From Cost Center to Value Creator: The CFO-Ready Playbook - UJET

CFOs are scrutinising every CX investment, and “better CSAT” rarely cuts through. In this CX Today interview, Rob Wilkinson sits down with Matt Clare, VP Product Marketing at UJET, to unpack how CX leaders can translate customer conversations into outcomes finance teams actually care about: lower churn, fewer interactions, reduced cost per contact, and higher customer lifetime value.

Matt explains why traditional VoC programmes often stay trapped in silos, how conversational analytics can consolidate signals from calls, tickets, CRM, and social into a usable root cause view, and why the best CX is often the interaction that never happens. They also explore what “agentic AI” changes in practice, including AI-generated taxonomies, natural-language querying for non-technical teams, and closed-loop automation that reduces swivel-chair work while keeping humans in control when risk is high.

Transcript

SPEAKER_02 0:58

So welcome. Thanks for joining me today.

SPEAKER_01 1:01

Welcome for our pleasure as always.

SPEAKER_02 1:04

So let's get let's get stuck in. What's the the background here? So there's a bit of a kind of what's going on right now. Why is it important? Uh especially when we look at this through a CFO's lens.

SPEAKER_01 1:16

Yeah, I think for me, the having spent the last 20 years in this industry, we've always talked about contact center, like changing the perception of contact center being the cost center and overcoming this. And I think we're at a point now where the tools actually enable it. And so there's a couple trends that we've been kind of monitoring in the space that are really putting an extra layer of scrutiny on, like, all right, your C levels aren't going to care about deflection rates and containment. Like, give me some analytics that actually matter.

SPEAKER_00 1:52

Um, and the first is I think a lot of companies have really struggled to find the ROI with AI when it comes to automation. Uh, the industry's push this narrative of human replacement, cost savings. And broadly speaking, when you look at contract center customer experience focused organizations, folks really kind of lay off staff. Um, in addition, they've spent a lot more money on AI than they thought they would because they they haven't properly accounted for it and it's gotten more expensive as things became more generative over the last few years. And so there's this added layer of scrutiny now on any CX leader that's trying to implement AI to really show that return on investment um and ultimately help orcs overcome this perception that contact centers at some cost and it's necessary evil. Um, so I think that's why the time is now to be having this conversation.

SPEAKER_02 2:48

Yeah, I I think you're absolutely right. I'd like to just kind of pivot slightly um because we did some recent coverage um around uh the spiral acquisition um that you did. And we talked in that around churn being quite a blind spot um in terms of data while we're talking about data. So um kind of in plain English for our our audience. Well, where are organizations feeling pressure most directly in that regard?

SPEAKER_01 3:16

I mean, I think the the problem with things like churn is you're not always going to get all of the answers out of contact center conversations alone. Right? There's product and service data that's being captured, there's CSAT and NPS data that's being captured in siloed tools, you've got all the contact center channels, of course, that are a source of conversations, you've got things like peer review sites, you've got people complaining on social media and Reddit and like these emerging kind of user forum uh type sites. And the the struggle historically has been someone churns. If a CX leader just looks at the CAS conversation and gets some analytics from that, they may get all right, customer edibility issue, uh very top-level topic model type um analytics. But the truth is buried in all those other sources. It's in what actually happened at the UX level when they were using your product or service. Uh, had they fill out surveys? Was this for customer frustration over years potentially of time? Um, so I think when we think about the blind spot specifically, it's that most contact center organizations are making decisions on a very small subset of their data. Um, if you think of a contact center supervisor and manager, they're looking at typically 5% of calls for the teams that they're managing, which means that you're going to prioritize your time on agents that you know are struggling and new hires. And so it's this very reactive model. Um, and so I think the tool chain itself has caused these blind spots to exist such that when a customer churns, executive skill Y, you get a little tidbit of information from the conversation itself, but you don't get the more holistic picture. And this is where kind of modern AI-based conversational analytics are really changing the game by collapsing all those data sources into a single source of truth that really provides voice of the customer at a level where you're not just getting that the customer had a billing issue, you're getting the customer had a billing issue, they had repeatedly been overcharged for X amount of time, and then you kind of get to the true root cause issue that allows an organization to go take action, ultimately understand things like churn or um etc. in in a deeper way than they've been able to historically.

SPEAKER_02 5:56

When you're kind of when you're looking at these organizations, because we spend a lot of time in lots of different organizations looking at all of this, exactly all this stuff. So what what triggers the moment, I guess, where the leadership team looks at it and goes, actually, well, there's a we need to do something now because um we're kind of we're struggling. Um it could be a turn event, it could be something to do with cost or even a whole conversation, I guess.

SPEAKER_01 6:20

Yeah, it's typically like the things if you think of the life of a C-level executive, the things that matter most are like business outcomes, right? Again, they don't necessarily care about handle time improvements because they can't map that to money. Um, whether that's revenue generation, whether that's operating costs. Um, so normally this is very reactive, kind of eureka moment where it's like we need to do something about this, where you're watching month over month downturn in terms of churn, you're watching revenue contribution, cross-sell upsell struggles, you're watching um maybe even things like employee churn, right? Contact center agent retention's always been a problem. And agents churn for reasons that are directly related to the interactions that they're handling quite quite often. So it's it's usually a very reactive moment where after a quarter of watching, you know, data points struggling, um, the executive class says, well, we need to do something about this. I think where people struggle is they go, you know, well, we've we've got that data. We've I am I'm a contact center leader, I've got all the conversational data, we have an analytics tool, and I can tell you if we have a billing problem or a product problem. But again, to my point earlier, without that kind of like consolidated truth from all those customer conversations and feedback sources, you're you're really only getting partial truth. And when you have partial truth, you're treating kind of a symptom rather than the actual problem at its source.

SPEAKER_02 8:04

How does that show up then uh for the customers and the frontline guys, you know, every day, day to day? Because all of this kind of has real consequences for the people who are just trying to get their jobs done, right?

SPEAKER_01 8:16

Yeah, I think for the frontline workers, the the problem is like if you think of the life of a contact center agent, like they're dialing for dollars, basically, right? In the traditional outbound sense, in the inbound sense, it's like let's resolve the interaction and get onto the interaction as quickly as possible. So uh one of the examples I often use is this like imagine that you had a problem with customers on your product because of a cloud region deployment in northeast Canada, and I'll use Canada because I'm based in Canada. Um contact center agents are based offshore, they're answering calls, they they know that there's a product problem, but they aren't correlating that that's a problem in eastern Canada. Right? Their job is to, all right, so you had a problem, here's how we're gonna resolve it, on to the next thing. And so the life of a contact center agent's job is really not to identify problems at their source. It's answer the inquiry as quickly as possible and get on to the next one. And even if you think of contact center supervisors and managers, their job isn't necessarily identifying problems at the source either, right? Their job is improving agent performance and doing things to drive CSAT, but more often than not, CSAT is something that, you know, there's a problem, there's a root cause issue there that needs to be treated for CSAT to improve. And so since UJET's acquired spiral, we've talked a lot about the value of this tool really being like avoiding unnecessary interactions. The best customer experience is the one that doesn't require you to reach out to me or me to reach out to you. And the what only way you can really get there, again, is by like having a tool that can help you identify problems at their source so you can treat them at the source. Uh so back to your early earlier question like, what does that mean to the life of the agent? It means that me as an agent, I'm gonna keep handling the same types of frustrated customer calls because the problem hasn't been fixed.

SPEAKER_00 10:28

Um, to the customers themselves, it's things like repeat callers, it's things like you know, very low first contact resolution. Um and so all of like by when you when you're treating the symptom and not the cause, you're just adding more friction to the customer life, which will increase increase customer effort and increase customer frustration, it'll lower CSAT, etc. On the agent side, it's basically doing the same thing. Handle times are going up, FCR is going down, agent frustration is gonna skyrise, which is a like the number one challenge with churn.

SPEAKER_01 11:06

There's contact center agents love being contact center agents. We often think of them as like entry-level jobs, but in reality, especially in enterprise, you find folks that have been answering phones and handling customer inquiries their entire career. Um, the reason people churn is because organizations don't fix problems at the source and don't give agents the tools they need to do their jobs effectively.

SPEAKER_02 11:32

Yeah, you've you've you've hit the nail on the head in that last sentence, I think. What what what you're seeing internally, uh, are teams kind of getting to the point where they're debating the data rather than going off and fixing things? Is that real is that a reality?

SPEAKER_01 11:46

I think the challenge procedurally, and like if you think of like there's two sides to this coin. There's the technology side that's a challenge. You need a tool that can bring you on this journey where you're actually identifying root cause and able to like build roadmaps to continually improve your product and service and operations. But then there's another side that's the people in the process. So it's it is organizationally overcoming this idea that voice of the customer is a thing that lives within contact center because it doesn't just live in contact center. Social media often lives with marketing, for example, right? Um, so you do need to fundamentally pivot your thinking, uh like in the customer first kind of world that we need to move to where you're breaking down the silos between these people and processes as well. In terms of debating the data, again, I think technology plays a role here. Um it's it's one thing for a contact center manager to hand a report to an engineering product leader and say, here's some problems with our website or our mobile app or a product, go fix the problem at the source.

SPEAKER_00 12:59

It's another thing to arm that product and engineering team with the tool and the analytics of themselves. So it's not just, hey, I'm receiving a report, I can go query the data myself and get the information myself. So one of the things we've actually seen with Spiral is we'll actually start a conversation with the contact center operationally, who's looking for an analytics solution to our voice of the customer. What we find on the back end is because the way the tool is under commercially uh licensed in a way that isn't C-based, one of in some of our accounts, the larger population of users of Spiral, our analytics platform, is the product of engineering team. Because they can just go query that data and get the data themselves. They don't even need to have that kind of handoff that traditionally existed between contract center, CX leads, and kind of product engineering. So I I think when yeah, just to kind of bring that all back together, it's not just the technology, it's the people and process, and then making the technology available to the people such that it's not just a like trust me, you have a problem, it's hey, you can learn it between yourself and figure out you have a problem. So much less debate in that world.

SPEAKER_02 14:19

Let's look at Spiral. Um what's different about what Spiral does um in terms of making it easier to get into all of that messy, scattered data that we currently have to contend with uh in terms to be able to then turn it into those great insights that that audience wants to get their hands on?

SPEAKER_01 14:39

Yeah, great question. And Spiral's a curious platform a little because it solves a lot of problems, like that customers will go, oh yeah, we have something to touch that. And it's like, well, do you really write? And so I think when you think about what makes Spiral different, it's it's a bunch of things. First and foremost, it is a hundred percent of your customer conversations and feedback sources. So it's designed not just to model and analyze your contact center transcripts and call recordings, etc., but it's pulling in MPS and CSAT surveys from third-party tools, it's pulling in social media from social media aggregators like Spread Social, it's pulling in your CRM ticketing and case management data, it's pulling in uh uh what am I missing here? Things like third-party peer reviews. So think like Google Play, Apple Play reviews, for example, we can ingest and model. And then probably most importantly, what really makes it different from like the data ingestion perspective is we can pull in customer-specific metadata. So that could, if you think of like a huge enterprise contact center with eight different BPOs, we can pull in all your BPO data and kind of show you exactly what's going on between the various BPOs very easily. We can pull in things like costs per contact or scheduling information, directly correlate, things like that. We have customers pulling in information about the release schedule. So if they pushed a product release last week, uh version you know 3.2 or whatever that version is, and they see a spike, they can directly correlate the version the customer's on, know it's a lease and release, identify the bug to the exact version, etc. Um, so it's able to really like the result of having all these different data sources is you're you're able to actually drill down deep enough to get to true root cause. And that's kind of the second thing. If you think of how AI analytics have worked the last couple of years, the last decade or so that conversational analytics has been around, it's been pretty high-level topic modeling. I can tell you you have a sales issue, a service issue, I can tell you what product it's about. But to my earlier example, like I couldn't actually get to the data that goes, hey, wait, all the customers that have been calling about this outage have been in Eastern Canada and are running on this version. So you something happened when you deployed your recent kind of SAS product in this specific region, go fix it. Um the other thing we do that's really interesting is we use AI and LLMs behind the scenes to automatically create a customer taxonomy. So with a lot of legacy tools, you'd manually program kind of labels and categories that you wanted to monitor. Spiral uses the power of AI and LLMs to do that all automatically, and the benefit of that is customers can get up and running in days as opposed to months, and you know, traditionally these were very expensive professional services projects.

SPEAKER_00 17:53

And then lastly, Spiral is not like we used to work at a contact center vendor a decade ago that would tell you we they had hundreds of reports, and it was like, great, I don't need hundreds of reports, I need one that gives me all the answers. Um and so the great thing about Spiral is it everything's created by AI, and it's as easy as asking a question in natural language. So if you think of like how to use Chachi Computer, Gemini, to ask questions, you can go into Spiral's AI agent type of question. CSAT's been dropping. Can you give me the top 10 reasons CSAT's been dropping and what I need to do about it? Spiral will actually give you analytics on this, it'll give you recommendations, so kind of a roadmap on things to go do to actually fix those problems at their source.

SPEAKER_02 18:42

What we need to then do is go, okay, what what so what? So from a CFO's perspective, going back to that, what what do they care about? So what does good look like to them? Because it's going to be different to what it looks like in the operation, and we should probably paint that picture so that leaders can understand how they need to communicate with their CFOs about the impact of these sorts of tools.

SPEAKER_01 19:04

Yeah, it's it's a great question, and it is a bit of a fundamental like mind shift, right? So in the traditional CX world, I I feel like I'm picking on the average handle times and deflection containment a little bit here, but those have been things that traditionally leaders have cared about a lot. Um, Spiral can absolutely help identify pain points and friction and help you build a roadmap on how to improve handle times, whether that's things like you know, agent coaching opportunities, kind of automated quality management use cases, things like that. Spiral can also do things like help customers understand what workflows are ripe for automation and which workflows are higher value that may require human touch. Um, so really helping customers understand like where and how to deploy AI and automation. Um But again, if you bring it all back to this kind of concept that the best interactions or the best customer experiences are the ones that don't require interactions, which you remember is kind of the core mantra for Spiral. Um it's it's actually things like interaction volume will go down, should go down because you're identifying problems proactively that need to be addressed such that the interaction doesn't need to happen. Uh, so you should see things like the volume of interactions go down, which of course has a direct correlation to human costs, um, because there's a cost per contact. Um so you should see cost per contact go down, you should see the volume of contacts go down, which will have a direct uh correlation. You'll see things like churn go down, um, which in parallel is directly correlated to cross-sell up, sell contribution, customer lifetime value. And so it is kind of your partner, your business partner, the tool itself is kind of partnering with you as a leader to actually change that conversation from all right, we're just gonna worry about average handle time and look at all these great coaching opportunities we've identified, and the humans are surrounded now, to all right, like money talks. And so when you get to things like churn and revenue contribution and customer lifetime value, that that's something that makes a CFO's eyes light up.

SPEAKER_02 21:30

Especially, I guess, when it is finally uh based on you know the report, not the the five five that we were something um I I'm aware of that we definitely haven't uh the fact that operation. Why why is that gap being so stubborn and and what should leaders consider measuring uh in order to prove that they can reduce the friction there?

SPEAKER_01 22:09

Yeah, it's a great question. I think the the gap is technology, right? Like for the last 10 years that AI has been in the contact center, everybody's been bolting on third-party AI for the most part to their existing stack. The same way that orgas have bolted on quality management, workforce management, etc. And every time you do this, you're introducing a new rub, right? Um APIs are limited, uh, integration possibilities, thereby limited in terms of what you're able to actually do. So, what we've announced with AXO is effectively we've taken Spiral and transformed it into a persistent AI layer that lives on top of CAS. So, to kind of for example, today's Spiral could ingest all your conversations and tell you what to go automate. But Spiral doesn't hook into a virtual agent to go automate. So with AXO, the Spiral Engine itself is basically ingesting all the conversations, building out the taxonomy, proposing a virtual agent builder workflows that they should go build and implement. So the human in the loop throughout the whole process. Um we're not doing anything magic, you know, and just we want to make sure that humans trust the process. So the humans have the ability to go in, tweak things, fine-tune things, apgar Rails, etc. Um, and that will deploy automation for the front end. Umxo also has a capability that allows um admins to basically specify how much or how little AI they want to use per flow, per customer segment, et cetera. So a new layer of granularity, and this addresses the part of the problem, which a lot of the AI vendors out there today will just be like, I need everything. And it's like, hold on, timeout, I don't believe that. Like there are conversations that will be higher value for humans to handle than bots in the future. Um, from there, AXO has virtual agent and AI virtual agents. So the ability for agents to take action, the ability for a virtual agents to actually tap into a supervisor to do like something that may be a little risky. So, use case you have a customer, they started a conversation with a chatbot on your website, and they want a discount for price matching purposes. You probably don't want your virtual agents just price matching willy-nilly. Um, so you tap in a human, the human approves that flow, the agent completes it. Um, and then of course, the AI component of AXO, also doing things like identifying when conversations need to get to humans quickly, uh, providing context once that transitions to a human agent. So not just doing next best action and coaching and uh kind of traditional AI agent assist tooling, but being able to like propagate across all your back office data points wherever that customer or business data may be stored. And instead of giving you like, hey, here's the last ticket, like you really just need that one line within that ticket, right? Why are we forcing agents to screen read through this fold to get the one piece of data they need? And then where AXO is really different is in kind of uh providing humans with the ability to execute agentic AI workflows that are not transparent. So the way we've thought about kind of workflow automation, historically, it's always been like it's just running behind the scenes. Um, AXO uses this kind of emerging technology called computer using agents or COOAs, that is kind of a it's an LLM-based evolution of RPA. And so if you think of robotic process automation and how this has worked for the last couple of years, it's like record a workflow on your screen, deploy an agent, and the agent will click. The challenge is if the UI changes or something's not exactly where it needs to be, RPA gets confused and can't execute a workflow. So these computer using agents basically see your screen and they think and they act just as a human would act. And the benefit here is we're actually able to break down this swivel chair that's existed within contact centers. So if you think of like use case, customer calls in, they got billed twice for a month, you need to file a claim, process a refund. Historically, this was swivel chair, swivel chair, swivel chair across different apps, and now it's click a button, file a claim, click a button, process a refund, voila, the AI agent will go do all the work, um, kind of report back. And then where it actually complete stuff on the back end, of course, the spiral engines there providing the analytics. Um, but it's this kind of continual feedback loop.

SPEAKER_02 27:00

To go back to the beginning of this conversation, it feels like we're on the kind of premise now of where all this is is happening. So super exciting. I can speak to you all day about this, Mark, but unfortunately, uh we're running we're running out of time. Um that that is that is also all we have got time for. But thank you very much for joining me today. Uh answering all the questions. Uh you've been brilliant. But just before we do close, for anyone who's watching this uh who wants to kind of dig a little bit deeper, explore in a bit more detail, what's the best uh way for them to find out uh a bit more about what ujet are doing?

SPEAKER_01 27:33

Uh get to ujet.cs. Um you can reach out to our team from there. Um there's plenty of information about Spiral by UJet, about the new AXL platform that we've developed, uh, and we'd love to talk to you and figure out how we can help you uh kind of reframe your thinking around the value of conversational analytics and then reframe your thinking around the value of AI in the context center.

SPEAKER_02 27:56

Makes sense, makes sense, absolutely. And and and obviously don't forget uh that you can also find a a wealth of uh related resources, uh, stories, articles, videos like this one uh at thextoday.com. And that uh that wraps everything up for today. I am Rob Wilkinson at the X Today. Uh thanks uh everyone for joining us.