CX Metrics In The Age Of AI: Stop Optimising For Speed

AI is already changing the contact centre from a transaction factory into a hub for complex problem solving. In this CX Today roundtable, Rob Wilkinson is joined by Martin Teasdale (Team Leader Community), Steve Morrell (Contact Babel), Sangeetha Rai (Barasch and McGarry), and Rebecca Wetteman (Valoir) to unpack what that shift does to the metrics leaders rely on.

The panel explores why legacy KPIs like average handling time (AHT) start to mislead when AI takes the simple work and leaves agents with emotional, high-stakes interactions. They also discuss what leaders should measure instead, how to avoid agent burnout, and why AI tools only work when agents trust them in real customer moments.

Transcript

SPEAKER_02 0:08

Hi, I'm Rob Wilkinson at CX Today, and today we're having a conversation around the evolution of CX metrics in the age of AI. Because basically, AI is already fundamentally changing contact centers from what they used to be, which is like a factory of transactions, into a hub of complex problem solving and a massive amount of really rich data and insights. So we're automating all the simple, predictable, kind of high-volume uh things for the agents to then have to handle 100% of all the difficult stuff. So the emotional things, uh all the high-stakes interactions. And unfortunately, um our management habits haven't actually caught up just yet to this new reality. So if we continue to manage this new environment with all the old assumptions, we're basically risk that we're gonna be flying by. Um we need to be optimizing for sorry, we don't need to be optimizing for speed. Um we need to be optimizing for experience because otherwise our agents are gonna burn out uh and our customer experiences are gonna go um well, gonna go down the Swannee. So I'm just gonna do a very quick introduction to everyone on the call today. So um Rebecca Wetman, principal at Valua, thank you for joining us. Steve Morel, Managing Director at Contact Babel, is with us also. Um Sangi Thore, uh, the Chief Operating Officer at Barash and Bagarry, and um Martin Teesdale, who is the founder of the Team Leader community. Thanks everyone for joining me today. How are we all? Great. Pleasure, thanks, Rob. So let's get stuck into some uh some details. So um we're gonna start off by looking at kind of a bit of a bit of a contextual piece and a bit of kind of setting the scene and about the market reality. So um you've tracked contact-centered data over more than 20 years now. Um we've got Rebecca arguing that average handling time possibly is is is now obsolete because AI has taken away the easy volume, you might call it. Is the industry actually ready to kill AHT yet? Or or is it still that golden carp that keeps giving?

SPEAKER_01 2:28

I'm gonna sound a bit reactionary here. Um because there are an awful lot of uh really good arguments in favor of you know dropping AHT, making it less important, things like that, but a few counter-arguments to that. So, number one, why do we measure it? Well, I think we measure it because it can be measured. Um, it's easily understood, it's objective, it's familiar, it's something that people at all levels of the organization they don't have to work with in a contact center actually understand. Um is it as relevant as it used to be? No, absolutely not. A call should take as long as a call takes, um, but it does impact on speed to answer and call abandonment, and these are the things that customers actually care about. And obviously, you know, the longer the call, the more it's gonna cost. So, you know, I'm kind of arguing the case that we shouldn't throw AHT out the window. Um, another way of looking at it is you know, if we're gonna try and move it on without binning it all together, perhaps include the number number of interactions that had been called previously, but now get handled by a self-service. And that might give an overall view of the kind of cost per interaction rather than just the you know the cost per channel, because as our data shows, every year phone calls get more and more and more expensive, they get longer and longer and longer. Um, so it's just about considering the number of calls, not just how long each one takes. Um, so really my view is that while it's not as important as it was in the past, there's no need to get rid of it completely, as it's a strong indicator of cost. Um and if an agent's on an unnecessarily long call, the queue's building up, and that does impact on you know things like customer experience and satisfaction. However, bring AI into this after what a minute. Um, if AI can be used for things like customer authentication, um, agent assistance, post-core work, summarization, wrap, you know, auto disposition, that's going to cut AHT without cutting quality. So maybe that's something to bring in here as well.

SPEAKER_02 4:23

Right, yeah, because we've always kind of chased those efficiencies in the past, really. That's been the kind of the reason that's been a metric. Um, I think there's a chance that because of these complex conversations that are happening, AHT could actually go up because we're dealing with things that take longer to deal with. Um hard work, you might call it. Um, how do we explain that to kind of the chief, the the chief finance officer that's kind of got all he knows is that figure didn't used to look like that and now it does?

SPEAKER_01 4:53

Well, just to explain it. Um, you know, I assume these people got where they are by being able to understand fairly basic concepts. Um, it's you know, it's it's like, right, what do you what do you want here? Is it all about cost? And then bring into the things like cost of cutting a call short. You how many, you know, how many people are then gonna call back, how much revenue you're gonna lose, you know, if talk to them is the things they understand then, it's not just about a simple cost per call metric, um, it's about everything that's around that. So, yeah, I mean, I'd I'd I'd like to think I don't work particularly closely with CFOs, perhaps you do, um, but I'd like to think they were capable of understanding something that was like in fashion 20 or 25 years ago and the world's moved on since then.

SPEAKER_02 5:33

No, it's a it's a it's a valid point. Um I guess uh I don't work with any CFOs myself, Rebecca. Maybe you can throw some light onto that. But previously, when we've talked around this subject before, um you mentioned that modernizing with the wrong scorecard, maybe is the is the word to use. It could be a trap. Um looked at you know, what is the single biggest mistake you're seeing companies now making when they try to overlay you know AI on their old KPIs that they've been using?

SPEAKER_03 6:05

Yeah, and and I want to address what what Steve said as well as CFOs, because I think it's a really important point. I think we're seeing people talk about changing metrics, but I think the the mistake we make, and this goes back to Steve's CFO point, is creating metrics as a proxy for something that isn't really an actual good proxy, right? I used AHT because it was a good proxy for a lot of things to measure. CFOs understand all about using the best data they have, not the perfect data, to make decisions, right? If I'm deciding to make an investment, I don't have all the data about what the future is going to look like for that investment, but I still have to make that decision. So they're very good at thinking about things like expected and worst case to understand what metrics matter and what metrics they want to launch. I think to more directly answer your question, Rob, I think the challenge is not with coming up with new metrics, but with the management and the incentives around that to make sure I'm actually getting and guiding toward the results I want, right? So I'm not suggesting that we throw out AHT, you know, if you you throw the baby out with the bathwater, you just get a dirty wet baby, right? Which none of us want. But what we need to do is look at what are the metrics are we really looking at and what are the incentives, what are the management strategies we're putting around them to make sure we're actually guiding toward the results I want. So, for example, if all of my more complex calls are going to agents now, because AI is taking care of 80% of the easy calls, I want to incentivize those agents maybe to take more time in certain areas, maybe to look more critically about what AI assistants may be recommending in terms of responses, rather than just giving the response that an AI agent gives me. So I need to think about how do I measure quality of outcomes and how do I manage toward those outcomes, not just about what is the new metric or what are the bunch of new metrics that I need to measure.

SPEAKER_02 8:10

That makes a lot of sense. I guess um the the key then is how that lands within the operation. And uh, Sangeetha, I know that you're kind of running an operation, so you've got that a really good uh understanding from that perspective. And also you've spoken in the past around we make a lot of uh mistakes that we keep repeating in our space, in our sector. Uh, we never learn from those historical issues. Um, and you're in you're you're a big advocate for encouraging people from you know um learning those lessons. Um, are we making another mistake now by expecting AI to just be a kind of cheaper human uh or to just by rolling it into the operation, or what concerns have you got around that?

SPEAKER_04 8:55

Yeah, unfortunately, I do think we risk repeating the same mistakes. You know, historically, companies often scaled contact centers by adding capacity. We you know, we were either offshoring, we did self-service. Rather than addressing the root causes, like broken customer journeys, policies not being clear, our knowledge gaps. So now with AI, we're all tempted to do the similar thing, and some organizations are treating it like cheaper humans that can handle more volume. But if the foundations, foundations are not strong, we all know that AI will amplify these problems faster. So, you know, you talked a little bit about it before, but what I see now different is the potential, right? So when I first led contact centers 15, 20 years ago, and even up to like a few years ago, the mandate was reduce cost per contact, manage AHT, increase, increase deflection, optimize staffing, uh optimize staffing. And the contact centers were treated as operational overheads. And I think with AI and omni-channel data, the contact centers can now capture this voice of the customer at scale. So whether it's product issues, whether it's friction and journeys, policy breakdowns, sentiment trends, all of this, AI can surface all of these patterns in real time and feed them back to different departments, you know, product, digital operations, so businesses can actually improve. So I think we have a real opportunity here. And it's this where we're going to transform the contact center from operational overhead to a source of business intelligence. So I really hope that we take a pause and we don't use AI to make the same old system cheaper instead of making it smarter.

SPEAKER_02 10:49

Understood. Understood. I think that really captures uh the opportunity, I guess, that that we we see in front of us at the moment that we need to take advantage of. So what I'd like to do is kind of just move on from the um almost setting the scene um and move into kind of impact and how this could be measured more. So um I think we've all we all like we all get what like being trapped in a in a bad bot loop looks like. Um that's not going to be new to anybody. Uh um Valvar's data suggested um a new metric could be used and um called effective escalation. And I really, I really like this. Um, it's basically knowing when the bot should quit. Um Sangeetha, in a and in a real operation, how do you measure if a bot is actually helpful versus just kind of blocking a door, I guess?

SPEAKER_04 11:41

Yeah, I want to take a step back, right? So when we first deployed bots, we had to ask ourselves this honest question: why are we doing this, right? Are we deploying bots to provide 24-7 service so customers don't have to wait and queue and get and they can get consistent answers? Or we are are we deploying it to be digital gatekeepers? And so once we aligned on improving the experience, and that's why we were in when we were deploying these bots, the metrics became clearer. So one metric um I do like is handoff accuracy, right? So does the bot recognize when the issue is complex and it just steps aside and without the customer having to fight for it? How many times have we said, you know, speak to an agent, speak to an agent until it actually transfers? Um and then the second one I would say is around containment with satisfaction. Um the bot that's handling the volume, it's not successful if the customers leave frustrated or they call back later. So looking at repeat contacts, their sentiments, and the resolution really tells you if the bot is helping. Um, and I also do like the idea of measuring effective escalation because one of the worst experiences is getting stuck in that bot loop, and many of us have nightmares about that when you need help. So, and we've all been there. So, if the core reason, if our core reason was building the bot for improving customer experience, so we can design the bot to be well, actually we can we can design well-designed bots to remove friction. Otherwise, if we want it to be gatekeepers, then a poorly designed one just moves it from one queue to somewhere else. So move the customer from one queue to another, another one.

SPEAKER_02 13:32

So that's key, isn't it? Because that's that's losing that resolution piece that we're trying to measure and trying to get a hold of. So um, Rebecca, when when in your review that you did, you you kind of um you mentioned something like uh the success rate um and and that that's a much better metric versus what maybe is currently a bit prevalent, which is deflection. Why um why is deflection often is it um is it a vanitary metric more than anything else? Does that hide what's going on underneath, really?

SPEAKER_03 14:04

Deflection doesn't necessarily mean that the customer got what they wanted. It doesn't necessarily mean that the customer's issue was resolved. The great thing about deflection with AI is I can gather all of that data about how the customer is responding, how they feel. And if I'm really tracking the data the way I should, I can tell if they try another channel later to get a better answer, suggesting that it was really a deflection and not a successful interaction. So what I want to be able to do with with with the AI when I'm doing these automated interactions is be able to take that sentiment data, which I can now do at scale with AI to understand the quality of the outcome, not just whether the customer dropped off. Right. And yes, the thumbs up, thumbs down stuff is is marginally helpful, but what's much more helpful is being able to understand in context what the customer actually said as the course of that interaction and really how happy they were.

SPEAKER_02 15:05

Yeah, and I that is key to all this measuring. It's it's what you do with that data then, isn't it? How do you put it into practice? How do you kind of report on it? What tools do you use to interrogate it and to understand it in more detail? Um, I think that stage is it's still a little bit away from us yet, although it's certainly becoming you know far more kind of at the top of people's requirements list at the moment. I think a lot of operations are still using kind of traditional dashboards and that kind of thing. So um it Sangeetha, but when you look at your kind of operation, the dashboards that you're using, how do you tell um where a handoff between AI and a human hasn't been successful? Is it kind of how long it's taken, or is it customer tone that kind of rings the alarms?

SPEAKER_04 15:55

Yeah, so when we look at dashboards, it's rarely just one metric that tells you that the handoff failed, right? It's so it's a pattern of signals. And so the way I look at it is that the goal is to make sure that AI actually solves the issue, right? So the first signal I look at is repeat contact. So if the comeback, if customers come back shortly after interacting with the bot, that's a strong indicator that the issue wasn't really resolved. That's to Rebecca your point earlier. Now, if the interaction does escalate to a human, the goal is that the experience feels like it's simply continuing, right? Not that the customer has to start all over again. I mean, how many times have we been on calls where we have to repeat the same thing we told the bot or repeat the same thing we told the first agent that helped us? So for that, I look at time to resolution after escalation. So if the interaction becomes much longer once the human joins, it potentially means that the agent had to rediagnose the problem or gather all of that information that they already gave the bot. Um, and then the second piece is the customer tone and sentiment. So if the sentiment drops right after escalation, it's usually a sign that the customer is already frustrated and even before the agent, the human agent, actually began helping them.

SPEAKER_02 17:11

Right. Okay. Um interesting. So we're measuring things, we're kind of um we're finding ways to tell when things are going right and when things are going wrong. Um it's normally useful if we can benchmark this stuff to help us understand whether these kind of levels are are right or or wrong. And Steve, you know, you've got your uh your access to a wealth of industry benchmarks. Do we have anything yet for this? Or is it too early? Um or maybe is everyone just making everything up as they go along at the moment?

SPEAKER_01 17:46

Um, yeah, it's um no benchmark that I've really seen. Um I think a lot of companies aren't even measuring this. Uh, knowing when a bot should quit is a pretty subjective view anyway. Um, a few things to kind of talk about here, to be honest, because when we're talking about AI and chatbots and this type of thing, it's very important to differentiate between rules-based bots, which basically can do certain things, and you know, AI bots that use generative AI and they've got an awful lot more sophistication. Now, the majority of the stuff that's being used today is still rules-based bots. So um what uh some stats we do have for you, um, at least 25% of chats end up on another channel. Um, that's from companies that actually measure it. Um, around 60% of chats that a bot initially handles will be dealt with by an agent. That's not necessarily a failure though, because it might just be the bot's job to um triage it and pass it on, um, you know, gather various information, take them through security and stuff like that. So that's not in itself a failure. Um one thing that's it is worth looking at why we have to kind of think about moving from the rules-based bots to to uh properly AI-enabled bots pretty soon is the length of time it's taking. Um even only in 2022, where 44% of chats took more than five minutes, that's up to 60% today. And there's very few that are being handled in less than three minutes. So, what bots are originally brought in to do, like take the easy stuff out of the equation, is another you know, self-service thing. Customers aren't really using it like that necessarily anymore. They're trying to get sophisticated, they're trying to do multiple things, they're trying to do complex things, and the bot eventually is just you know shaking its head and saying, Look, I can't help you pass you to a person, and that's not saving any money. Yeah, the person didn't uh didn't want to talk to a a a business in the first place on the phone, which is why they used a chat bot. But so we've maybe got the assumption that people are using bots in a way that they're not so much anymore. Um, so that is something else we we kind of have to throw in there as well. So the complex work we talked about being passed to a phone, it's also seems the data is saying it's being passed to live agents doing web chat as well. Um that's taking longer and it's costing more. And this year I was really surprised to see we've just published our chat bot report today, actually. Um there's been a reverse in chat automation. Last year, 53% of web chats, um, whether it was full automation or triage, were handled in some way through automation. Um that 53% has dropped to 46% this year. So they're getting longer, they're getting more complex, and there's less being handled by automation, which was a surprise to me. I'm not easily surprised after 25 years.

SPEAKER_02 20:26

So that's actually there's there's some other risks inside that those numbers because there's there's a risk also that um customers are are getting almost trapped in with the bots, um, and it's taking longer than they want. So, okay, we're managing to like not have to deal with it in the operation, but actually the customer experience is is is is not stacking up. It's that's a really big risk. Um, and no wonder they end up coming through to other other people. If they do, they might just leave and never come back. So that's a really big flag. Um so we need to come on to uh talking around the impact um of all this uh change and all these kind of stats that we're talking about. Um and let's get it into the the realm of um the people, so the employees on the front line and the kind of the customers, obviously, we've just touched on there, the risk to those guys. Um I'd like to kind of get into this a little bit, Martin. You you you have got a uh a community that lives on the front line, you're you're you are speaking to CX leaders every single day. Um Rebecca's research highlights um something called suggestion relevance, and so that's whether agents actually trust and use the AI's advice, uh, if the uh agents are ignoring the AI co-pilot, and is that an agent problem? Uh or do you think that's a tech problem or is it a hybrid thing?

SPEAKER_00 21:49

It's it's such a lovely phrase to to start with. Um for me, that for me based on interactions with all of the people. within the in the community there's there's probably three factors and often in operations you feel like things are done to you and your day changes overnight and often that's done with predominantly good intentions you know it's this is to make your life easier and to make the delivery of what we do for our customers better and it's and it's going to help you. The extent to which that usefulness lasts andor is relevant um I think is determined by at what stage operations have been involved in the development of the tool and and you know now we're seeing finally um AI tools that are there to help agents and help team leaders you know team leaders are kind of a the last of the the people that are getting access to tools to help them prioritize their day um coach their teams better use um feedback better knowledge management there's so many great things that are starting to to happen but their their usefulness um will be determined by have they been involved in how this is going to be used where the knowledge is taken from because for AI now read paper scripts when I was an agent and the moment you become competent you would kind of think well I don't think I need this anymore I'm not gonna I'm not gonna use it and so it's have they been involved in the outset the second one then is both competency and tenure so I think it's ri we're seeing a lot of AI tools really really helping new starters and maybe people that are having performance blips with kind of ongoing in in the moment support you know choose this next action here's what you should be referring the customer to this is invaluable for comforting people giving them um support in the moment especially when you take into consideration there's so many people still working from home that they aren't able to um in the moment speak to their team leader so they're being supported through AI tools so if you're a new starter if you are um struggling with a particular performance area absolutely great no you know no problem the these things are are are definitely having a uh a positive impact and then I think the third factor is the extent to which the industry you're in and the service that you're dealing with has uh is it stable or are they is it a constant period of change um because the the brains behind the AI whether that's the knowledge that's being fed into it to to help steer someone to take the right action if that's becoming outdated or replaced and not updated within the agent facing world very very quickly people can lose faith in the support that they're getting um and you you can you know I've seen great tools great strategies die a death in in a contact center operational floor pretty much in an afternoon where the teams have decided no this is it's wrong it was wrong once therefore we're not trusting it and we're gonna we're gonna carry on with our cumulative knowledge so that there's a number of factors I think I think we still need to see more done for uh team leaders. We're we're asking them to do a lot and uh we could definitely help them a lot more with kind of prioritize prior prioritization of their day and their time management. I I do worry that the development of our leaders is not keeping pace with the development of the technology around them. You know whether that's how to be an emotionally intelligent leader, a culturally intelligent leader, we're not doing enough. So that that does give me some concern but I think Rebecca's phrase is is is really it's it's a far more adroit way of explaining what I've just spent a long time trying to explain. So yeah it it but I as I say I think there's three there's three factors. Are they involved at the outset? What's the tenure of the workforce and competency level of the workforce and how are you keeping it up to date that's exceptional actually because it's it it it I wasn't expecting there to be that much good news coming out of it.

SPEAKER_02 27:13

I appreciate there's a lot of like negativity there's a lot of risk and there's there's a lot of lessons not being learnt which is the whole point of us having this discussion today uh from what you're saying though that there are some tools being adopted there are some new things that are there's some green shoots maybe on the horizon for for agents and team leaders which is which is awesome um I guess we need to kind of just focus on that a little bit more because you know the agents are our frontline they're the people who are dealing with the customers when they do get through to them and now they're dealing with the you know only the hardest the most difficult the most emotional um types of interactions um that's got to be like a massive cognitive load for them and and especially when these guys haven't necessarily been trained up in this type of environment originally so their roles have fundamentally changed how how do we change the metrics that we measure them on in order to protect them from burnout because otherwise this is kind of a it's like a it's like a bomb waiting to go off.

SPEAKER_00 28:17

This is one of my favorite topics as a I I started in this industry industry as an agent and the transactional queries were an unofficial break. If you wanted to change your direct debit or change one thing come to me I loved it because it was it was a chance to take a cognitive break right now if we are under the impression that um talking to our customer facing colleagues and saying to them great news AI's taking all the transactional stuff you are going to be a problem solver. All the complex stuff is going to come to you and we're not gonna change anything else your pay will be the same your breaks will be the same the support around you will be the same you are not going to have thousands of people cheering and saying oh this sounds fantastic we we have to be really really careful I did some research at the end of last year uh where we research we surveyed over 500 agents and we asked them when you look at other people leaving what do you think are the factors that determined why they left so it's easier to answer that question than one where you're employed and you're talking you know what are the factors that might make you leave and the single biggest reason was fear of burnout um and when you when you dig into that it is because we know the job is tough right but asking people to deal with highly emotive urgent complex customer issues time and time and time again without changing the environment around them we're heading we're it is not a good place to be going to um we could have all we could have great tools around them we could utilize AI but yeah we are heading for a burnout crisis if we if we don't look at we can do so many great things with scheduling the good news is you know 70 80% of the people I can't remember the exact number but the people that we surveyed were able to take an unscheduled break after an emotive or difficult customer call but we need to be doing far more than that I'll get off my soapbox now.

SPEAKER_02 30:41

No no though look that's a a great solution isn't it there's there's there's got to be other ways of of helping I guess in order to know when we need to step in though we need to kind of and and Rebecca I want to come back to this great um this great measure you've you've said this suggestion relevance how can we use that better is it is it just is it just clicks or is there anything deeper that we can understand from that to see whether that AI is adding the value to the conversation and and whether that's detracting from the you know the conversations that the agents are ending up having to do.

SPEAKER_03 31:15

Yeah I think I think Martin's point about knowledge and currency of knowledge is really really important right because even in a in an environment that may not be evolving a lot I need to be able to understand the data what's coming into the contact center how those queries are changing if I have a particular topic for example that is not being addressed well by the AI so that I can quickly update knowledge provide that information to increase that suggestion relevancy. And Martin's actually absolutely right if if agents try it and it doesn't work they won't use it. And if they try it once and it doesn't work if they're given something else that looks if it looks like a duck it grabs like a duck they're not going to use that either. Right? So I think we have to think about how we phase in adoption how we look at suggestion relevance but also how we look at all the data to to understand again why is the agent not using it and how do we make sure to to Martin's other point that it's a tool to enable and empower them not to take the mental cognitive breaks of their day.

SPEAKER_02 32:29

Yeah yeah I think yeah we really need to get more kind of weight behind this because it's um yeah it's it's like it's it's just it's just waiting to happen. I think um I'd like to look at these kind of evolution of these metrics a little bit um and I'd like to just bring in both Steve and Sangeetha here on this one. So there's a there's a there's a risk that um what we we do we call them big brother metrics that we're currently using you know monitoring for compliance and that's it. If we try to move towards coaching metrics um which would be you know to support people to to to kind of Martin's point that's what we need to be doing more of how how do we achieve that are there any um are there any kind of quick wins for our audience that they might be able to put into practice is there any kind of um data to to support where this kind of alternative approach is actually succeeding and and and and working well um right what yeah that's a tricky question some maybe I'm a bit cynical here but whatever you do if people are rewarded on metrics then the metric then becomes the target.

SPEAKER_01 33:39

Whatever metric you decide to use should really be kind of aligned with the experience that you want to give customers and just you know kind of be honest about how you want your operation to be run. If it's about it's we want to be reducing costs that's that's it that's our number one fine your choice and and then you pick metrics that support that. But if you want to do it differently then you know start with the honest outcome you want not what you think sounds right um and then work out backwards what behaviour needs to be supported what metrics need to be applied in order to get to that point. It's almost like start where you want to be at the end of it and then work that rather than like right here's a load of metrics let's just go after that and what do we end up with it's you know what outcomes you want and then pick the metrics accordingly and and I I do think these days that you know the big brother metrics because we we the met the the stuff that we look at each year we say right which of these hello contact centre leaders which of these metrics are most important to you and you know we've tracked it over many many years and things go up things come down. One of the things that still amazes me is um like adherence um adherence to schedule is still super important. Customer satisfaction first but that's a bit woolly because what drives customer satisfaction then it's like you know adherence to schedule it's like really um that's it it's still a thing so it was we're still there we've not kind of moved on from that because that's the way it's always been done and that's what people understand. But you know this is you know this is a brave new world and maybe it happens over such a long and slow period that we don't really realise the world's changed. And it's only looking back perhaps five or ten years into the day track and actually see oh wow that really has changed um maybe you know when you're on the ground it's it's different.

SPEAKER_04 35:23

Speaking of being on the ground Sangeetha you're definitely you're definitely there can you help Steve answer that um adherence question or have you got more uh light you can throw on moving to coaching metrics no I I agree and I also agree with what Martin said before I think you know the development of leaders is not keeping up with the pace of uh development of technology um we talk a lot about upskilling um teams but I think the real shift is you know actually we're shifting the mindset for leaders as well and you know we talked about people being fearful of change and it's not just the teams it's the leaders as well so it and it has to start from the top not team supervisors but really even higher than that and for a long time many leaders have used metrics for oversight things like average handle time adherence compliance metrics and things like that and those big brother metrics where the goal is to check whether people are following the rules right um so as we have AI becoming part of the workflow we need to teach leaders to think differently so the goal is not monitoring people but it's helping their teams perform better. So things like you know you need to measure outcomes not activity right so um it's and also looking at measuring collaboration between the humans and AIs or using metrics for more coaching conversations instead of just compliance checks. So we want leaders to change their mindset from monitoring behavior to supporting and enabling performance of their teams.

SPEAKER_00 37:02

I I love that it isn't it fascinating I think one of the things that um a lot of people including myself say I'm proud to be in this industry for two reasons. We are at the cutting edge of the application of uh technology and AI we're we're using it every day and we and and there's not many industries yet that can say that but the main reason is that we're a people based industry it's the people that have kept us in this industry. I can remember my first steps into into leadership and then going through to senior leadership your dashboard and we've talked about some of these historical and still use today metrics my dashboard had two people measures on it absence and attrition that was it latter we've started to see um employee surveys and pulse checks and again I think they can they are great if they are used the right way it it shouldn't just be a vanity metric so your contact center can win an award and say look at our employee survey results we've got this much engagement that they should be used to check in with people on scale where where they're at and then dive into some of the detail and then that should influence a lot of your strategies and and tactics. I I've heard through the community a really interesting use of employee surveys to focus in on confidence because you can only be confident in your role whether you're an agent or a team leader if you're competent. So they really started to look at the confidence levels of their of their team members and I and I thought that was fascinating. You know if you can um elevate that to C level and uh the you know that kind of senior leadership I think we're gonna hopefully address that discrepancy between technology versus versus people skills. Because equally the same day I heard about that I heard about a head of department struggling to get the board to pay attention or even do another employee survey. Well we did that last year we don't need to do it again do we you know and so that's the spectrum is wide the the spectrum is wide and I think metrics and our um the you know our attachment to them now is a time to look at and and evaluate is it helping us is it preparing us now and is it helping us for the future what what should we be looking at? So I was I was very pleased to hear about companies kind of being a bit disruptive and just thinking right what what are the what's important to us?

SPEAKER_02 39:53

Yeah no I think that that's great um well it's it's a shame in one respect that that's a conversation that's having that's happening but it's great that there are you know positive there is positive news coming out of there um I think as well we we have an opportunity to kind of I I'm the last person trust me I'm the last person to say we need to throw technology at something straight away but I I am aware that we are moving from uh you know surveys of a handful of customers into measuring every single conversation from a customer's perspective we can also do that from you know an agent's perspective as well because if we're listening to those calls using AI we can identify when not just the customer's unhappy but when when an agent is struggling and I think we could do that across every single conversation. So that's that's you know maybe another conversation for another time but I think you know that's where AI can potentially help us with that as well. I'm just conscious of time guys and and and I'm I don't want to waffle on but I'm gonna do a very quick I'm gonna put you all on the spot a little bit here and I'm gonna do like a very quick round robin where kind of one by one I'd like you to tell us tell the audience um so if if you could ban that's ban one legacy metric or big brother metric whatever you want to call them um from the boardroom slides forever uh which one would it be and why? And uh just going around the screen we'll start with Rebecca.

SPEAKER_03 41:18

I I I I think AHT is one that we need to rethink moving forward. I think the alternative is understanding what not many folks are measuring today which is the level of complexity and emotional complexity of the calls that agents are actually giving.

SPEAKER_01 41:36

Awesome and Steve. Okay um not quite sure it's what you're after but I'm gonna go for MPS. It's too simple. Any customer service got response bias and it's being gamed as people are rewarded based on customer scores.

SPEAKER_04 41:52

Fantastic I don't I would lean towards AHD but I would say anytime we remove something we have to be a little thoughtful and deliberate about it and not just do it because the rest of the industry is suggesting it right so you know occupancy, AHD, these are old old metrics and I would say like really thinking through and saying what makes sense for your business.

SPEAKER_00 42:16

Steve mentioned it earlier I think schedule adherence and the the reason I'd I'd want to kind of ban that is that we we need to we need to think about how our schedules reflect the way people work today and what we're asking them to do.

SPEAKER_03 42:32

So I think basically what we're saying Rob is that we need to rethink all the metrics.

SPEAKER_02 42:36

So we can't ban them but we need to go back to the drawing board. That sounds like a really good idea.

SPEAKER_04 42:42

Yeah even FCR is another one it doesn't make sense like it used to previously right with AI so we we really need to rethink all of them.

SPEAKER_02 42:52

Yeah because yeah it might not even be resolved just because it's been finished and complete by an agent an AI agent it doesn't mean it's actually resolved and I think that stat is being actually talked about at the moment um I gave one final one this is uh before we before we finish and and thanks everyone uh so far it's been great but I want you to give us one actionable takeaway each um what is the first new number that a leader can put on their dashboard tomorrow uh in this new age of of of metrics when it comes to AI it has to be the success rate as measured by did the customer actually get their issue resolved not just did they suspend or end their interaction Steve um okay start with the new one customer effort score um I think that's all all roads lead to customer effort in the end of it regarded what channel you're choosing, how long it takes you, the amount of you know emotional effort um so I I think that's a very underappreciated metric hardly very

SPEAKER_01 44:00

Rarely seen, but I'm a big fan of it. Not an easy one to measure, though. That's exactly right. And Sangita?

SPEAKER_04 44:06

I would say time to value for AI implementations, how quickly the AI contributions actually improve resolution times and quality. So a lot as leaders focus on strengthening the foundations that we talked about and redesigning workflows, and we change the mindsets of the leaders and the teams. I'm thinking that the time to real uh time to value is gonna go down. So that's a metric that shows impact and not not just busyness for um teams.

SPEAKER_02 44:36

Absolutely, yeah. We don't want busy fools. And um, just to round things up and to finish that one off, that's uh for the final one over to you, Martin.

SPEAKER_00 44:44

Sorry, I'm gonna be biased. I I think every metric we have flows through or is influenced by team leaders. So I would um ask senior leaders to think about the team leader experience uh and to have that on their on their dashboard. What are they what are they focusing on? How competent are they, how are they supported? Because if you get that layer right, then uh you can you can get everything right.

SPEAKER_02 45:10

That's brilliant. That's an uh and that's a such a fitting way to to finish this discussion. Thank you everyone around the table for joining me, for answering all my questions, uh, and for being brilliant. Um I hope the audience at home felt the same and uh come along and look at the next one we do in a uh very, very soon. Cheers, guys. Thanks very much.