Why are companies preparing for agent-to-agent customer service when many customers still cannot get a chatbot to answer a straightforward question?
In this episode of Tech Talks Daily, I speak with Latané Conant, Chief Marketing Officer at Parloa, about the state of customer experience and what businesses must repair before agentic AI becomes another barrier between customers and support.
Parloa's State of Agentic CX report assessed 10,000 enterprise websites, 4,000 chat interactions, and 100 phone trees. According to the company's findings, fewer than 10% of the tested chat conversations achieved the customer's goal. Only 1% of enterprises demonstrated readiness for automated agent-to-agent interactions.
Those results raise a difficult question about years of customer experience investment. Businesses now have websites, chatbots, mobile applications, email, messaging, and phone systems, but customers frequently struggle to find help or complete the task that brought them there.
Latané argues that part of the problem comes from treating customer service primarily as a cost center. When the objective is reducing contact volume, organizations can unintentionally make themselves harder to reach. This overlooks the commercial and operational information contained within customer conversations.
Calls can reveal onboarding problems, unexpected product uses, recurring faults, and potential sales opportunities. Latané explains how analyzing service conversations can give marketing, product, operations, and executive teams a clearer picture of what customers are experiencing.
We also examine why so many chatbots reproduce the frustration of traditional phone trees. Although the interface looks conversational, the system underneath may still rely on rigid categories and predefined routes. Customers then find themselves trying different words or repeatedly requesting a human agent.
Latané describes a better agentic customer experience as being closer to talking with someone who already knows you. A personal AI agent could remember previous interactions, understand preferences, work across voice and text, and complete a request without making the customer repeat information.
That possibility also introduces questions about trust, permissions, personal information, and oversight. Latané discusses the need to monitor what AI agents are doing, identify when conversations move away from approved subjects, and use supporting agents to detect potentially harmful behavior.
Human involvement remains particularly important when a conversation involves distress, vulnerability, or emotional care. In Latané's roadside assistance example, AI can arrange a tow truck for a flat tire. If it detects signs that the caller is in distress, the conversation should move quickly to a person.
We finish by considering what this means for customer service employees. Latané believes experienced representatives and operations teams can become builders and managers of AI agents, applying their customer knowledge across a much larger digital workforce.
If the existing customer service front door is confusing and unwelcoming, should businesses repair that experience before inviting AI agents through it? Listen to the episode and share your thoughts with me.
Useful Links
[00:00:00] AI agents are only as strong as the data that they're given. When provided with an outdated data set, your agents could end up doing more harm than good. But not with Denodo. With an AI data layer built within your platform, your agents are provided with real-time data changes. So with Denodo, your agents can finally make the right business decisions. Simply visit denodo.com to learn more.
[00:00:27] Why are companies racing to build AI agents when customers still can't find a phone number, escape a chatbot, or explain a problem that doesn't fit neatly behind button three? Well, today I'm joined by the Chief Marketing Officer at Parloa. And together we're going to examine what the customer experience looks like before businesses ask AI to reinvent it.
[00:00:57] Only then can we truly understand the opportunities that are available. Now, Parloa's State of Agentic CX report assessed 10,000 enterprise websites, 4,000 chat interactions, and 100 phone trees. And guess what? They found that fewer than 10% of the tested chat conversations
[00:01:20] achieved a customer's goal. While 1% of enterprises demonstrated readiness for agent-to-agent interactions. And I'm going out on a limb here, but I think maybe that is not a surprise to many of you listening. We've all purchased something from a website, whether it is a product, a service, or a trip. But everything is great. Some big promises on there. Until you need to get hold of someone,
[00:01:46] or speak to an agent or support, suddenly things go dark very quickly. And this is why my guest is going to explain today why customer service has been treated as a cost-to-contain, and that's led to bad service. But the good news is conversations can reveal valuable customer insight. And why replacing an old phone tree with a shiny chatbot often reproduces the same frustration
[00:02:13] in a new different window. But the good news is, we'll discuss personal agents, human escalation, workforce preparation, and what CX leaders should be fixing first. So none of us have to experience episodes like this again. And with that scene officially set, it's time for me to beam your ears all the way to Amsterdam today, where my guest is waiting to join us.
[00:02:38] So thank you for joining me on the podcast today. Can you tell everyone listening a little about who you are and what you do? Sure, Neil. Thanks for having me. So my name is Latney Conant. I am the Chief Marketing Officer at Parloa. I'm a four-time Chief Marketing Officer and really excited to be doing it at Parloa because Parloa is fixing a part of the customer journey that I think desperately needs to be fixed, which is the
[00:03:08] service component. That's where we're starting with contextually aware, frictionless voice AI experiences. So it's a pretty exciting place to be. Awesome. And for people that are hearing about Parloa for the very first time, how would you describe it? And what would you say makes it different from all those other solutions out there? Yeah. So it's AI for customer experience. Yeah.
[00:03:34] And the things that make it really different is we started in voice and voice is the hardest modality to make voices natural sounding, to make sure that there's things like barging. So we've done a lot of work on our infrastructure around that. Another thing that's really important is like we're building agent building capabilities for the business. So those of you listening that are in CX,
[00:03:58] you know, with Parloa, you can build and manage and run your own agents. And I think this is really important because you wouldn't take your staff, your human staff, and give them to IT. That would be ridiculous. And like when we work with a really large travel company, and if you think about it, let's say a storm hits the Northeast, you don't want to file a JIRA ticket. And so we've really designed our product to be able to be highly reliable, but a business user can really own their,
[00:04:28] their new AI workforce. So that's really exciting. And then the last thing that we just released is the ability to uncover your dark journey. And what does this mean? And this, this gets me really excited as a marketer because we've spent millions of dollars trying to understand the customer journey. We've bought CDPs, we've bought CRMs, we've, you know, we've done so many things. We sat in rooms
[00:04:56] with post-it notes to try to, what is this customer journey and how do we do better and how are we more personalized? But the whole time it's been dark. And what was missing was what customers were actually saying to us. And so what's crazy is now we can listen to every single call. So every time a customer calls customer support, we can start to understand, oh, why are they're using the product in a different
[00:05:24] way? Maybe that's a use case we should start marketing. Oh, this is an onboarding problem. Like we should go fix that. Oh, this product line has, you know, these issues. And so we listen to every single conversation and uncover that dark journey for executive teams. Because so much of that rich customer insights is in the service data. And so that's really exciting, I think.
[00:05:53] Yeah, it really is. And one of the reasons I invited you on here is I came across your research. And when it comes to listening, I mean, you looked at 10,000 enterprise websites, 4,000 chat interactions, and 100 phone trees. But I'm curious, looking at all that vast amount of CX data, what did you learn about the state of the customer experience today? And which findings surprised you the most? Because this is not your first rodeo. You've seen and heard so much throughout your career.
[00:06:22] Anything surprise you in there too? The whole thing was actually pretty darn surprising, I got to say. When I started at Parloa, you know, your first job is say, what's the current state? And so this study for folks listening was designed to go and mystery shop large enterprises to figure out what is their support experience today.
[00:06:43] And what we found, and my underlying assumption going in was that companies were actively trying to not talk to their customers. And I assumed that was going on, right? But I didn't know to the degree that it was. So only 43% of companies have a visible customer support phone number. And again,
[00:07:07] when I go back to that rich data, like company, like people want to talk to you. Think about my, my career has been trying to get companies to talk to my brand. And I have spent so much money. Meanwhile, there's all these people that actually want to talk to us, and we're sending them away. I mean, it, it, so that just blew my mind. And then the next step was, so we had a lot of great
[00:07:34] findings from the mystery shopping. But then I was like, okay, let's see our customers, our consumers cool with the experiences they're getting. And then we ran another study where we went out to all these consumers and said, how does this experience work for you? Is this working for you? And what the unlock there was 86% said that their service experience directly impacts loyalty. 86% said they cancel a
[00:08:02] subscription immediately after a bad service experience. And voice, so being able to call is preferred three times more than any other channel. So on the one hand, it's, oh no, this is bad. On the other hand, think about the unlock there is for companies that really start to focus on this. It's
[00:08:26] reduced churn, it's reduced brand reputation issues. The other thing we found from the study is that 33% tell friends and family about, about a bad experience and 29% post on social media. And then that loyalty number, I mean, 83%, like that's the whole podcast right there. Yeah. Yeah. It really is. I mean, that's that 43% of company websites failing to provide a clear
[00:08:54] route to customer support. And we've all kind of been there, you know, AI, you just want to speak to someone. You've got an AI bot, you've got an email from a do not reply email address. So that rules that out. Then you go onto the phone and it's confusing phone menus and lengthy hold times all the time. As you said, you just want to speak to people, but I've got to ask after years of investing in digital CX, why are businesses still making it so difficult for customers to get help? We've got
[00:09:23] more communication tools than ever before, right? Yeah. It's fundamentally been thought of as a cost center. Yeah. And it hasn't been thought of as the center for customer insights. It hasn't been thought of as the loyalty unlock. It hasn't been thought of as a new revenue stream. And some of my favorite case
[00:09:49] studies here at Parloa are ones where when we got in and we started planning their journey and thinking about the best use cases, we found that people were calling in with upsell opportunities, cross-sell opportunities. So some of our best case studies are not just, oh yeah, we, you know, we reduce the cost to support. It's actually, we created a whole new revenue stream. And this is over and over and over
[00:10:17] again. Like think about extending your support hours, better scheduling, like faster time to schedule. So I think that there's a real mind shift that needs to go, go on with what is the existence of this function and how it contributes to the revenue team and the revenue cycle versus this is, this is just an operating cost that we want to try to avoid. Yeah. And another big stat, and I think it was fewer
[00:10:43] than 10% of chat conversations you tested actually achieved the customer's goal. And again, myself, you and everyone listening, we've all encountered these kinds of problems. It feels like if you're not asking a generic question, like, can you reset my password or what are your opening hours, then it doesn't know what to do when you drift from that script. But why are so many chat experiences failing
[00:11:07] and what separates automation that genuinely resolves a problem, gives you what you need very quickly from automation that simply creates another barrier and leave you just furiously typing, speak to agent, speak to agent. So I think the fundamental issue is, you know, the IVR tree is not, as we would say, conversational, right? It's completely rules-based. And so what a brand needs to do is they need to guess
[00:11:35] why would someone call in and they need to try to interpret what they're saying and put it into a menu tree. And when most people call, and we find this over and over again, because one of the first things we do is analyze all the calls, and we find there's always like four or five different intent categorizations, which I think is just hard for an IVR. And we've all done this where you call and
[00:12:01] you're like, it's like, do you have a claims issue? Do you have a dispute? Do you have a technical mumbo-jumbo? And you're like, I don't really know. I just need to know if this is covered in my plan. Like, where would that be? Or I'm calling about an MRI. Well, you're calling about an MRI because you want to know if it's covered, because you want to know which ones are in the network, and none of that comes up on
[00:12:26] the menu tree. And so I think that's where this big disconnect and source of frustration comes. And so what's happened is chat is seemingly conversational because you're like chatting in, but it's based on an IVR tree. So it's just taking it and plotting it into that same rules-based flow,
[00:12:51] which is why it's having the exact same result as an IVR. And we all know the first thing you do, 70% of customers admitted to gaming the IVR pretty much immediately. Yes. Yeah. I think we're all guilty of doing that on occasion. But I suppose the good news here, though, is generative and agentic AI now promises a very different customer experience because the
[00:13:17] systems can understand our intent, reason, and take action. So what does a genuinely agentic customer interaction look like compared with the nightmarish scenario I mentioned a few moments ago with chatbots and automated phone systems that we've become accustomed to? We say it's as easy as talking to a friend. Yeah. And the reason that's so important is you know your friend's phone number, they pick it up,
[00:13:45] they pick up the phone when you call, they know about your past interactions, so they have memory and context. They're able to understand your preferences. And it's multimodal, right? They can text, they can do phone calls, they can email. And so that's really what the ultimate experience is. And what we're working with our brands and customers to do is to actually create a
[00:14:14] personal AI agent for every single one of their customers. So back to that travel example, hey, I'm, you know, I need to change my hotel. Oh, okay. Are you traveling by yourself? Are you traveling with your kids this time? You know, are you ending up back in Chicago? Or do you, are you flying to New York, which you go to a lot? Do we need to change it there, right? It's like talking to someone who already knows you, knows your patterns, which is just a completely different
[00:14:42] experience. And so that's really what we're working on with our customers to be able to provide. And it's, it's really, really exciting. I think it's exciting for the end consumers. Another thing that we found in this, the second study that we did is 55% of people admitted to crying, throwing their phone,
[00:15:07] extreme stress with some of these interactions. And that's not great for the person on the other end either. I mean, that's, that's really stinks as a customer support rep as well. So I think if we can, I think there's this opportunity to make it way, way better. Yeah, a hundred percent. And maybe the biggest wake up call in the report is a scary stat that only 1% of enterprises demonstrated
[00:15:32] readiness for this next generation of automated and agent to agent interactions that we're talking about here. So I've got to ask, what does ready actually mean? And what are the other 99% missing here? Yeah. So what's interesting about that is that stat is about me as a consumer. So rather than me calling to cancel my hotel, I have an AI agent that goes and cancels my hotel on my behalf.
[00:16:00] Yeah. And that's what we're seeing people are not ready for. The first step to get ready for that is to actually have a much more conversational, like AI conversational interface, because, you know, the AI is not going to be pressing one or pressing two or, you know, things like that. So, so that's like a really important first step. And then there's some other steps as well that need to be taken.
[00:16:24] And I don't think that's something people need to be ready for tomorrow or maybe in the next two months, but I think it's sooner than we think. And so, you know, I think it's probably six months to a year and really thinking about that. It's already happening on websites. And so when you think about your front end website experience, making sure it's friendly for an AI.
[00:16:49] Yeah. And I think a while ago, what we're talking about here would have felt like science fiction, but we're already approaching a world where customers AI agent could contact a company's AI agent and resolve something without either person picking up a phone or opening a website. So how does agent to agent commerce and service service, how does this change the way that companies need to think
[00:17:13] about things like identity, trust, permissions, and customer relationships? It feels like everything's changing so quickly. And this stuff is here much sooner than the many listening might actually imagine. Yeah. So, I mean, a key is observability. Yeah. So a really key criteria to a good agent management platform, which is what Parloa is, is observability. And so being able to see exactly what those agents are doing, what they're talking about,
[00:17:42] are they drifting? You know, are they trying to get someone off topic? Are they trying to get someone to give up PIR? Things like that. And then there's actually little, they're called subtask agents, but helper agents, if you're working with a good platform, there's helper agents that actually monitor the AI agent and the conversation to make sure some of the right things get done and prevent
[00:18:06] bad things from happening. So it's just a, it's a level of sophistication that is getting built into enterprise grade platforms. Yeah. And I think for many listening, there'll also be a very real concern that businesses might view AI as an opportunity to remove people and reduce service costs rather than improve outcomes. It's very easy to get distracted like that and end up with short-term gains and a lot of long-term
[00:18:34] problems, but where should humans remain part of the customer service and how should companies decide which interactions can be safely be handled to, handed over to AI? How do you get that balance right? So I think this is a really important question and there's a great book about the, called the effortless experience. And most people want their experience to just be easy, like for most of the
[00:18:58] reasons people call. But when it's not that effortless type of call, you know, for different intents, it's critical to be able to quickly get to a human for the most kind of caring and important things. So a good example for us is we do roadside assistance for a company. And if you have a flat tire, that's, you know, we can get you a tow truck, no big deal. The second any sign of like distress
[00:19:27] is sensed, that gets like escalated immediately because like something much more important could be happening. So it's like understanding those dynamics and like when to escalate to a human or not. We actually on the flip side have some conversations where people are more comfortable talking to an AI. So it's really just, I think, being, continuing to be very, very customer centric and thinking about like what's the right approach for the right situation.
[00:19:58] Wow, that's fascinating. I always try and give people listening a valuable takeaway. So if we do have a CX leader listening and they recognize that their organization and many of the things we've raised in these findings today. So what should they fix first? What practical changes should they be making today to prepare their data, their systems and customer journeys for an agentic CX model? It's not as simple as just buying something off the shelf. What should they be doing to prepare now before
[00:20:27] they get there? So I think the elephant in the room is like, what do I do about the team today? Yeah. What we see in a typical support organization is actually a 40% turnover year, you know, every year. So a lot of folks leave on their own. And so what you want to do is protect your best people, right? We really want to keep those amazing service reps and keep them motivated.
[00:20:52] And we also want to start to be getting them upskilled because there is a future world where each one of those amazing agents and the people in operations on the team actually become agent builders, right? So they go from having, you know, maybe they manage a team of eight human agents and all of a sudden they go to managing a team of a thousand agents. And so the scope becomes much, much bigger
[00:21:18] and it's actually like an up-leveled role. So I think there's a lot of things that as a leader, you can do to start to prepare your team. You know, we do these like agent builder workshops and competitions and things like that where anyone can go and learn like how to build an agent. So I think getting like AI literate and understanding is really, really important so that you can kind of get your team to the place that they need to be. And then when I think about the first use cases,
[00:21:47] success is about prioritizing the right use cases. And to me, the front door is what has to be replaced. The front door right now for most people is a scary, not inviting house. It's press one, it's press two, it's IVR technology that was built in the eighties or nineties. And that
[00:22:09] AI can absolutely crush and do so much better. And the business case is, you know, the IVR is cheap on the surface, but when you look at all the downstream problems that it causes and what it actually deflects versus what people just game to get to an agent, it's really not effective. And so that's what we always say is like, let's get that front door replaced. And then we'll think
[00:22:35] about what and how we can continue to work towards that personal AI agent for every one of your customers. Brilliant. I can hear light bulb moments going off around the world and everything that you've said here and some of those big stats. And for people listening that would like to find you or your team online, find out more about anything we talked about today, including that research. Where can we find out more about all things Parloa?
[00:23:01] Yeah. So they can just go to www.parloa.com. They can also find me on LinkedIn. I'm Latney Conant on LinkedIn. And so either one of those, I'd love to connect and get to know your audience. Well, I've just loved chatting with you today about breaking down some of the uncomfortable truths for the AI chatter and all the noise out there is that enterprises are not currently set up for CX
[00:23:26] success in the AI era, but there's a lot of opportunities to change that. So I will include links to everything that you mentioned, including the study that we've referenced today. And I urge people to reach out to you and the team there and find out more information. But thank you for raising this today, such an important topic and something we've all experienced. But thanks for sitting down with me. Amazing. It was great to be able to be on the podcast, Neil. Talk to you soon.
[00:23:53] Wow. So much practical advice here about simply beginning at the front door. If customers arrive and want to talk with you and find a maze of menu options, rigid scripts and technology that cannot understand why they called in the first place, obviously adding another AI layer into that same mythology will just simply make the maze even more expensive, complex and frustrating.
[00:24:20] But agentic customer service can become useful when it remembers context, understands your intent, completes ordinary requests and recognises when a person needs human care. And the roadside assistant example that she shared, I think captured the balance perfectly. Yes, AI can arrange help for a flat tyre, but any sign of further distress should trigger an
[00:24:48] instant human response. And I think her view of customer service teams becoming agent builders and managers rather than being pushed aside by automation is also important too. But over to you. From listening to the conversation today, what should your company be repairing first in your customer journey? And you know from your own experiences as a customer for other enterprises. I'm curious where
[00:25:15] you're going to start. Let me know. TechTalksNetwork.com. You can find out where you can meet me on the road at a tech event, work with me, or just simply browse through 4,000 interviews, whatever it is, we got you covered. But that's it for today. Thanks for listening. Bye for now.