will see more customization, even within how these things operate. And I'm going to skip our last one because this is too much fun. So we're just going to finish up here, but I want to. If somebody's listening to this and they're like, whew, I really just need to get into Gen AI, I haven't really experimented these two are really telling me I have to. What should be their goal, you think, for 2025? How should they go into next year thinking about this?

Andy Crestodina: 27:38

I'm just going to share my favorite use case for AI, which is to use it to find gaps and make the thing that you're making better. Whatever you're doing I mean, I'm a marketer, so any page that you're writing, or any email or any copy, any presentation, an ad if you're a paid marketer any ad it probably is missing a few things that are important to the target audience. So if you can first train the AI on your target audience. I have persona prompts for this. Connect with me, I'll share. There's also, it may be like an ICP that you have or a documented kind of persona. Give that to the AI, along with the draft of the thing that you're making, and ask it what's missing. You will suddenly become much more confident. You're going to be happy.

Andy Crestodina: 28:30

You've learned some AI tricks today, made you smarter and it didn't remove any of the parts of your job that you love. Right, being creative or doing analysis. These are fun things for me, so I like doing those. I don't want AI to do those for me, necessarily, but AI powered persona driven gap analysis amazingly helpful. Like oops, I forgot to talk about this. My audience cares about this but only works if you first train the AI in your audience. I often joke before you train AI on your target audience's information needs AI might as well stand for average information. It's not targeted, it's not specific, it's generic, it's water. It tastes like water because it ingested the internet and summarized that corner of the internet. But then train it. First train it on your target audience and it suddenly becomes breathtakingly useful, especially if you just use it to get another point of view on whatever your deliverable is.

Jeannie Walters: 29:24

I love that and I think this I would offer the same thing for, uh, I keep thinking about time and how, right now, the expectation is that I need that now. Whatever it is, right, like I need it now. If you tell me something's available, I want it on my doorstep tomorrow. If you are communicating about information, that is a lot to learn. I want to learn it now. And so where are the places that you can lean into shortening the time of whatever that delivery is and that would be? You know, I use it a lot for things like if we have some customer data points that we want to summarize, it's very good filtering through that and really discovering the priorities, the right priorities for those personas. So, yeah, it's pretty cool stuff.

Andy Crestodina: 30:18

It's a very dynamic category. We're in right. It's a fun time doing this.

Jeannie Walters: 30:23

I know you think about where we started and the technology we were using back then and where it's coming. in

Andy Crestodina: 30:28

In one lifetime it's so crazy. I

Jeannie Walters: 30:30

I know.

Andy Crestodina: 30:30

Really there's never been.

Andy Crestodina: 30:31

This is really amazing to see the growth of the internet and browsers and then mobile and social and all like we've seen these things. Just yeah, I, I gotta say like that I, uh, I actually feel for and understand the people who are like, oh, I'm, I'm 60s year now, I'm gonna sit this one out, like yeah, I appreciate that too.

Jeannie Walters: 30:54

I know, I know and the cool thing for me personally and I'm gonna say this because my dad listens to every single one of these but my dad was a systems analyst in like 1961 and two, so he was trained like there wasn't even a degree right like he was learning by IBM and all these others, and now my son is a computer science major at school and watching their interactions is really interesting because they the span of what they both witnessed in their lifetimes is pretty remarkable.

Jeannie Walters: 31:27

So, yeah, it's really cool stuff and we get to ride the wave together, right. Well, Andy, this was so great. Thank you so much for being here and thank you for all you do. For those people who haven't followed you, I highly recommend. There's always stuff to learn and I know you're always very gracious as well with your teaching and being out there for people, so thank you for that.

Andy Crestodina: 31:51

Thank you for this. I would never miss the chance to collaborate with you, Jeannie. This is awesome.

Jeannie Walters: 31:55

Thank you so much. Well, and thank you everybody for being here with us on the Experience Action Podcast. We will talk soon and talk again in 2025. Thanks for all you do.