Greg Kihlstrom - Continuous improvement of AI & customer experience
Greg Kihlstrom is a customer experience and marketing technology transformation consultant, author, and keynote speaker, as well as the host of The Agile Brand podcast.
In this episode, we talk about continuous improvement of AI and customer experience. We break down the concept of AI maturity and discuss how agility goes hand in hand with continuous improvement. We also speak about the need for responsible AI adoption, which Greg wrote about in a recent CMSWire article.
Links & mentions:
"Unless you're preparing for this, unless you're looking at things like the transparency of your models and how you're getting your data and all those kinds of things now, you're never going to be prepared for the equivalent of a GDPR or something like that, that's going to be coming soon."
Welcome to the agile digital transformation podcast, where we explore different aspects of digital transformation and digital experience with your host, Tim Butara, Content and Community Manager at Agiledrop.
Tim Butara: Hello, everyone. Thanks for tuning in. Our guest today is Greg Kihlstrom, customer experience and marketing technology transformation consultant, author and keynote speaker, as well as the host of the Agile Brand Podcast. We recently connected with Greg through a great article about the need for responsible AI that he published on CMSWire, and which we'll link to in the show notes, and we figured that this would be a perfect topic to cover here on our podcast.
So today, Greg joins us to talk about continuous improvement of AI and customer experience. Greg, welcome to the show. It's great that you're joining us today and that we get to discuss this with you. Do you want to add anything here before we dive into the questions?
Greg Kihlstrom: Yeah, no, just glad to be here and thanks for reaching out and, yeah, looking forward to talking today.
Tim Butara: Awesome. So the first thing that I want to kind of clarify, just so to make sure that we're all on the same page here, is what specifically does continuous improvement mean in the exact context of the conversation that we're having today?
Greg Kihlstrom: Yeah. I mean, I think I'm a fan, you know, my podcast is called The Agile Brand. I'm a fan of agile principles and, you know, I've seen them work very effectively. And, you know, some of those principles are related to continuous improvement, which, you know, in this context, I don't think there's a single thing that we should be doing that we shouldn't also be questioning and trying to make better over time. And there's not always a way to do that. And it doesn't always make sense to improve things based on the amount of effort to get the return. But we should always be questioning that. And there's nothing, that's a positive within organizations.
So I think, you know, when we're talking about AI, we also need to be continuously looking at that. I mean, you know, it's a fast changing space and there's, you know, even in the last few months, lots of things have changed and lots of ways of working have changed. So I just always, you know, want to be kind of questioning what can we do better and when should we do it.
Tim Butara: I think that a lot will have changed just from the time that we're recording this to the time that this gets published.
Greg Kihlstrom: Probably till the end of the show even, but yeah.
Tim Butara: Damn, yeah, right. And so you mentioned that it's important to be thinking about continuous improvement also in the context of AI, but I'm wondering, why we're talking about AI and customer experience, both in the same context?
Greg Kihlstrom: Yeah, I think customers have continually rising expectations and, you know, it started, we started, customer experience as a profession has been around for decades, but I think the marketing, the customer kind of centric parts of the organization have really been talking about customer experience for the last few years.
What they quickly realized was customers want personalized content, offers, experiences, all of those things. And when you start talking about personalization at a rudimentary level, you can break things down maybe by segment. So grouping large groups of individuals together and kind of making a judgment call on what might appeal to that group.
But you start finding more and more that the more you can individualize, the better the experience is going to be, the happier they're going to be. But you also... that's externally, internally, you start realizing, wow, we don't have enough people to make all these personalized offers and content and experiences.
And so AI comes into play here because it augments what humans are able to do. Like, we're always going to be the strategic, at least for the time being. When that changes, we have a whole other set of problems, but the humans are going to be the strategists. But AI can play such a role in just creating the breadth and depth of content and offers and experiences that need to be created to have a good experience.
Tim Butara: And that was true before the launch of, the public launch of ChatGPT and this explosion of generative AI and large language models, right? Because I'm guessing that for most people listening, it makes sense that, you know, at the end of 2023, that AI and CX would kind of go hand in hand, but it's, it's been with us, you know, either directly or indirectly in one form of another for quite a long time, actually. And I think that it's only been in recent years... I think that the documentary, The Social Dilemma was like the first time that the public kind of became aware of like, oh yeah, there's all of these algorithms, all of these ML and AI tools kind of in the background, running everything.
Greg Kihlstrom: Yeah. Yeah. I mean, yeah, like, marketing automation dates back to, you know, the early two thousands, if not earlier than that. And, you know, so that from an evolution from there, we always want to A) automate things to make our, you know, increase operational efficiency on the inside. But, you know, B) we want to make sure that customers are getting ,the more targeted we get... I mean, the stats prove that you can, you know, the stats are easy to find out there that point to customers that get something that's more tailored, have a higher, you know, conversion rates and lifetime value and all those things.
So, yeah, anything, I think, yeah, with ChatGPT, the AI conversation, I mean, I've been working in automation and personalization for years and years, but it takes something like a chat GPT or, you know, just some, something for people to really wrap their heads around, okay, this is, this is what they're talking about when they talk about automation and, you know, machines helping us and all that kind of stuff.
Tim Butara: And one other thing that I really want to discuss is the concept of AI maturity. So what is AI maturity and why is it important, especially in the context of our conversation?
Greg Kihlstrom: Yeah. I mean, there's a lot of like maturity. Every consulting firm has their own maturity model, I'm sure. And they're, you know, 10 times over, I've made my own before in various areas, but I think, you know, when we talk about maturity here, a first level might be, yeah, some of your team is probably using ChatGPT right now, but you don't know that they're using it. You don't know how they're using it. And you don't know if you're truly protected because maybe they're using some free version of some tool and your stuff... I mean, you know, we've seen the last few months, I forget the name of the company. I probably shouldn't name it anyway. But, you know, their source code ended up on the internet because someone entered it in a generative AI program.
And so, you know, step one is like, let's just acknowledge that the teams are using it, whether you want them to or not. So let's put some guardrails on, you know, it's going to be used, but let's use it responsibly. And then, you know, it kind of, it goes upward from there, where more sophisticated companies, they're not relying solely on external tools. They're building their own or they're training their own models on their own data because, you know, it's just specifically tailored to them. So, you know, there's certainly a graduated, you know, level of level of maturity. But, you know, I think it starts with curiosity and but, you know, companies out there who don't have a stance on this really need to get there quickly.
Tim Butara: Well, and also another major factor will probably also become, I mean, we're already seeing kind of the beginnings of this, all the regulations and kind of the legal stuff around that. And probably it's hard, even for the most sophisticated and the most kind of advanced companies, to reach really high level of AI maturity if they don't really know what the legal requirements and consequences of something will be, so we're probably, we're probably likely to see that develop over the next few years alongside these regulations and everything.
Greg Kihlstrom: Yeah, definitely. I mean, I think the easiest correlation here is the consumer data privacy stuff. So, you know, like, GDPR in the EU, like. Europe is probably going to lead and then, you know, others are going to follow there. That seems to be the trend and it seems to be happening right now in the EU, at least having something in writing out there and, you know, the US will be a few years behind.
But, you know, I lived through the GDPR stuff. And now currently in the United States, there's a big push for healthcare, like personal health information, that aspect of privacy, and it is disrupting the industry in a bad way. Like, you know, there's good disruption that like takes things forward. Healthcare providers in the US are not prepared for this, you know, some announcements that happened now months ago, you know, at the beginning of this year.
So I use that parallel to show like something's going to happen and exactly to your point, unless you're preparing for this, unless you're looking at things like the transparency of your models and how you're getting your data and all those kinds of things now, you're never going to be prepared for the equivalent of a GDPR or something like that, that's going to be coming soon.
Tim Butara: Well, but it was a major scandal in that context that kind of led to GDPR. So I'm hoping that, you know, some kind of, I don't know, a solution is made possible before a similar size scandal occurs with something like ChatGPT. I mean, but you know, something like the example that you mentioned before about company data getting leaked, but because somebody entered them into ChadGPT, that's all kind of on the border of that kind, that type of scandal.
So maybe if there are more of these smaller scandals or like the stuff that we saw, like with ChatGPT getting somebody on Fiverr to bypass CAPTCHA, did you see that?
Greg Kihlstrom: Oh, yeah. Yeah.
Tim Butara: I mean, like, CAPTCHA exists solely for the reason to not let AI pass off as a human. And then ChatGPT, or maybe it was another, another similar tool, but I think it was ChatGPT, you know, if they're able to not just bypass CAPTCHA, not just think of the option to bypass CAPTCHA in such a way, but actually trick the particular human into like confirming that, no, don't worry, I'm not artificial intelligence.
Greg Kihlstrom: Yeah, I mean, it keeps getting, you know, it' that there's I think the criminals take one step forward and then, you know, the cyber security, you know, component is, sometimes they're ahead, sometimes they're like, you know, a step behind. It's sort of that continuous, like, battle, you know, neck and neck sort of race between things.
So, yeah, I mean, it's going to keep, and it definitely, you know, there's some scary stuff, but there's also some stuff that just reminds me of like the early days of photoshop where, you know, everyone was like, oh my God, we'll never be able to tell if a photo is real or not. And now like, sorry, but I can tell when something, like, I think people's grandparents can tell that something is photoshopped now versus, you know, 20 years ago. It was scary, I don't want to discount because you know deep fakes and there's a lot of things that are getting incredibly realistic and that's kind of scary, but I do feel like, I have no scientific evidence to support this, so, please, you know take this with a grain of salt, but I feel like we're evolving to be able to spot this stuff better over time. I think we're a few years behind the technology in that way, but I do feel like we're evolving to be able to spot at least some of this stuff. I think you're definitely right.
Tim Butara: But the problem that I see here is that the technology is also evolving. And I think that the technology, like you said, that we're a few years behind this. But the technology is evolving at such a pace that even if we weren't, it would outpace us. So I think that it was even like the examples of, like, hands in, like, AI generated images have gotten really, really, really better and more realistic just over the past, like, not even 12 months, maybe like six months or something.
So, that's definitely something to, to keep top of mind. But to your point, I think that it's also true on the other hand, that, you know, at least for those who are in touch more frequently with AI generated stuff, then at least those will be able to kind of develop a keener eye for this and be able to spot this more frequently and more accurately, even as the technology evolves to be more accurate itself.
Greg Kihlstrom: Yeah, well, and I do think, you know, this is where regulation is a good thing. And, you know, sometimes it can be onerous and not support innovation, but I think we need it just like we need it with consumer data privacy. Again, I'm a marketer. Like it makes marketers lives a pain to have, you know, all these consumer data privacy laws, but we need them because there are, it's not even bad actors, it's just people trying to do their job, but it interferes with someone's, you know, privacy, we need this for AI because we just can't help ourselves and, like, I don't, I know there are some bad actors out there, but I believe on the whole, it's not about that. I think commerce gets in the way and good intentions for other things get in the way of what's actually good for, you know, people. So I think we need regulations for that and it makes it easy.
Tim Butara: Well, and it's also just, just the, like the inherent feature or functionality of something, right? Let's say that if you have an algorithm whose sole objective is to drive engagement, then the algorithm will do everything it can to help you drive engagement, even if that leads to some unethical or, or maybe... Practices that are unhealthy or damaging to people's mental health or to the overall society consciousness. Because it's not even about, you know, it's not about a positive or a negative goal. It's just a business goal of having more engagement. And it's as simple as that. And then that's why I guess a lot of people are talking about the risks of AI, because something as simple as a primary objective of a particular system can have, you know, unpredictable and devastating consequences in the longer term.
Greg Kihlstrom: Yeah, yeah, I mean, you know, one of my favorite movies is 2001: A Space Odyssey, you know, Stanley Kubrick, like very simple, you know, early and this is in the 60s, like very simple lesson, like the mission was the objective that the AI needed to achieve and the people were extraneous to it. So what did it do? It killed off the human or tried to kill off all the humans because the mission was the most important thing.
To your point, I mean, marketing, we're not kind of dealing at that level hopefully, but you know, if marketing is the objective or if getting better customer engagement is the objective at all costs, you know, your point is validated and that, you know, we need to put some guardrails in place because the machines don't know, they're just trying to accomplish the mission.
Tim Butara: This leads perfectly into the next question. So I wanted to kind of touch upon the article that you published recently for CMSWire that I mentioned in the intro and kind of kind of talk a little bit about why it's so important to develop digital or customer experiences responsibly. And I assume that this importance has greatly increased in the last few years, let's say.
Greg Kihlstrom: Yeah, yeah, I mean, definitely. It's, you know, it goes back to businesses need to stay competitive. Customers keep seeing things elsewhere. You know, the example here would be, you know, like, Amazon next day delivery has spoiled us all. And now there is an expectation, if you buy something online, it could be from the smallest one person company in the world to the largest, you better get that next day or else, you know, Amazon ruined it for us, basically.
So that expectation and that, but that's one example of many, it's like, someone sees something and whether they're a B2B, B2C, whatever customer, they have an expectation from somewhere else. And so we've got to meet those demands or else, you know, there's going to be only one company out there that, you know, serves those needs.
So companies have to be competitive, but they also, again, can't just get distracted by the mission and not treat their customers' data and privacy and all those things well as well. So, you know, it's this continuous, like, balancing act of how do we do both? And, you know, I think one easy way to do that is when you ask for something from your customers, give them something in return.
Like that's, I think brands that are being successful, they're like, they ask for stuff and they know a lot about their customers, but they also use that information to benefit the customer. They don't just use it to sell them more stuff. They actually use it. The customer is happy to do that because they get something back, you know, customer loyalty programs.
That's an easy example of, you know, if you keep using the same airline or staying in the same hotel chain or whatever, you get something back. And so you don't mind that they know that you stayed at their hotel last week It gets you free stuff. You know, that's a very simple transactional version of that, but I think it goes broader.
Tim Butara: Well, and personalization is an interesting topic because sometimes over personalizing can have a negative impact on the customer experience, right? For example, if you buy a product, and then the AI constantly keeps showing you the same product, or the same product in a different color or something, maybe it's the kind of thing that you don't really buy like once a month or once a year, maybe you buy it once in a lifetime, once in 10 years.
So this point that you made just now just really highlights this kind of importance of, you know, right, we know a lot about our customers. But we want to use this knowledge to make their experience better rather than just sell them more of our products, which could have a negative impact if this advertising aspect was not approaching the right way, I guess.
Greg Kihlstrom: Yeah, well, it also goes to data quality, right? Because I mean, a lot of, and this is why, you know, in addition to consumers' data being protected. It also, it's better when brands have first party data about you and know, you know, I just got a new car a few months ago. Okay. I'm still seeing, to support your point, I'm still seeing banner ads to get another car. I don't need a different brand, let alone the same brand again. I'm good with one car.
So like, you know, that kind of thing, it's like as a consumer, yeah, I don't really want all these third parties knowing that I, you know, want to buy X type of car or whatever, but also, I just don't want to be shown cars anymore. I'm good. Like, either I made the right or wrong decision, but I got to live with that for a few years at least.
So, you know, this is where consumer data privacy plus brands owning their own first party data about about customers is a win-win, you know, everybody wins. And, you know, I think this move is so much better versus again, a company buying some information. I made some inquiry on a website 6 months ago. They're still selling me as a likely car buyer. No, that's not true anymore.
Tim Butara: And the car example is a really great one because a car is by default something that's supposed to last you at least a couple of years, right?
Greg Kihlstrom: Right.
Tim Butara: So if you keep getting, like, I know that the algorithm doesn't see it like that, obviously, but if you keep getting like, if you buy a car, and then you keep getting ads for a new car. It's like some kind of way of saying that, oh yeah, your car that you just bought, it probably isn't working anymore. So you need a new one, which is like, you know, why would I buy a new one if the one that I just bought isn't working anymore after just a week?
And it, I know that that's not the logic behind it, but to the consumer, it could seem like that. You know, obviously if you have faith in the product and the product is supposed to last years, then, you know, if you know that I made the purchase, then it would be detrimental to kind of keep advertising the product because that would signify this, that I just mentioned.
But yeah, a lot of interesting stuff with advertising, and so it's, on the other hand, it can be a real treat when you see companies actually using your data, you know, to kind of benefit you and to advertise in a non obtrusive and maybe a helpful way. So that's definitely the other side of the equation that you're then even happier to see because you're sometimes so disillusioned by this other more negative aspect.
Greg Kihlstrom: Yeah. Yeah. I guess that's an interesting way of looking at it in that most of it is so bad that when you actually see something relevant... that's an interesting study in and of itself. But yeah, it's true. Like I fall for, I mean, marketers fall for marketing probably more than anyone probably, but like, you know, I fall for marketing all the time when it's tailored to me and it's not a trick at that point, like it's actually benefit, like everybody wins, you know, and I love that part of it because again, I don't think companies, they're not going to waste advertising dollars just to like stalk me around the Internet when I don't need their product, you know, they're just, they're throwing money away because they, they don't have better information.
And yeah, to those that are less kind of tuned into this stuff and how it all works, like it's just creepy, but useless, you know, it's one thing if it's a little creepy because you don't quite get how internet tracking works or whatever, but it shows you exactly what you need, then honestly, you get over the creepiness factor, you know, pretty easily if you're served with something you need, otherwise, yeah, it's just, it's a lose lose the other way.
Tim Butara: That was a very good point, yeah, because if it's useless, then the only memorable and significant thing about it will be the creepiness aspect. So that will stand out even more.
Greg Kihlstrom: Right, exactly.
Tim Butara: Okay, so to kind of tie everything together and start driving the conversation to a close, how can agility or agile practices help with continuous improvement of AI and customer experience?
Greg Kihlstrom: Yeah, definitely. I mean, you know, kind of going back to the maturity level, you know, every organization, what I like about maturity model, I mean, what I don't like is there's too many and I think people use them wrong often, but what I like about a maturity scale is, you know, everybody is at a different level and that's okay.
And the idea is to continuously get better wherever you start and wherever you finish. And so, you know, to me, the agile approach of, let's always be looking at this stuff. I mean, you know, there's external factors. The software is getting better and changing and there's regulations and all those kinds of things.
So there's a lot of things changing there. But also internally, people are just getting better at... it's like search engines, you know, 20 years ago or so, people are getting better at googling, you know, nobody called it a prompt engineer back in the day, but, you know, now, now people are just getting better at finding information.
And so similar to that, people are getting better at using ChatGPT and like these no-code/low-code tools. And so, you can't just say, okay, we're going to, you know, we're going to adopt the set of AI guidelines and ways of working today, and then, you know, keep it for the next two years. Like, you've got to be continuously adjusting, larger organizations need to be seriously thinking about building their own internal tools that are proprietary and, you know, domain specific and stuff like that as well, because, you know, it becomes competitive advantage, just like customer experience is a competitive advantage if you do it well, AI and your own models and ways of doing things, that's going to be a competitive advantage soon enough. I think most orgs are a little bit immature in the space, but that is fast changing.
Tim Butara: So continuous improvement and agile definitely also go hand in hand also in this context.
Greg Kihlstrom: Yeah, yeah, absolutely. Yeah. It's just, it's always looking at what can we do better and, you know, what's changed, internal, external, and then feeding that back into a loop to, you know, do it on a on a regular basis. If you don't build in, if you do a transformation and you adopt AI and all that, and then you just leave it, then you're going to be in the spot that a lot of organizations are with technical debt and just rethinking and changing and really expensive changes.
If you build in continuous improvement from the start, then you don't have to do a redesign, re-, you know, re-transformation, whatever you want to call it, two years from now, because you'll be on the same path, you'll just be continuously improving.
Tim Butara: Well, and also, things are changing so fast that if you don't do it this way then you'll just be kind of eternally playing catch up because you'll always be transforming for yesterday, which will have been obsolete by the time that you finish your transformation.
Greg Kihlstrom: Yeah, I mean, an analogy there is, you know, I owned a marketing agency for a while and we built a lot of websites, right? So the traditional way of website design and development is, like, every, you know, depending on the industry or whatever, every three to five years, it's like buying a car or leasing a car or something. It's like, you know, every three to five years, it was like, okay, we got to redo the website. So what do you do?
Non scientific? No, it's completely arbitrary. Let's scrap the old site and completely redesign from blank slate. And then, you know, it's a, depending on how big you are, it's a, you know, 6 to, you know, 24 month process, depending on the complexity.
And so what's changed in the last 2 years since you started redesigning your site, to your point, you're already obsolete by the time you launch and then you need to be planning the next redesign. Like, it's inefficient. It's not a scientific test at all because you change every variable. So there's no way to do an apples to apples comparison, versus continuous improvement.
And, you know, companies like Amazon, for instance, you know, when's the last time they redesigned their entire website? You know, one way of looking at it is, like 20, 30 years. The other way is they're constantly redesigning everything because they're always testing, but that's the way you do it these days.
Tim Butara: That was a perfect example and a great note to finish our conversation on today, Greg. Just before we wrap things up, if listeners would like to reach out to you, learn more about you, connect with you, where can they do that?
Greg Kihlstrom: Yeah, two places I'd recommend. I'm really active on LinkedIn, so please, you know, search for me. I'm always happy to connect there. And then you can reach me on my website and it has links to the podcast and books and everything. It's gregkihlstrom.com.
Tim Butara: Awesome. We'll add both of the show notes as well as the link to the CMSWire article, as we also promised in the intro. Well, Greg, thanks again for joining us. This has been a great discussion. I love how we just talked about our experiences with this stuff for a while. And yeah, thanks for joining us. Happy to have you here.
Greg Kihlstrom: Yeah. Thank you so much.
Tim Butara: Well, to our listeners, that's all for this episode. Have a great day, everyone and stay safe.
Thanks for tuning in. If you'd like to check out our other episodes, you can find all of them at agiledrop.com/podcast, as well as on all the most popular podcasting platforms. Make sure to subscribe so you don't miss any new episodes, and don't forget to share the podcast with your friends and colleagues.