Beyond AHT and SLA: How to Measure Success in AI-Powered Contact Centers

00:00Hi, guys Hi, advice from a call center geek, a call-center geek with a focus on all things call center. We'll cover it all from call-center operations, hiring, culture, technology and education. We're here to give you actionable items to improve the quality quality of yours and your customers' experience. This is an evolving industry with creative, people like this guy. Not, Not only is his passion call center operations but, he's our host. He's the CEO the CEO of Exp. All right, everybody back back to another episode of advice from a call center geek the call Western Pennsylvania, here in the states, work for a variety of different companies, from financial services to retail to healthcare, to tech support, and they handle any type of outsourced customer support, sales, that customer need, also I'm the founder and CEO of auto QA, O-T- T-O-Q-A.com. And auto QA is an AI startup startup where we are fully scoring contact center QA calls with with using AI. So I hope everybody's doing well. We Last week was just brutally cold the hard part right it's the February March trying to kind of you, and it's, it's something that, again, there's no book, book, written on this. This is all kind of needs to be kind of sorted out. So just kind of want to put a couple thoughts into kind of what are the KPIs that and how do they need to evolve AI, centric world that we're living in, in most contact centers. You know, we've had, you know, these, these core metrics of and handle time and really tracking the agent, right?

03:02Right? And how do we do that now with AI? And is it the same metrics that we need, right? Or do things need to evolve? And I think they do need to evolve. And I want to give some thoughts on what, on what they need to evolve into here. Because it is a different world, right? And I think we need to start, start to look at, lookUTF you know, from the impact of number one,

[Gap 03:25–03:55: no reliable transcription]

03:55How do we really measure if AI is KPIs today that have to kind of deal with a that self-service model to kinda how we're looking at this kinda advanced self-service model with AI. So again, we look at things in today's and I'm saying today's most contact centers, maybe have a little bit of AI and they're starting down that road. But most contact centers are still human being staffed, right? So we're still looking And then we look at our today's quality metrics, right? Do AI quality metrics have to change, right,

[Gap 05:43–06:43: no reliable transcription]

06:43there. So that gives you a benchmark of kind of what percentage things are being kind things that we need to start at. And we can drill deeper into then the wise, we can kind of get into analytics. Analytics to see, if there metric to have, and again, these are asked if you're going to AI, you got a a little So how much time is that customer taking with that that agent and again the agent and now the virtual agents are going to be less expensive so maybe they could composure but then how does that affect your customer experience or your nps like all that stuff we're still kind of figuring out but i think that these are the number that's a huge number to start to start to again that's your that's to talk about, I think there's hallucinations and it can give bad responses. and I'll have a really good KMS and all of this information but no one is actually QA, right there, AI. And what is kind of the AI I'd like to score? And I'd like to I think you can use an auto QA platform, right, to either randomly sample or sample all of those.

09:31But I think it's really important it's important that you're giving out the right information. doing it, doing AI at scale with thousands and thousands of customer interactions, if it's telling the right stuff, without you kind of looking at that. So again, that's another metric.-, we would QA our agents, why would KU-A the AI?, we fine tune that thing so it's really crazy accurate and we're getting better at that. But I think that's something that we don't really talk enough about. So, again, I think that those are some core AI metrics that kind of just efficiency and look at how well you have your setup, right? Is it doing the job that it is supposed to do? Is it handling your agent or your customers? It most of your issues. It's, doing it, in kind of a relatively quick manner compared to saying that we're using agents, but if we. Agent assist-assist, right, we're using other things that that are helping our agents.

11:01What impact is that having. And again. and again, we want to measure that right because we don't want to just listen to our AI chatbot companies telling us this stuff. But how do we want to start measuring it? And I think, you know, here are some thoughts that I have on this, right? to have an AI assist rate. How frequently are agents HU using those AI tools during an interaction, right? And I think that's important, right? To see, number one, is that tool actually providing but to make sure agents really need to be educated in some other things as well. But I think that that's a really cool,

[Gap 11:42–12:09: no reliable transcription]

12:09that I'm, so if I'm using an AI tool, like an agent assist, how much time am I saving, right? Because if I-m not saving any time, is there a cost savings there?

[Gap 12:22–12:52: no reliable transcription]

12:52,,,,,,, knowledge retrieval time is something that that kind of we talk about here, is it worth it? And that kind of goesincerity is an agent assist tool, is it worth what we're doing, right? Are we looking and we're taking a lot of time to do that? And if if we win, what is that time?. And then we can kind of calculate the savings that will get from it from an agent assist type tool. your homework a little bit ahead of time So we need to start to talk in terms of how is AI affecting the customer experience, and So as an industry, I think we need to start to think about these things a little bit more. And I think when you talk about quality metrics, I see a lot of organizations out there that that are just kind of giving some overall summaries and calling that quality.

14:52And I t hink that that is a huge, a good thing, to have a, you know, here's kind of some coaching chips for the agent. Here's kind of the summary of the call, what happened.

[Gap 15:05–15:32: no reliable transcription]

15:32So I think, you know, looking at auto QA accuracy is a huge... vastly massive thing, right? You shouldn't, you know, dealing with anybody that's not in that, and you want them to get them at scale, and so I think you know looking have an impact in a change of what is what is your contact center, and the other really cool thing with AutoQQ is the the session only, only analytics. You can get deep into why customers are can't are they saying you're a a competitor's name, right?

[Gap 16:32–16:57: no reliable transcription]

16:57We've been scoring agent and customer and agent sentiment scores, but looking at shifts, looking at what agents do they start with really negative customers, but do a really good job at the end of the call of turning that into a positive sentiment customer. All of these type of things that we're now, we're drilling deeper into what we can score from a QA standpoint and from an AI standpoint. So how do we start to utilize not just what age the end of the customer and start to kind of give that some score, to the end of, during calls with the kind of the intent of the the We can do that very easily, how much of the cost were each of these agents, and how much

[Gap 18:34–19:04: no reliable transcription]

19:04What we're doing when we look at quality, when we look at sentiment, when we look at things like CSAT and NPS, when we look at the overall cost of our contact center. or is things staying the same,

[Gap 19:23–19:51: no reliable transcription]

19:51to think about here in Expivia and that auto QA to,