Author transcript
Hi and welcome. I am Rob Wilkinson at CX today and today we’re taking a closer look at why 100% conversation analysis only matters if it actually changes how contact centers coach agents improve quality and act on customer insight. So, if you’re challenged with turning more conversation data into meaningful coaching or if you’re exploring AI powered quality management as an opportunity to make feedback fairer, faster, and more useful, stay with us cuz today I’m joined by Taylor Whitman, senior solutions consultant at UJet, who works very closely with contact center leaders evaluating these technologies. And Taylor has a really great view of where the market promise meets operational reality. Welcome, Taylor. Thank you so much for joining me.
I’m so excited to be here. Thank you so much, Rob.
Great stuff. So, before we kind of dive in uh and get into too much of the detail, just for our audience’s benefit, can you just give us a little bit of a background to this topic and why does uh the gap between analyzing 100% of conversations and coaching from a much smaller sample uh make so much difference right now?
Yeah. So, I’m a solutions consultant. So, I’ll kind of frame this in what I’m hearing daytoday. I’m in buyer conversations pretty much daily and almost every evaluation right now is going to start with some sort of can you analyze all of our calls? And at this point, we all can pretty much every vendor can analyze 100% of these interactions. And this previously was a big deal because QA would look at maybe 2% of interactions grabbing a few calls from each agent. Uh but that’s table stakes now. So when we typically are in these conversations, what I actually like to ask is, okay, if we have all of these interactions, what is changing in your organization? And that’s where people kind of pause and they’re uh kind of like that’s a good question. Uh so a lot of these teams already have a full analysis, but they’re coaching from this tiny sample. So the problem’s really just moving. they’re they they don’t have they went from having not enough data to having more than they know what to do with which is why it’s mattering right now. So, uh we’re looking at these organizations that are having all of their AI budgets scrutinized and leaders are saying what’s actually in that data. Um and if it’s really just 100% of calls and a bigger dashboard uh and the same coaching then it’s really hard to tell uh and kind of back the value in that tool.
I can see the disconnect for sure. Yeah, that makes a lot of sense. Um, as you say, lots of vendors are, you know, saying they can do all of these things. Um, what you’re saying is, yes, but so what? You’ve got this recording, you’ve got all of this, it’s no use if you can’t do anything with it. Um, exactly. So, so let’s put that I don’t know in plain English for our audience then. Where is it falling short for? If a supervisor can’t actually turn those insights into useful coaching, what’s that impact?
Yeah, I think a lot of it has to do with the tools and who they actually were built for previously and now. I feel like a lot of the analytics were designed for an analyst. So, you have entire teams that are built to dig through all of this data. That’s their day-to-day job. Uh but when you bring it to for example the operations team uh or the frontline supervisors and they have minutes to make decisions for escalations or for their agents um we really don’t know uh what to do with that data. So I see this in demos all the time when I’m showing dashboards. The QA team is loving it. Like they know exactly what they’re looking at. But when you ask the supervisor like what difference will this make in your dayto-day or what are you going to do with this data on a Monday after the weekend when you’re logging into work they’re they don’t really know they they look at you and they’re like this is another dashboard but what do I do with it? So what the supervisors, the people who are dealing uh day-to-day in these uh operation centers, they need to see patterns. They need to see like value attached to agents and they need to know what they should actually do with it. So that’s where we’re seeing these kind of fall short. It’s the analysis. They have it all at the fingertips, uh but they don’t know how to use it to coach and they don’t know how to make decisions with this data.
I guess that’s that that’s going to cause problems because the coaching side of stuff that’s taking the detail from the front line that’s taking the customer’s voice um in 100% of interactions and it’s turning it into strategies to help them improve. So what what does that look like for the supervisors who are grappling with all of that then? They’ve got maybe a larger team. They’ve got uh limited coaching time. Um they’re always struggling to find time uh for their, you know, their teams. More interaction data than ever before could be quite overwhelming. What what’s the kind of where does that show up for those guys?
Yeah. Well, I I I think it’s twofold. So, number one, the supervisors are the ones in the room during an evaluation. So, they’re the most honest people. And uh when we’re doing a buying process with them, they’re like, I know that we have this data, but I don’t have the time to find it. So they they don’t necessarily know or have the time and focus to be able to sort through everything. And then the second part of that is that a lot of them uh came from the floor themselves. So, a lot of them have worked their way up through the contact center and they want to be a good coach and they want to know uh how they can truly help their team because at one point they were in their shoes. So, it’s a bit of guilt of making sure like I know how I can be a good coach having been in that shoe. Um, so it’s, you know, making sure that they know how to find that information, but also making sure that they are able to truly make these developmental conversations. Uh, and a big part of that is not just going to be another platform that they can log into. It’s it’s actually surfacing information that’s valuable and useful and they can enact in their day-to-day with their agents. It’s uh being able to tie it to specific agents, the analytics that they’re gathering, being able to tie it specifically to a problem and to a person um and really just allow them to utilize this data accurately.
So, it sounds like we’re taking this massive ocean and rather than boiling it, we’re kind of going actually there’s some really specific actionable things that you can do here and surfacing that for them to to target that and get to the kind of the big wins like straight away. Does that sound like a good summary?
Yeah. Yes, exactly.
That sounds like a a nice promise. Um, I guess when QA feels like someone’s catching you doing something wrong, cuz QA’s had such a bad rep for such a long time, and this feels like maybe an opportunity for that to change and for QA to become quite a supportive and, you know, really uh really helpful um process for the for the teams on the receiving end of this potential coaching. So um what changes when the coaching feels that way? what what when people’s development needs are being met, h how does that impact what does that impact elsewhere is is it it’s got to be the the kind of front line, it’s got to be the customers as well, right?
Yeah. Um I think what changes the most uh specifically I think it’s a few things. I think number one you said you know this policing kind of like making sure that we’re like you know watching our agents and from the agent perspective if you put yourself in their shoes like they could have 1,000 calls per month and with QA the way it was historically done we’re pulling three to four of those calls and considering that the full sample of all the interactions they’re having and if I were them imagine I had a bad day and it was the worst call and that’s what I’m being scored off of. So, of course, I’m going to be defensive and upset because it’s kind of like a highlight of my worst moments in my job. And then I think the other part is that um with scorecards traditionally, you are learning to perform for those scorecards. So, you know what you’re being graded on. And so, it’s less about how do I really develop my skills and it’s more about how do I uh like make sure I’m getting 100% on this test and I know the answers. So, they’re asking for feedback and they want to see their own trends when it’s less about, hey, you hit zero out of 10 on all of these five things and they’re really wanting to um make sure that they’re doing well when it’s more about development and support. So, when an agent or a supervisor is coming to these agents from a compassionate uh standpoint with a a f full view on their performance, I think that’s when we’re starting to see uh some of these really positive shifts in these agents of wanting to do better, not just to have a higher score, but to actually um improve their performance. And so I think uh kind of all of that to say is that the coaching will follow the agents progress over time instead of these small samples of all of their interactions. Uh so it’s going to be balanced. I think the system, these QA tools will start to find your strengths overall and not just grade you on all of your misses. So that’s where we’re going to see the shift into like these more supportive conversations. uh it’ll be less about surveillance or um docking new points for your negative calls and more about uplifting you and making sure that we can progress your development based on a full view of all of your interactions.
Sounds like it really does move away from this kind of tickbox exercise that it’s been for a long time and into actually, dare I say it, something with a bit of robustness to it that’s a representative sample. You know that that’s got to be so much easier to buy into than you know somebody who might not even they might have a a concern that someone on the QA team doesn’t like them or anything. Um so I guess on that point though there there is definitely like um a fairness angle to consider here because if QA is based on a small sample of interactions like it it has been and that those scores influence the coaching and confidence even maybe career progression I guess um there’s going to be the possibility for bias or inconsistencies to creep in there. Where have you seen that kind of impact and and and and is that something that this can help?
Yeah, absolutely. I think this could help it. Um, and where do we see this impact? For example, 70% of frontline CX workers are women. According to the US Census data um, and that’s not always reflected in leadership doing the evaluations. So, when coaching is inconsistent, it’s not necessarily a neutral problem. It’s landing on the people whose careers are depending on the QA performance. And even deeper than the QA performance and that score, what is good is really dependent on um for a lot of these historical scorecards that were created years ago. And whoever wrote that form decides how you’re coached, who’s getting promoted um and who is managing out of the agent uh seat, for example. So the categories and the scorecards are not very neutral. Um, and so full consistent coverage of all of the interactions for all of the agents is one of our most concrete equity levers that we can use in the QA uh CX space. It doesn’t necessarily fix everything, but if people are evaluated in all of their interactions, not just a handful of these calls that they traditionally were, um, this should help to uh eliminate some of that bias or inconsistency.
Yeah, it’s easy. It’s easy for people to start to consider these things as as magic bullets and the kind of the gold rush towards AI in the early like a couple of years ago was evidence I think of that and and a lot of people made a lot of mistakes by by trying to run too quick but the technology now is is so much better. I mean in such a short pace of time it’s getting so much better and you know the use cases like this uh are constantly improving. I guess we have to therefore move away from talking around the kind of the people side of things and let’s have a little bit of look at under the bonnet on the technology side. Um but but we’ll jump in there with a quite quite important question because if there is the if this is the are the possible now people in these roles need to really understand what that 100% conversation analysis can actually make possible for them. Um the supervisors you’ve mentioned and the QA teams as well. How does you know what does that open the doors up to for them?
Yeah, I think uh like we previously talked about, first off, it will eliminate a lot of that defensiveness because uh we’re not just being scored on our worst call of the week, for example. Um and in turn, being able to look at 100% conversational analysis, uh is going to really kind of drive trust between our supervisor and our agents. If the supervisor is going to sit you down and say, “Here’s what you need to work on,” they can back it up. they they know across the entire scope of this agent’s uh performance what they need to truly address. So these are traceable answers. Um as well as if we’re going to ask questions like why are escalations up with my team, you now can get an answer with receipts attached to it. For example, like the skill gap with our agents drove 250 escalations this week and here are five calls. Um so they’re going to be real examples around for example that escalation uh increase. So it’s decision grade intelligence. Uh it’s something you’d be comfortable bringing not just into the CX contact center space but you can put this in front of engineering or do coaching one-on-one with your agents based on these accurate uh views into the CX space. So once you have that, you’re able to spot that pattern, review the calls, and then coach the agents accurately and measure performance even moving forward. Like we said we were going to work on this. Uh we’re looking at our uh most recent calls and I can see that there was an an improvement. So it’s less about scoring and it’s more about how are our agents improving?
Decisionra intelligence. I love that phrase. That’s a I’m going to pinch it if I can. Yeah, certainly said it does what it says on the tin though. So I I love that. Yeah. Um so so UJ has raised a point previously that a lot of teams have built their kind of QA scorecards a long time ago and unfortunately it sounds like like a lot of the time we move into the technology we run we kind of the silver bullet piece. we’re kind of all systems go, but we forget to actually change things that are around that that technology. And in this instance, I guess the scorecards are potentially going to be, you know, not fit for purpose anymore when you’ve got all the, you know, the art of the possible that exist. Uh even I guess not just the the scorecards, but the dashboards and the the coaching programs themselves when you open up all these possibilities with the the access to this data. Um, what how are you seeing that play out in in kind of the businesses that you spend time in?
Yeah. Uh, a lot of what’s being sold right now in the market is AI scorecards. So, it’s the same experience they’ve had. It’s now just filled in automatically, which saves time. Um, but if a scorecard hasn’t been updated in a very long time, you’re just grading inaccurately faster. So grading at scale isn’t necessarily coaching at scale, which is what I think we’re trying to drive home today is the coaching is what needs to improve based on the data that we have. So we encourage teams when we’re in these conversations to treat a QA or a scorecard like a living thing. So let’s look at what’s actually coming up daily, monthly, quarterly, and if we’re seeing new issues uh driving repeat calls or uh seesat scores dropping or skills, for example, suddenly separating your top performers, this is what we should be looking at in our scorecards and the coaching plan. So technology can be utilized to help you notice what’s actually going on. So the team should be able to decide what looks good uh based on today, not previously, not three years ago, not 10 years ago when your contact center started.
And right now means solving the issues that’s going on right now and not waiting till next month for, you know, those reviews to get done or whatever, which is always a kind of a problem when, you know, you don’t you only do that kind of 2% of of conversations. I I can really see yeah I can see all sorts of benefits here. I guess it’s not just the front line as well. You’ve got the customer improvement angle here as well and the opportunity to analyze every conversation that has got to mean some sort of change in terms of spotting the recurring issues that customers are having. So in the same way you can see the recurring issues from an agent’s uh training requirements, there’s going to be repeat issues from a customer’s perspective and and that might be a process that needs fixing or just a need to stop those same problems coming back again. Does this help to do that as well?
Yes. So, a lot of teams have sentiment analysis, which is useful up to a point right now, and it’ll tell you when your customers are customer customers are frustrated. Uh, but it’s not really going to tell you why or what happened uh in this call last Tuesday to actually cause that sentiment score to drop. So, with full analysis and traceable answers, you get to the why. You can see the increase in repeat calls about a specific issue. for example, when engineering just did a new update or the step in a process um that’s going to send people back to the cues when they’re trying to reach billing support. So you can see these real examples in the interactions and then show that to the teams who own the process. So, it’s less about um focusing on the full picture per se and more about finding the outcome of the customer problem that’s going to uh eliminate a lot of those calls coming back in. So, agents will also benefit from this. Uh they don’t want to deal with 40 calls a day talking about the last engineering release that’s crashed everyone’s mobile app. So, it’s less about um it’s let’s stop coaching agents around problems that could be fixed internally or in other departments.
I love that example that that’s a great analogy. Um but it’s true, you know, um we we we if we’re not looking at the bigger picture, we’re we’re just kind of firefighting rather than actually getting to the root cause of the problem. Um, so that makes so much sense. I guess you’re in a really unique position, Taylor. You speak to buyer evalions every single week. Um, and you hear both the I I dare I say there’s probably a lot of excitement around this sort of stuff, but but equally some corners will have healthy skepticism if we want to call it that. Um, if the context leader who’s watching this right now is looking at AI powered QA or or all the conversation analytics over the next sort of 12 months or say from your perspective, what should they be asking the vendors before they buy and what are the things that they can do early to make sure they make a more about success of it?
Yeah, I think they should listen in the conversations and maybe play like a little buzzword bingo. Um, so we hear a lot of we analyze a 100% of conversations, AI powered insights, real time, automatic sentiment, AI scorecards, which is great. None of these are bad, but these are kind of table stakes. This is the baseline right now, and it’s not really a differentiator. So, I think people should be looking around like what happens after that. So, are the categories building themselves from your conversations or is someone setting them up once and this is going to be um a stagnant scorecard? Can you trace the answers back to real calls? Um are we closing that information gap? And is it actually going to improve the agents performance next week or is this just a better dashboard? So, I would think uh I I would recommend thinking about not necessarily like the now, but what is it going to look like next Monday when you log on? Um what is it going to tell you about your agent performance, about your contact center health, and about your day-to-day? How can we improve the agents and uh focus less on just scoring their performance?
That’s really clear. That’s I I like that. Um I think you’ve Taylor you’ve done a really great job and I could speak to you all day about this. So um thanks thanks for that thanks for sharing. Um for me what comes through very strongly from this conversation is that analyzing every conversation is is is really only the start even though a lot of people won’t be on that journey yet. Um that for when people do get on that journey that value is going to come when those insights help supervisors to coach better. uh that’ll lead to better development of agents, improve confidence, uh and businesses that act quicker. Um you know, in order to make customers, you know, get what they need as well. So, and before we we wrap up there because we we pretty much run out of time, Taylor, for anyone who’s watching this and they might want to learn more about UTE’s approach to conversation analytics and quality management, where should they go next?
Absolutely. They can find us online at ujet.cx. CX on LinkedIn at.cx and on Tik Tok at.cx.
Very good. Any channel you want. So, uh, perfect. That’s great. And of course, you can discover more articles, resources, and videos just like this one at cxday.com. Uh, but for now, that’s all we’ve got time for. So, uh, thanks again to Taylor at Eugjet for sharing. I’m Rob Wilkinson. Thanks very much for watching.
