Customer Support Leaders
Customer Support Leaders
273: Balancing Automation and Human Touch in Customer Support; with Cory Brown & Elliot Fox
Discover how to master the delicate art of balancing automation with human interaction in customer support, as Corey Brown and Elliot Fox share their expert insights. Facing the pressure to adopt AI while keeping a personal touch can be challenging for businesses. This episode uncovers the potential pitfalls of over-relying on AI, such as generic responses, and emphasizes the importance of human intervention to ensure customer satisfaction. Listen as we explore how AI can serve as a co-pilot, boosting efficiency and enhancing customer experiences without sacrificing the essential human connection.
Join us as we dive into the practical application of AI in customer service, especially for high-volume, low-complexity tasks and round-the-clock availability. Corey and Elliot discuss the challenges AI faces in nuanced environments and highlight the benefits of data-driven insights like sentiment analysis. By examining both the strengths and limitations of AI, this episode offers a thoughtful approach to integrating technology in a way that elevates both customer satisfaction and team morale. Tune in to learn strategies that create win-win scenarios in the ever-evolving landscape of customer support.
Hello and welcome to episode 273 of the Customer Support Leaders podcast. I'm Charlotte Ward Today. Welcome Corey Brown and Elliot Fox talking about finding the right balance between automations and humans. I'd like to welcome to the podcast today Corey Brown and Elliot Fox a rare two guest episode. Nice to meet you both. Thanks for joining me. Nice to meet you. Thank you for having us. Should we do a quick introduction and then we can dive into what we're going to talk about?
Cory Brown:Yeah, sure, my name is Corey. I'm the co-founder and CEO at SimpleSat, and SimpleSat is a customer feedback management platform which lets businesses collect and understand customer feedback.
Elliot Fox:And I'm Elliot, I'm the head of customer success.
Charlotte Ward:Thank you so much. So we're talking about a topic that's a hot topic of the last year or so, I feel right. We're talking about the use of automation, ai and all that technology and specifically, about, like, that balance between all of these bots and all the other technologies that, uh, that um, like I'm making significant inroads right into the world of support and success, but, um, it's about hitting the balance right, and I think that's what we want to talk about today. Isn't it where that balance is, how we find it, um, what, what that experience looks like, and so on. So so where do we begin? I mean, I think let's talk about the proliferation of of options, maybe, first, and the proliferation of technologies and, uh, the different ways they're impacting the business.
Cory Brown:Maybe that's a good place to start yeah, and there's almost like a like a FOMO that consumers are feeling at this point because all of, like you said, the proliferation of all these technologies and we're getting bombarded by all the help desk and CRM tools that we're using, all these AI updates AI is coming soon. So I think everyone kind of feels like we should be doing it, but maybe not slowly taking a step, taking a step back and thinking you know what is truly right for our business and our customers, and that sort of thing.
Charlotte Ward:Yeah, yeah, that FOMO is very real and I feel, like I feel whenever I go to a conference, that it's more than fear.
Charlotte Ward:It's almost like a terror, that like, if you miss out, you're not just gonna have this kind of sense of of kind of uh, concern that you're falling behind, but like that I think, more than any other evolution I've seen, particularly in this business, the rate of of tools coming to the market range, the massively increasing range of ways in which they can help you as a support or success person that acceleration, I think, is heightening. That fear, isn't it, and it really feels like do this or you're going to be so far behind the game that you're going to be out of business or something right. I think it's a real terror, it's a tomo.
Cory Brown:Yeah, it's that. But it's a kind of double whammy where maybe managers and business owners are thinking about you know, we got to adopt all this, or else we're going to go out of business or competitors will surpass us. But there's a lot of support people who are thinking I might be out of a job, so it's scary for them as well as well.
Charlotte Ward:So yeah, yeah, um so. So how do we? How do we um reduce the fear on both sides? That's a big question isn't it if you've got the answer to that, I feel like you'll be making a lot of money tomorrow. Yeah, but uh elliot's.
Cory Brown:Elliot's actually like elliot's, a funny guy. He's super technical but um also really likes the man, the personal kind of manual touch. So I don't know, maybe elliot you can explain.
Elliot Fox:Yeah, I'm kind of I'm kind of against ai and support processes. I feel like you know this push that companies are doing with ai, you know especially it's very topical now. Um. So, like you guys were saying, you see it, you see it everywhere, it's very prevalent, but a lot of companies maybe are pushing for it too early or try to try and replace people too too much. Um. So I think you know it is a process that that can scale. But if you're scaling at the expense of your customers like that's what you want to avoid doing you don't want to just push more work onto your customers and I mean that's the thing that annoys everyone with. Like you know, you call a hospital or something and then you get in a phone tree and you're there pushing buttons and listening to recordings. Nobody likes to do that, right, but you get a hold of a person, you can ask your inquiry directly, you can have a human conversation.
Charlotte Ward:Um, I think it's important not to not to lose out on the personal humanization aspects of the business I, I really agree yeah yeah, I really agree and I I think that's an interesting point you make, because I think in many ways, in that rush to get on the ai bandwagon, we're almost forgetting that is just another technology and and we've had technologies to replace a lot of these processes or to co-pilot a lot of these processes in a different way for a long time now. And the one that you touch on there, the phone tree, you know, the something that recognizes what tone you just press, never mind your voice has been with us for a very long time and it's always probably for 30 years now has been super frustrating for all of the consumers of those phone trees. And yet here we are, we all live with them because they work for companies, not customers, right? But the idea that we think about AI as something different than another technology we could be considering, it's kind of weird to me that we treat it as a special child. It is vastly superior and you know its ability to contribute is much less confined to some of those bots and some of those.
Charlotte Ward:You know phone technologies and things that are really just one path and that's all you do. So AI opens the playing field, but it doesn't it. That doesn't. That doesn't really help us if we only think about that. I think we have to think about it as just another technology. And then and then the question always is whatever technology you're buying in, is it the right thing to do for our customers? Particularly, as you say, and I think that's the key Is it the right thing for customers? Because I know phone trees continue to piss me off to this day.
Elliot Fox:I think it's an important balance High effort, high quality implementations and AI might be.
Elliot Fox:You know, just as critically thinking, or appearing to be critically thinking, as a support agent could be, with a considered response, but low effort implementations where it's just plug and play with your help center, it's just regurgitating some answer that may or may not fit the context. Regurgitating some answer that may or may not fit the context, it's like you know, the co-pilot uh aspect that you mentioned. Uh, it lets agents have kind of oversight of that process so you can ensure that you know you're not just giving a convenient response, um, an immediate, convenient response to a customer's inquiry yeah, yeah, the co-pilot is an interesting and, I think, really valuable proposition and probably the best application I've seen.
Charlotte Ward:I do recall, like it was about a year and a half ago, I was on a panel at Support Driven and I was asked on that panel what was the most overrated technology or the most overrated tool in customer support. At that time this was a year and a half ago, it was not a decade ago and I said AI as the most overrated thing, and you know it divided the room, that's all. I'm going to say that same. In that same conference, I was wandering around the floor um later that day and somebody approached me and said that was really interesting. What you had to say, um, and the point I made to them was you know, I don't care what your tool is. Stop trying to sell me on the basis that it's powered by AI. I don't care if it's powered by a box of gerbils. That was my point.
Charlotte Ward:It could be powered by a box of gerbils, as long as it helps my team and helps my customers, and I think for me that's where Copilot sits. It's helping the team and the customers at the same time, and actually I think you know it could be anything. At that point, it doesn't matter that it's AI. It could be a well-structured, well-formed, accessible knowledge base. You know, I think I think it's doing the right thing, rather than the fact that it's AI and I like, I agree. I think that a good implementation can help. I'm not averse to finding that I'm talking to a bot in a chat you know, conversation with my energy provider or something If it's quick and it gets me the right answers as a customer, because there are situations where I'm happy for it to be depersonalized. But there are many, many situations where I'm not happy as a customer for it to be depersonalized, but there are many, many situations where I'm not happy as a customer for it to be depersonalized, right.
Cory Brown:Right, yeah, it's kind of like the high volume, low complexity cases are the ones that you don't care. It's personalized and it's just there's like a customer satisfaction metric, ces, the customer effort score You're going to get if you just it was very low effort to reset your password or to get the status of your order or something like that. But if it's maybe something where the company has a 30-day refund policy and you're asking for an exception, that's when you could get really upset Because let's say you have a really good reason but you're talking to an AI that's not going to give you the exception. So kind of reserve the more complex or sensitive issues for a human.
Charlotte Ward:Yeah, I think that's right and I think, um, I kind of look at it as, uh, you know, the human should be the gatekeeper of what goes out the door, you know, and I think that that is naturally going to change the agent experience, isn't it? And maybe, elliot, you can talk a little bit about, um, the agent experience here because obviously your agent experience is is the shape of their jobs is going to change if you do something like this? But also, um, managing that against how much of their role just becomes gatekeeping, those answers, you know, their fact checking and how much is actually value add over what the AI might be doing internally.
Elliot Fox:Well, part of what worries me with engaging these co-pilot tools like an agent engaging with it is it's not a knowledge replacement tool. It gives you an answer and it can seem very confident, but it could just as easily be the wrong answer. You still have to know whether it's right or wrong to be able to validate. And so it is a gatekeeping, because if you just take that AI response and you give it to the customer, you're also taking that AI's confidence, you're putting in your own confidence and it's still confidently wrong. And with like technical topics, like if you're talking about an API or something, no matter what AI tool you use, it's just like they're not gonna know how every API or technical matter, the specifics of a very like niche or small tool, how it functions, but they will answer you as if they know everything about it, and so you really do have to know the specifics.
Charlotte Ward:Yeah, that's interesting.
Charlotte Ward:That's an interesting point actually, because you are checking the accuracy In some ways.
Charlotte Ward:I think that one view I've always had of these kind of co-pilots is they're serving you the answer, and then what you're doing is kind of sprinkling in a bit of humanness, a bit of humanity into into the answer that before you push it out, you know, um, I mean, we've who hasn't asked chat GPT to do something like here's a technical answer, like from our engineering team, or something. You just make this a nicely formed customer response and, as well as seeming confident, I think they seem a bit generic, and so you know the kind of notion that you might put a bit of humanity in it as well as as well as fact check it. I think. Um, I think is where my head went of, like, actually, are we just are we devaluing the expertise of the human, because that's what the AI is potentially replacing and we're just asking humans to to humanify these responses? But you make a valid point. They are. They are a bit overconfident as as contributors, aren't they Copilots sometimes are a bit overconfident as contributors, aren't they co-pilots sometimes?
Elliot Fox:Yeah, and even with the genericness of the response, because usually the way that you interact with customers is you do tone matching right. If somebody talks very short and succinctly, you match their tone. If they're more verbose and they have more of a casual chat, you try to match that energy level and you can dive into a little personal details a bit, but ai's responses are, you know, generally generic it's very true.
Charlotte Ward:It's very true. Yeah, yeah, yeah, absolutely, absolutely. Um, so we've covered kind of some of the things we should be thinking about. Then, uh, in terms of whether's appropriate, whether it's something we we should even be even fearing, you know, experiencing that FOMO about, there are I think, as we touched on Corey at the top of the, at the top of the conversation organizations out there. There are support leaders out there who are feeling a bit like they should be doing something, but they're probably possibly and I would say maybe quite widely under pressure from the wider business as well to do more with less, to take advantage of this wave of AI, from being the support leader in that situation where there are other folks in your org, your peers or your boss or your CEO who are saying we're all in on AI and we need to get some tools in the door before we get left behind which is what we were talking about at the top how do you, as a support leader, think about making the right choices? How do you present that to the business making?
Cory Brown:the right choices. How do you present that to the business? Well, I think I guess, first and foremost, you need to jump into AI and just start using it for yourself. And you know, like I heard something someone said you're not going to lose your jobs to AI. You're going to lose your jobs to people who know how to use AI. That's very true. So become that person. So you know.
Cory Brown:Of course, it's not okay. You could wipe out. You could just fire your whole support team and do chatbots, but we already talked about how that could be bad for customers. But think about how you could be even a better support hero with um, with ai on your side. You could be resolving more tickets. You could be doing it faster. You could be. You know getting information, getting context information faster as a team, you could be surfacing. You know doing qa and stuff faster. So I guess, just trying to see how do we make this actually be better for our team, improve internal morale and internal satisfaction and customer satisfaction, like there's. There's got to be at least I believe, and at least you should start out believing this that there's got to be. At least I believe, and at least you should start out believing this that there's got to be a balance or or a um, not a balance, a win-win. Where you're, you're actually doing more, like you just said, doing more with less. It's not like the phone tree, where that was a zero.
Charlotte Ward:That's a zero-sum game for the company or the customer so, yeah, yeah, it is about finding that balance, isn't it, um? So are there any? Are there? If we're looking for win-wins, are there any easy wins here? What should we do first?
Cory Brown:so let's see here. So one we kind of talked about already was the um, the high volume, low complexity. So I think that's an easy win where you know if again with like the password reset or something, the customer doesn't care if they're talking to a, to a robot, they just want to get that done. Um, maybe like for a 24, seven availability if you don't offer that already. So you know off hours or over the weekends. So like customers most customers probably don't expect live support on the weekend, but you also don't like hearing that. You know you need to call back on Monday. You know if you just want to get it done on a Saturday or Sunday. So that could be a thing of just closing the gap. And also, I guess, just like data-driven insights as well. Or you know topic detection or sentiment analysis, or you know kind of automated qa, something where you're taking a whole bunch of data and then categorizing that in some way to help you improve or triage requests or something like that. I don't know, elliot is there anything else you'd add?
Elliot Fox:yeah, that's the one that um stands out to me because uh, just doing like that timely don't know, elliot, is there anything else you'd add? Yeah, that's the one that um stands out to me because uh, just doing like that timely qa process where you're going back and you're reviewing all the support conversations, it's ai could help you go through everything your company has and flagging things for you know where you could look into it deeper. So it's not really replacing something, but it's drawing your attention to what you should be focusing on faster.
Charlotte Ward:I find that a really interesting one because I think also, volume here is a key right.
Charlotte Ward:I think that I've tried sentiment analysis and my current environment doesn't have the volume either for surfacing knowledge or for sentiment analysis, or for generating answers on the front line.
Charlotte Ward:We just don't have the volume.
Charlotte Ward:It's low volume, high complexity and it's just the worst environment to apply AI into, even in the back end, even in terms of sentiment analysis.
Charlotte Ward:And that's because I think AI still is not clever enough to spot the difference between a customer saying I've got this problem and here's all the technical details and is there a chance you could help me by tomorrow? All the AI sees is super long response and they use the word problem and it's now surfaced as like a potential quality issue without catching the fact that we've responded really quickly. We've given the customer a one line, you know, code change or something, and the customers come back and said thanks, and that is just the typical tone of an engineering conversation, poorly on the sentiment charts because of the, the, the boundaries of what that ai understands to be a real human conversation. I think. I think that for me, the, the challenge is that, um, it's just not clever enough, frankly, I'm going to put it down to, and boil it down to one sentence. It's just not clever enough to understand that that's actually a good interaction in many ways.
Elliot Fox:um, yeah I know cory must have thoughts on on that, because we've actually been delving into sentiment analysis and kind of a topic attribution to feedback recently and it's not so. We couldn't just implement sentiment like a sentiment analysis tool. We had to, you know, go a bit further, right.
Cory Brown:Well, yeah, so we're building this. We're building a tool that, um, it's it's mainly about topics. So it's it's reading the, reading the answer and then applying a topic to it, but also a sentiment associated with each topic. Um, and it was just pretty cool actually, in in some responses there's negative and positive topics. You know, like, I love your product but your response time is too low or something like that. So it's, you know, product quality positive, response time negative. But I am noticing too that it's only going to work for high volume customers. And because it's, you know, it's subjective. If a human or a robot's labeling it, it's still, you know, you could argue with what they did. And just if you're only getting a few responses a day or survey responses a day, you could almost just be doing that manually as well. So, just like you're saying, if this stuff only works, even if it's accurate and not clever, it's only going to work on high-volume accounts.
Charlotte Ward:yeah, Do you see that changing at some point in the future? Well, probably.
Cory Brown:I mean, it's only been a few years that this whole thing just kicked off, so it'll get better and better. Yeah, I guess better.
Charlotte Ward:Yeah, I guess I mean I I can't wait for that day where, you know I can, I can talk to an ai that is secure enough that I don't have to worry about my, my organization's data getting outside of the boundaries of where we want our data to get, but that the ai has a wider experience than our data, because, you know, when I, when I review one of our conversations, I have a way wider experience than our data Because, you know, when I review one of our conversations, I have a way wider experience than just my last four years in my current role. You know, and I'm applying all of that experience and I think if there's a way that an AI can build that experience outside of my own data set, that would be amazing. I think that would be powerful in ways we can't imagine yet, and that is probably a world in which we want to, you know, be one of the people who are who know stuff about AI rather than.
Cory Brown:That's a great point. I've thought about that as the same thing as well, where right now you're just used to chatting with a bot that knows the help center, right, but like GPT, right now you could chat with it about anything. So it's like it's kind of getting close where it would know the help center, but you could also say give me a you know fettuccine Alfredo recipe, and it's like okay, I can do that too, you know, just knowing everything, your, your business, but also competitors, um, and just more nuance of how to how to have conversations. Like elliot was saying, tone matching, it's, it's definitely getting there.
Charlotte Ward:Yeah, it's a scary world actually, in many ways as well. I think when we get there it feels like we're now getting a bit 1984 or something, but but yeah, uh, but, uh, you know a world where actually you can't tell the difference necessarily as a customer.
Cory Brown:Oh, we're definitely going there.
Charlotte Ward:Yeah, yeah, yeah, yeah, yeah. So final question from me, then, let's put the future to one side for a minute, because it's too scary. What is, what is some practical advice that our support leaders and success leaders listening to this today can take away if they have done nothing yet about AI in their company? What would be a good first step to engage, to think about, to look at, to go and explore?
Cory Brown:I mean I'll start with a few ideas, like a hybrid approach could be good, where you know, don't just think that you know, is it time to replace our agents with AI? You know, it's probably never going to be that. So how do we always think? How do we work with AI? Not just having AI replacing something, and just you don't have to do this overnight. You can try new things, you can try a little bit, you know, with a few agents or with one group and just kind of small, continuous improvements and doing what's right for your organization, just like you're saying, maybe your ticket volume that you're dealing with is too low for like this mass AI stuff, you know. So every organization and support team is going to be different.
Cory Brown:Yeah yeah, that makes sense.
Charlotte Ward:Elliot, have you got a thought?
Elliot Fox:Yeah, I guess it would be to engage. It depends, like, what tools your organization is using, like if they have AI features out of the box that you can engage with or try out, or like, if not, something you could like easily bolt on with low effort or go outside and ask questions to and, you know, figure out how those tools work and how you can use them for some basic like content generation first. So, like, engage with the tools, try to see how they're useful and don't like use it as a co-pilot, I would say first. So make sure that you understand, like where its failings are, like where it can break down, where it's inaccurate, where you need to actually like, have the knowledge Cause. Then you're going to know whether you can trust it or not, and that's the most important thing to me, like, if you can't trust an AI to give the right response, then it doesn't really have a place and you know, within a customer facing interaction, you have to ensure that you're giving your customers accurate information, otherwise it damages your company's reputation that's very true.
Charlotte Ward:That's very true. Thinking about where you place it in the business is really important. Um, and I do like the idea as well of going for the easy approach first, is there something that you can just pick up? That's a feature maybe. Maybe you can get it on trial or something for a couple of months. Just, you know, something that's really easy to slip into your current processes without having to rework everything and then rework it again if you don't like it.
Charlotte Ward:I think I would add one final consideration, which is, as you look around you, what don't you want your people to be doing? Like, just ask yourself that. What's a waste of your folks' time right now? And when we think of AI, we most naturally because it's the thing that's in the media most we most naturally think of chat, gpt, we think of content production, we think of answering customers, we think of LLMs, we think of natural language. But AI can be applied in so many other ways, and I think there's the sentiment analysis, which is an interesting one. Doesn't work for me, but you know it's an interesting one.
Charlotte Ward:Automatic quality is another one. These are behind the customer scenes as well, so I think they're safe territory to experiment with you can pick them up and put them down really easily, right, and even beyond the language stuff you know. I think that back to my point of like, what don't you want your team to be doing, or what's taking a lot of time right now? There are other applications of AI that aren't language-based, and for thinking about my environment, one thing we're considering is how can we do things like automated anomaly detection, because we run a managed service, and things like that, and there are some really interesting applications of AI once you get into data and once you get into services and processes as well, and I think for me, that's the big thing what don't you want your team to be doing? What's a waste of their talents right now? Today?
Elliot Fox:totally agree yeah I think, another easy application is maybe with like help centers, like if you're, you know, wasting time searching a help center and ai, could you know digest all that information? Potentially, you know, give you an answer and tell you where the article is, especially you as they get bigger and bigger and things become harder to navigate. But it's a very easy to implement kind of direct application that you could start with.
Charlotte Ward:Yeah, yeah Again, it's almost almost Virgil on Copilot, but like just being that, that easy access search, because no search tool is ever perfect and yeah, yeah, yeah, absolutely. Thank you so much for joining me today, both of you. It's been super interesting conversation. We come back either together or separately. Um, I think it would be great to have another conversation there's I I know that, uh, there's a lot more we can explore on this and other topics. Would you please come back and join me another time?
Cory Brown:we would love to awesome.
Charlotte Ward:Thank you so much for joining me.
Cory Brown:Yep. Thank you as well, Charlotte.
Charlotte Ward:That's it for today. Go to customersupportleaderscom. Forward. Slash 273 for the show notes and I'll see you next time.