Disambiguation: AI and the Customer Journey - Transcript
Michael Fauscette
Welcome to disambiguation. I'm your host, Michael Fauscette. Each week we interview experts in AI generative AI and business automation to help business leaders understand how to use these tools for the biggest business impact. In our show, today, we look at AI in the customer journey, I'm joined by Jeff Nicholson industry veteran, several decades in AI before it was called AI, CRM process automation space for several decades. And he had a hand in creating a bunch of different categories across technology segments. Jeff, welcome. Great to be here.
Jeff Nicholson
Michael. Good to be back. Yeah, thanks.
Michael Fauscette
I appreciate you joining today. I, you know, we're going to talk about AI cuz, you know, we always talk about AI on the show, but, but before we kind of jump into the AI piece of it, I kind of wanted to start by setting the groundwork a little bit around this concept of customer journey. And I know it's a term we use all the time. But, but I'm not sure we all, you know, think about it in the same way. So, I'm curious, how do you think of it? How useful do you think it is? Are there pitfalls, shortcomings around it? And then just in general, how can companies improve it?
Jeff Nicholson
Well, I think it's an interesting hot topic, especially when we're talking about the concept of AI, which is a lot of hype, a lot of noise. And many are looking for, you know, what's the practical application. And I see a lot of vendors that make it seem like you just turn on their system and automatically will solve customer journey. The customer journey is an interesting thing. Because if you really have an outside in perspective, maybe that journey touches you a few times if you're lucky. And you may touch your marketing, if you're lucky, there's a lot of research that is on the part of the consumer and buyers today, that's not involving you whatsoever. My touch on marketing and like touch your sales, touch points, autonomous touch points that are selling, even humans in the mix, they still exist. Any successful making that sale, customer service, customer support functions, building functions. And the journey is cutting across all of these things. And if you take a step back and look at how enterprises are set up, they're not really set up as one big thing, right? That's why sobbing journey is so hard. And when you throw AI in the mix, I'm seeing, in many cases, AI being deployed in isolation in a silo doing one small thing. But it doesn't really cut across the journey. And Michael knew, and I, you know, I've talked about journey in pasture, the concept of journey maps, as if you could just put it on a piece of paper, and then it's done.
Michael Fauscette
You know, I never could figure out a way to get the customers to follow my map. Right. And, you know, you hit on another thing, too, that I think is really important as we carry in to the AI conversation to the silos that organizational silos, we know are a serious problem when you talk about customers, but the data silos that we've created through the years of all these different systems, I mean, that also has to be a real hindrance to the use of AI in any of these processes, right?
Jeff Nicholson
Well, absolutely. And we're seeing organizations try and get their heads around this. And they were forced a few years back in many cases to start thinking about their data from a governance perspective and a different way through legislation like the GDPR CCPA. And where's the data? Where's it held for? How long? Where's the PII? Where do we have to redact it and de identified for use cases, whereas it should be used and not used. And so many businesses thankfully had had a running start, but now that they're they've been able to push right to the, to the rate comes to AI and masses of questions that organizations have to solve for here, when it comes for bias in your mind's eye using data and using it in the right way. So yeah, all that should be top of mind.
Michael Fauscette
Yeah, that makes sense. I mean, it what, as you start to think about AI and generative AI, which is obviously the topic of the year, what do you what are the most exciting, promising ways that you're seeing companies use these tools to build a better digital experience enhance the customer journey?
Jeff Nicholson
Well, there are quite a number of them happening now. And I think what we should be looking for are those practical use cases. There are things that will happen someday at scale. But what are the things today, we touched on what we touched on marketing, we touched on sales, we touched on customer service, and I think there's use cases in each of these. Probably a good to start with the marketing side. And, you know, one of the areas is kind of the personalization zone, but not the kind that we've talked about in the past person has just been around several decades. Right Michael? Started from dear insert your name here and an email and that was not otherwise I still get those Michael. I don't know if you do?
Michael Fauscette
Oh yeah!
Jeff Nicholson
So that's still out there. But then we got more clever. And we said, oh, this, this offer is the offer of the week. And then we said, well, this offers for you. And it was the same offer. We offered everybody else in that segment. And maybe we customize the image to be because you clicked on Italy. Here's a picture of Italy. And that's all great stuff. Well, I think it's gonna buckle is towards real individualization. You know, is not just the marketing message that's going to be shaped here. What are your thoughts?
Michael Fauscette
Well, you know, I mean, you hit on that. I love the word individualization. And I've kind of bounced that around in my head for years when I think of personalization, because I, you know, when you, I always have this story about one of our large online retailers, that, that I'm a photographer, I buy high end photography equipment, and I've done that through this retailer, and others for years. But one time I bought a Christmas present for my mom, it was a little $149 point and shoot camera. So now, almost every day, at least weekly, I get an offer. Here's our most popular, you know, little cameras, here's another $119 $149 thing that you must want. Because one time 10 years ago, you bought one. And that drives me crazy. Because they have lots of data, why don't they use it? Right?
Jeff Nicholson
And that's generally the attitude. If more and more customers are aware of that you have their data, they're expecting it to at least use it to help them along that path. If you get it wrong enough, they're gonna take that right back.
Michael Fauscette
That's the value proposition, is that I give you I let you use my data, I'm supposed to get something back, right?
Jeff Nicholson
Precisely. So, AI is one of those areas that you can apply it to help you down that path. But really, there's some examples, given, if you look at how marketing teams are using AI to do, they're not using it for it to help them summarize content, right, from the right and abstract and take any book and boil it down to a nice little subject line. That's, that's great stuff. But where it really has the opportunity to, to help is actually taking it a step further to the individual individualization you talked about and consider that even the products that we're offering can change in that market message. So think about an individual that may be clicking on sport coats might be clicking on everything blue, not buying and moving through the site, and you're leaving those behaviors, those footprints that you talked about Michael, but we're just presenting the same things that we may or may not have in stock, frankly, right, where it's gonna go, I believe, is to the air of actually wrapping that product around, and what if you actually can show that sport coat in blue with texture, because everything person is flipped to the head texture, and it doesn't have three buttons, it has to because everything has had to for some reason, they're always clicking on a jacket with no pocket on it. Well, that doesn't exist. But with this technology, there's AI that can actually generate that image. Firefly are things that we're using today, Stable Diffusion is not a one, Michael,
Michael Fauscette
The data, we have lots of data. And I mean, you know, like we knew some of the, you know, product analytics kind of companies like Amplitude, have behavioral databases, and they track all these different breadcrumbs across the whole interaction that you may do there. And, frankly, you know, intent data from across the web, so you get it from other places, too. So, it does seem like a logical time that we, we could now take that data, and have an engine that can really individualize that for each of the customers as they go across that experience, which is exciting, frankly.
Jeff Nicholson
It absolutely is exciting. And it's using it to wrap it around you that individualization it's not just a marketing message. So, you get that small code example the product is built for you, it might take you 14 days to get it. And by the way, brands are doing long versus letting my daughter disorder one letting you design your own shoe, but you design this is where the using those digital footprints at mass scale, you can produce these images using generative AI for image not just text alone. Yeah, well multimodal in many ways here. And that that and what makes it super interesting to me as a lifelong marketer that when you think about the things that you can sell them next at Mass, your system should alert you if it spots trends in these customers. And then you can spin it full circle into your segmentation strategies. And yet again, rinse and repeat, rinse and repeat.
Michael Fauscette
Well, I mean, you hit on another thing too. I mean, that actually even ties back into the back end of the business right to the supply chain or manufacturing or whatever it might be dependent on the business. You can actually start to predict demand of different types of items or variations of items, right colors or textures or whatever. And actually make sure that you have that just in time and save both on overhead and also improve customer satisfaction. I mean, that's amazing.
Jeff Nicholson
Use it for good, not evil, right?
Michael Fauscette
Yeah, exactly. You know, I was I was reading something this morning. One of the conversational, you know, sales tools that people use a gong, announced that they've now added a generative AI capability that improves forecasting by I forget their number, but maybe it was 30% or something, because they're taking the data that they're collecting, and all those conversations across all of the conversation for all those years and abstracting that into ways to predict. I mean, it seems like that from a sales perspective, some type of tools that make the salesperson more effective would be a really important place to focus when you when you look at AI and the ways you could use it.
Jeff Nicholson
Yeah, and that's what I'm, that's what I'm seeing. And that's what I think the hotspot is when it comes to sales automation technologies, we're seeing kind of two hemispheres happening. One is on the productivity on the seller, and I'm seeing a lot of that, where you can use Gen AI to, for example, summarize the conversation. You can use it to extract action items and meeting notes, scheduled meetings and things like that. And those are, you know, hugely helpful, but they're kind of party tricks in some sense, right? Yeah. You do what you shouldn't be doing anyway. They're just giving your time back. And my question is, what's you doing with that time? And that's what the example you just shared? I think that's why I think it's so interesting, because you can use AI to actually shape what you are doing materially to affect them.
Michael Fauscette
Yeah, I like that idea, that you sort of have two paths with AI in one part is, what can I do that removes friction from that person's role, their process? Whatever, right? So that's automation. And, you know, like you said, summarization, data entry, all those things that people shouldn't really have to spend their time on when I want them to sell. And then the other half is, what do you do to give them better insights into what they're doing? So, they can focus correctly and make better decisions? Right?
Jeff Nicholson
Absolutely.
And the example you gave was going where the AI is learning at scale, what happens to be working on that working, really surfacing that to you to make you more effective, not just efficient. And that's where I think that the real Holy Grail is. And if it's all wrapped around that journey, that we talked about using those footprints and doing it right.
Michael Fauscette
Well, you know, so we're talking a bit about the customer. And I think that's really important, but I just wonder if you turn the lens around a little bit and look inside to and I guess one of the things we've mentioned the automation idea, or the or removing friction, that probably does improve the employee experience, but how are businesses doing that? Like, are they applying Gen AI internally, to improve the employee experience as well? And what how does that play into that? And CX? I mean, how does that play into interaction with the customer?
Jeff Nicholson
Yeah, absolutely, I'm starting to see that. And you think about, you know, the employee experience is part of the customer's journey, they are not just touching autonomous touch points, they're touching employee type things. And they're gonna have to surface insights to you from their enterprise applications. And by the way for themselves just in their daily life. Today, they're, they're interacting with one application at a time, that's one of the major differences. And today, they're, they're querying us something specific and surfing to try and somehow find a needle in the haystack and the data in the future. They're not just gonna be clicking, they're gonna conversing. And I'm starting to see the applications emerge and conversational AI platforms are out there, let you actually have a conversation. But I think is the probably the most interesting part, Michael was that the conversation doesn't happen only necessarily with a single application, but it can happen across it. He'd be asking, you know, how much time if I don't take my time off through the year? And how much will I get back after taxes considering the following things?
Michael Fauscette
There's no system that's got that ready to go, but no, or even what, you know, if I change this in my process, what does this do? Or if I, you know, whenever I can, I cannot get rid of some of this data entry work or have an assistant that does things. I mean, that to me, that seems like that would make a big difference for the for the employee. And then if you took that into the customer service world, and you can literally give an assistant to that employee that would help them solve customer problems faster or actually, you know, be a part of the process. That's that, that also seems like that improves both that improves the employee experience and customer experience or without a doubt.
Jeff Nicholson
And for years, we tried to surface kind of custom scripts into the screen prints for someone to some of it dynamic on logic. But now Jenny, I can actually surface something more meaningful. And even just think about knowledge articles, and how they've typically been in your frontline employees trying to speak to a customer at the very same time as they're reading content. Yes, 30 pages long. It's just challenging job. So, if you can use Gen AI to summarize that, not just the article, but in the context of the customer query that's at hand right now. And here's what I think the issue is, then we're talking about something more manageable.
Michael Fauscette
I mean, the I think the industry standard is that a chat says customer support person on chat can handle three, a good one can handle three to five interactions at a time. I mean, to me, I always thought that was crazy. Like I couldn't barely do one interaction at a time. But if I was trying to read things and figure out how to solve a problem, I mean, that could make a huge difference in my ability to manage that many interactions at once.
Jeff Nicholson
Absolutely. So it's not just what you can do at the same time, Michael, it's how you do it, and then you wrap it around the journey, and you and all these different forms of Gen AI, will be able to help visual audio, text based, all these multimodal emergent approaches that are popping up, and your ability to understand the context, wrap it around, the customer is one of the major game changers.
Michael Fauscette
well, you know, one of the other things that as I was thinking about marketing, one of the things I was curious about, and you know, obviously from your perspective, as a practitioner, how do you keep consistent branding. you've got tools that generate emails, and you've got email campaigns, and you've got campaigns on social and you've got, you know, eBooks and content marketing and all this stuff. How do you keep the brand consistent? I mean, is that something that we're worried about from, it seems to me like that could go in all sorts of crazy directions.
Jeff Nicholson
You nailed it. And this is one of the greatest challenges and opportunities when it comes to GenAI. And when you can send it out, technically, anybody can generate content from across systems. The danger is that that's just getting created and sent over the over the hill to your, to your market to your customer. Without the checks and balances in place. Is it hallucinating? Is it right, is it in our brand tone. And actually, there's a there's a wonderful analyst at Gartner named Andrew Frank that has been doing some really forward looking thinking around the subject of brand. And he believes that it's not only gonna be able to help you police that brand, as content is going out the door and understanding where things are on and off in terms of tone, in terms of color, in terms of sharpness, of image, all these types of interesting things. As it generates content. It knows how to keep it in tone with your brand tone in a way, frankly, you could never even think of doing before that scale.
Michael Fauscette
So, it actually lets you set guardrails that could increase the consistency in a way that you probably never could have done even in one team, let alone across all the organizational silos that you're talking about that do touch the customer.
Jeff Nicholson
Today, I mean, that's yeah, it's always been a challenge, Michael, for marketing teams with content going out the door, in all different directions being created at all points across the organization. Now I both is making that exponential. So, credit, almost a bigger problem, but it's also introducing a solution we've never had.
Michael Fauscette
Yeah, that makes sense. So that kind of brings me to another question. And this just sort of hit me. And I know for a while CDP's, customer data platforms have been a big thing. People talk about him a lot. But every time I look at them, for the most part, now this is starting to change. They were talking about it as a marketing tool. I mean, do you think that CDP's play a role in this from an AI perspective? Do they give you the capability perhaps to overcome some of these silos? And is that the right sort of way to think about it?
Jeff Nicholson
There's some debate around the lifespan of CDP's and kind of where it's heading. I think, ultimately, it's around taking step back and saying what does the data we have? What's the data we have a right to use? It and what does the customer expect us to have? And how do we use that ethically in a way that's proper for that customer, respecting PII, respecting all the different things we should be respecting? And having a system of governance in place across all the data sources and it's a major challenge. I wish there was some magic one to that, but it's one that every CIO is gonna have to face.
Michael Fauscette
Yeah, but one of my guests not long ago, we were talking about compliance, and you know, regulations and governance and that sort of thing. And one of the things he said that I thought was a really good lesson for the audience and I'll share it with you see what you think but his statement really was companies have to be intentional about that governance and compliance process upfront. So, as they're planning what they're going to do, they need to include that strategy at the time versus we have this thing working and out, oops, we need to, we need to figure out a way to, to comply. And it seems like a simple thing. But it also seems like something that probably is a thing that would be missed, it can easily be missed.
Jeff Nicholson
And whenever you're thinking about data, it's interesting, when you think about it, as you don't need to use all the data all the time, right, you need to use specific sets of data for specific use cases. So, I always think about it take a step back, what is the use case that you're even thinking of? And how will it impact the trajectory of that customer journey? If you only could? And then what data do you need? And let's make sure you're focusing on the right problem.
Michael Fauscette
Yeah, that I mean, that, again, that makes sense. And it seems like that could tie into best practices. And so like, in fact, that's probably a good place to go next is to think about across those different functions. I mean, what are what are some of the best practices that you're seeing today, for people that are integrating generative AI and AI systems into their, into their customer management systems?
Jeff Nicholson
Well, a couple, first and foremost, is actually getting in the game. So not putting your head in the sand, your competition is doing these things are getting more efficient, more effective. But you don't have to solve it all at once, and you can't, and the technology is changing, you couldn't possibly frankly, Mike, right, the technology changes so fast. We've already blown past GPD, great belly, four, and on. And now there's Google bar. There are other types of approaches that are just exploding across everywhere, and you can't be an expert in absolutely everything all the time. And so don't, don't be overwhelmed by like, get started. And when we think about these, you need to be implemented systems of trust. And that's trusting the systems from an employee standpoint, so that you can achieve custom in what you're doing from a customer journey standpoint. And to do it is about testing and learning. Let's test this out, let's watch it closely. Make sure in the early days, we do have humans in the mix. And I think of it very much likely the old days of progressive profile and not the old days, you should still be doing. Marketing, where you don't have out of the gate of complete profile. But progressively you're building and you're learning, you continue to get a better picture. The same thing is happening over time, when you're deploying these AI, we have the humans in the mix, if you're finding that 100% of the time in this type of a setting, when we're using this AI, it's actually producing the right the right type of content, and its low risk, I can let that fly. But these other areas, building confidence, I'm not quite there. And so, you need to be able to regulate these things and build systems of trust. And my advice is also not just from employee standpoint, but also look at it from a customer standpoint, try if you can to get feedback from the content but my advice is don't go nuclear with it. So, I'm talking with thumbs up thumbs down. How are we doing type stuff I'm talking about here filling complete the survey with a pop up this slammed into your screen? Nobody likes that.
Michael Fauscette
No, no, that's a sure exit strategy for me if I see that I'm gone. Right? I'm not doing it. Yeah, that's good. I and I think it's important to say that you're collecting feedback. And you're going to use that to refine and I like they're particularly in most of the, you know, GenAI tools that you play with as a consumer, they have that little thumbs up thumbs down. That's it, you know, it was the answer good, or was it not good. And, and, and that, over time helps it tune and learn and, you know, perform better? I think that's really effective. And quick, right? It's frictionless or nearly anyway. So, you already sort of hit on a little bit of the advice, right? So, you're saying they should be doing it, and they should be keeping humans in the loop until you build trust and, and, and that sort of thing. But I'm curious, what else would you say from a standpoint of a business? I'm thinking about jumping in with some of the different customer areas within AI tools. What advice would you give them?
Jeff Nicholson
Well, I think, start engaging with the various experts that are out there in the industry. There's great thinking coming out of McKinsey out of Forrester, Gartner, Accenture is doing good things and also new challenger vendors that you trust and have them share with you and challenge them to tell you what are the things that you should be thinking about their technology, they should be close to it, but they may be bias.
Michael Fauscette
They will be biased.
Jeff Nicholson
yes, yeah. Challenge them to for their recommendations because, again, people coming at this very episode slightly different perspectives with different flavors of AI. And what I'm seeing right now, Michael, is that there's no one kind of atomic flyswatter that does everything. When it comes to AI, you might test it a Google might get there someday or meta. But right now, flavors are good for very specific things. And you're often going to take these open source ones and be very concerned with your data privacy and make your own flavors of it. And they're going to chain different models together for not just different queries, but even different pieces of a query. So that's why you need to start building up your knowledge internally on it. And again, challenge the vendors that are coming to explain to you where did they believe that they can help you most and, and honestly, we're not.
Michael Fauscette
Yeah, I like the use of the word Trust. And I've seen this now. And a few of the, you know, enterprise vendors that I've taken a look at that are starting to embed AI in their systems. And, you know, like Salesforce calls it a trust layer, Sugar CRM, I did a deep dive with them around their trust layer. And one of the things I've seen is, there's a certain amount of automation that's starting to happen in here, too. So, like, if you're, if you're passing information into a model, some companies are using a filter type approach, which just prevents things from going through that filter that you don't want in the model, API mostly right. But then the other approach, I thought was really novel, is having a, basically an algorithm that knows that can identify those things and, and modify them, or mask them so that they aren't identifiable. But they still set the context for the model that you'd need to get an even more accurate response. So, to me that both of those are obviously useful in different times, they probably both apply. But I like the word I like the use of the word Trust, I think that's important in this and I guess the vendors should be able to help with the, you know, customers, their customers with that trust layer and trust level as they as they implement, right?
Jeff Nicholson
Without a doubt. And we really saw some early thinking around the subject coming out of a Pega systems and they put in place T switch, they're switched up for trust and transparency on their next best action technology, which is which is AI, right, the right thing for you. But it's using a plethora of analytic models, and absolutely real time. And those models rely on data. And cases, even regulations will mean that you might have had to explain meaningful insight into the logic involved in that decision. But if you're using I'll pick model that really is pulling from many different types of data and using different ways that you can understand, you're not willing to be in violation with that you're introducing risk into it. So being able to leverage and turn off and on certain models for different use cases and have that trust and empathy built in the system. I credit Pegasystems for really starting that way. And now everyone seems to be talking about which is the way I think it needs to be.
Michael Fauscette
Yeah, I mean, that, to me, that does seem like the logical approach to give enough control in the system that you can, you know, that you can, can meet your own compliance standards, but also regulatory standards. And as we know, privacy regulations are mishmash across the world. I mean, GDPR, and you've got, you know, California has their own lot of states have their own. We don't have a national one yet. But I know in the in the executive order that Biden, the Biden administration put out on AI, that was one of the tenants there is they wanted Congress to take that up and pass a national, you know, legislation for privacy across the country to make that consistent. So that's, I'm sure that's a confusing one, and you need to control in the systems to be able to get there, right.
Jeff Nicholson
But without doubt, and I think it's a wonderful thing that we're considering having a uniform set of legislation across, you're still gonna have differences between states, but they won't be as wildly different now. And you can actually have a system that can better govern that across your processes right now. It's very complex, very hard for organizations that are serious about it.
Michael Fauscette
Yeah, it's hard to keep up with me just the changes and how things go. So as we talk about this, I mean, this isn't an investment in time and it's an investment in money. Certainly. What metrics should businesses be looking at? How do they measure the ROI of these investments? And, you know, look at how it's improving CX, of course, but how do you make sure you're getting what you wanted out of it?
Jeff Nicholson
Well, some of the traditional measures will still apply, you're going to look at, are they clicking? Are they engaging with content that you're producing? Is it converting to a sale? You might even look at the amount of time it takes to convert to a sale? And are you able to service that customer better. So, getting to a first contact resolution faster and less lower, ah, to your average handle time for these sales. But in my mind, I take a step back and say, you need to be looking at both how we started the conversation around not just efficiency, but effectiveness, not just the amount of time that it takes to do something and being more productive and efficient and new processes which you need to do. But are you more effective? Are you changing? And if you work from the outcomes backward and focus on that, I generally always believe you're thinking of the right thing.
Michael Fauscette
That's good. I think a lot of companies today seem to like they really focused around productivity. And that's important. I think there are a lot of productivity gains here. But your point of Yes, but look at the outcomes and make sure that the outcomes are what you expected. You can do something really fast in the wrong direction.
Jeff Nicholson
And it's very easy to automate the wrong thing at scale. Yeah, yesYeah,
Michael Fauscette
Yeah, that's exactly right. I mean, I sort of felt that way around sales technology for the past few years that a lot of the sales tech was built to make you do the thing that already wouldn't work, and only more of it. Not sure if that's a good strategy or not. Right. Well, Jeff, thanks for joining again, this. That's all the time we have today. I mean, we I could definitely keep going on this topic for a long time. It's, it's always an interesting one. But before I let you go, the one question I always ask everybody at the end, could you recommend somebody you know, a thought leader, author, mentor, somebody that's influenced you and you believe that would be beneficial for companies that are thinking about getting involved with AI and generative AI and their customer journey that they could look at and learn fromAbsolutely, there's
Jeff Nicholson
Absolutely, there's a lot of there. Frankly, a lot of great thinkers out there, the one I'm probably the greatest fan of is, is Andrew Frank at Gartner. He's been in this space for a long time, and I've long followed him. And he's the one that is thinking about things such as that brand aspect of where we can apply these types of technologies, and things that we can never possibly do before much less at scale. And I think he's just really has a just an incredible way of thinking about the opportunities and challenges. And he's also featured, I believe in a great podcast, which is Gartner hashtags. And if you're not following that, I suggest to check it out a number of great thinkers, and they're penny less, because they're Ben bloom others. And it's always it always gets you thinking in a different way, which I think is rare.
Michael Fauscette
Yeah, do I think that's great. And that's certainly in the world we're in, it changes so quickly. It's good to have some source that can kind of help you keep up and keep that in the right context, too. So that's great. Thanks. I really appreciate that. So, Jeff, thanks for joining today's great conversation. I really appreciate it.
Jeff Nicholson
My pleasure, man. I was great. Thanks.
Michael Fauscette
And that's the show for this week. Thank you all for joining remember to hit that subscribe button. And for more on AI and other software, research reports and posts, check out the area on research.com/blog and slash research reports. And don't forget to join us next week. I'm Michael Fauscette. And this is the disambiguation podcast.
