Busting Burnout: How you can use AI to predict and dramatically reduce contact center attrition

Hi, everyone. This is Vicky Harold, executive director of SWPP. And welcome to our sponsor webinar today. We are so excited to have interim with us today, and they are gonna give some great information on busting burnout and how we can use AI to predict and dramatically reduce contact center attrition. And I know there’s gonna be lots of great tips and tricks in here that you’re gonna wanna make sure that you check out once we get started with this webinar. We have two great presenters with us today. We have Ted Lango, and we have Chris Busby from Intradiem, and we’re happy to have them with us. Wanna make sure that everybody knows how to interact with them when you wanna they ask you to put any information in the chat or they want some feedback from you, we want to use the chat. If you look on the bottom right hand side of your screen, and you pull up that chat and make sure it says send to everyone. And so if you wanna go ahead and pull that up and put your chat in there. It looks like Minerva has already told us that it is evening there, and she is in the Philippines. So if you would put your chat in there and let us know where you’re listening from. That would be great to make sure that everybody knows where the chat is. So it looks like Princess is in Georgia, and Ron’s in Ronald is in Los Angeles. It’s a brisk morning in Fremont, Michigan. It is not brisk here in Nashville, but it is it is much nicer here. We’ve got we’ve got eighty degrees coming up this week all week, and that’s been a lot nicer than our nineties that we’ve had. So somebody in Texas, are y’all still having the heat wave there? Got Minnesota. I’ve got more people from the Philippines. Alright. You guys are figuring this out. Beautiful fall day, and State, New York. I wanna I wish I was there with you for that because we are not quite to fall yet here in the south. Oh, Tenson Scotland. That looks great. So it looks like everybody has found the chat. I do wanna let you know that this webinar is going to be recorded, and we’ll be sending out the recording after the event. So you can share it with anybody who happened to miss this. And so it looks like everybody is ready to go. So I’m going to turn it over to Chris and Ted and let them get started. Alright. Thanks, Vicky. Hey, everyone. My name is Chris Busby. I’m vice president of product management at Intra diem. I’ve been with Intra diem for eleven years this month. I’m responsible for our product roadmap. I’m very excited to talk about a new product that we are bringing to market today. Ted? Thanks. Thanks, Chris. Good morning, everybody. I’m Ted Lango Senior Vice President, Business Enablement at Intradiem. I’ve been with Intradiem for a year now. Prior to that, I’ve spent the previous twenty years fighting the same types of challenges that you all fight every day. Supporting workforce management organizations and contact centers as a whole. Today, we’re pretty excited to share with you of, of something that we’ve been diving into, that is a topic that I’m sure keeps a lot of people up at night, attrition and contact centers. Know, we’re not just talking about a minor inconvenience. While some companies are seeing a little relief post pandemic on attrition, Many I speak to are still experiencing yearly turnover rates that range from thirty percent to even over a hundred percent. I’m still hearing in some cases. The financial toll this takes on companies is significant as we talk to people and get ideas of what they price the replacement cost. The low end we hear is like twenty thousand per agent up to thirty five thousand. The key to, there’s a a key to dramatically reducing these staggering costs. It lies not really just in the actions that we take to support employee satisfaction, but also in prediction. We wanna share that with you. But before we do, I wanna dive in, before we dive into the topic, I really wanna pull up a poll question. We’ve got three poll questions just to get us started. Love to hear from all of you, in terms of what you’re seeing in attrition rates. As I mentioned, I’m hearing thirty all the way up to a hundred percent, but love to hear what everybody’s, seeing currently post pandemic with their attrition rates. A poll question should have appeared right now. If you can’t, answer the poll, sometimes people can’t click on it, you’re welcome to type it in chat as well. Or if you’ve got just other comments that you wanna share. And as we’re looking at, those answers, and as you guys are submitting your, your answers. Let’s dive into the real world impacts. You’ll see on the screen a dynamic calculator ticking up with the costs associated with rising attrition. You know, they encompass, you know, not just training. We know training is the core cost, but recruiting HR, IT, all significant investments. And, of course, new agents who replace experienced ones are less efficient. Further inflating our cost. That efficiency cost really can add up quickly. Just to give you a snapshot, thirty three percent agent, attrition rate and a one thousand agent operation can easily cost seven million dollars, get up to fifty percent, and you can be you know, over ten million dollars. It can be twenty, twenty five percent of your budget. So let’s take a look and see if we’ve got some answers on the poll here coming in. Yeah. And it looks like a attrition is is still still a problem for folks who are are sharing their numbers. Twenty to forty percent We still have people who are, forty to sixty percent. I don’t have anybody who’s over the sixty percent. Oh, no. Actually, a couple. Or in the sixty to eighty percent range. At the end of the day, these are expensive costs. If you get up over fifty percent, as I said, this can quickly you know, consume, twenty, twenty five percent of your entire budget itself. But what are the top reasons that your agents leave themselves as you’re selecting, As you selected the first answer, I wanna pull up a second poll question, to look at the top reasons why agents, leave today. And as you’re selecting your answer for the second pull question, let’s talk about a major culprit behind attrition, and that is employee burnout. I recently had the opportunity to co lead a breakout session in a financial sector industry at a conference. The focus on our breakout session was the future challenges and opportunities in contact centers, and attrition was a hot topic. One of the things that we heard about was an innovative approach that caught everybody’s attention that came from a credit union. They had implemented an employee battery system, a dashboard that supervisors used to gauge agent well-being. This system takes into account various metrics from attendance to our traditional KPIs and call centers, even quality of one on one interactions. It was updated regularly helping supervisors identify agents who might be on the verge of silent leaving. The concept really struck a chord with everyone and is something that we’ll expand on shortly. But let’s look at our poll results on our our second question. Just pulling that up myself right now. Ted, this is Vicky. Sorry. When I opened that first question, it opened off three of them. So everybody answered all three at once. Everybody answered all three. You’re jumping ahead of me. I well, I’m just gonna take a look at the The, the second one here, we’re seeing one on one meetings with supervisors. Oh, nope. That’s the third question. Sorry. Here. I just wanna take a look at your second. Top reasons people are leaving. So work environment, leadership, manager relationship, work life balance, stress, compensation and benefits. That’s always a big one. Found a better opportunity. Don’t know. A number of you said don’t know as well. And at the end of the day, a lot of times we don’t know. At the end, the the answers are across a number categories, which I think quite frankly is, is one of the challenges. And so a number of you have already, seen the third question here are final ones, and that is what are you taking to combat attrition? And you guys can continue to answer if the poll is still open. The question that you would have been looking at is You know, what are you doing to combat attrition? Is it more training, more, one on one meetings, wellness breaks, those types of things? And you know, as you select your answer, I’d just like to talk a little bit about some proven strategies to rejuvenate your agents. Or the human batteries if you you speed if you see a, you know, this concept as as meaningful here. On our screen, we’ve got terms like stretch, support, recognition. They sound like buzz words, but really they’re actionable steps that we can take to significantly impact agent well-being. On the right side, you’ll notice this term rest periods. This isn’t merely just a suggestion or a feel good thing. It’s backed by science. I was going through a Microsoft study that found short break can reset the brain reducing stress and enhancing focus. That that I think can be applied directly to call centers just as back to back meetings can learn to lead to burnout, so it can back to back calls. Microsoft’s research itself really showed that, you know, these micro breaks are a meaningful thing. So how do we detect when agents’ batteries need recharging? The journey to understanding agent well-being is far from straightforward. The factors contributing to burnout are really immense in fluid encompassing everything from time metrics to customer satisfaction levels. At Introdeem, we had theories that there was a signal that could be found in the data. I’d like to hand it over now to Chris to share more of those findings and the technology that we’ve built around those findings. Thank you, Chris. Ted. Yeah. Thanks, Ted. So our our team thought that there were likely signals in the data that intratum already captures through our existing product ties integrations, like, ACD and workforce management system. So we concluded that we may be well positioned to make an impact on Asian attrition with our data. So We did a few things. We created a machine learning and artificial intelligence team. We developed a machine learning model that intakes data from those existing integrations with the goal of being able to predict attrition. And I’m very excited to officially announce today our brand new burnout and attrition indicator product we’re gonna dive into and and look at a few more of the details. So let’s take a look at what’s involved, on the next slide, Ted, when predicting attrition, using machine learning. There’s a a few steps involved. In order to investigate this problem area, we we needed to capture the data that we needed to feed into a machine learning model. So first, we created data pipelines from our provider integrations and platform to a new data lake house for storage of that data, that we needed for the model to be able to investigate. And then we went through a data analysis exercise. They’re enabled us to stand up a model and make tweaks to that model, making sure that we were using the right inputs or features such that it was as predictive as possible. The next step was aggregating and presenting that data produced by the model so that it’s both useful and usable for diagnosing burnout risk at both the agent and organizational levels. And then, lastly, offering suggestions for reducing risk at various points throughout an agent’s burnout risk journey. In other words, different risk categories that we’ll talk about in a little bit mapped to different remedial actions, which themselves come in a couple different flavors, manual actions that a supervisor can take on their own, as well as automated actions that are taken by the Enterium platform. And so on the next slide, we can get into some of the details of the model itself. Intridium captures a lot of data, like billions of data points across numerous contact center technology integrations. And so we evaluated a large number of potential features again, or inputs to the model. Some near term data, some longer term data with a focus on, agent performance data, what we determined was that a combination of the two yielded the best results. And we evaluated the resulting data set over a period of several months to ensure the model was able to account for any weekly or even monthly seasonality in the data. But the bottom line is we demonstrated that Trudium successfully differentiates between high and low attrition risk agents with eighty percent accuracy. We’re super, proud of this result But let’s take a look at the actual model inputs on the next slide and the resulting burnout risk categories that we’ve been talking about. To give you a little bit of a better picture about what’s going on. So we’ve got four burnout risk categories, the result from the the machine learning model, and those include low, moderate, high, and critical risk. Those are the different categories that that the the model will classify each individual agent in as part of the daily, compute. And then the input features are the the things that are fed into the model that help us to help it to classify agents into these four buckets include average handle time, time, and after call work, time on hold, average time on call, average AHT, and average occupancy. And you see AHT on here in a few different flavors, the three that you see on the right, the three averages are actually multiple inputs per average, because we’re taking averages over multiple, historical periods like current day, one day in the past, three days in the past, five days in the past, seven days, and and up to even fifteen days in the past. So we’re providing different averages for the, same data point to allow the model to become more aware of patterns in the trending data increasing the predict prediction accuracy. So so we’re constantly looking at new categories of data that we can use to augment this model to increase the prediction accuracy. These categories of low, moderate risk, and, low, moderate high, and critical are calculated daily. At the agent level, and they represent an agent’s journey to attrition through the lens of agent burnout. And so the resulting data on you can see on this slide is displayed in a user friendly interactive dashboard. And there’s a main the main table of this dashboard displays which agents are at the highest risk of burning out based on their current day burnout risk categories assigned by the machine learning model. So it it bubbles up those agents to the top of this table, allowing supervisors to quickly understand where the highest risk is and which agents need to be interacted with. Some additional columns on that table include who their manager is, the current burnout risk category, and the number of days they’ve been in that risk category. The monthly aggregated burnout risk category for the agent, and then some supporting data around, average handle time and occupancy for each of the agents. There’s also some supporting dashboards below. The main table that show the organizational risk, across b a business unit and manager dimension. So, in the indicates there could be something unique to a particular business unit or particular manager. Causing elevated levels of burnout in those groups. So we’re trying to present not just the overall risk, but also across the organization, in those two dimensions. There’s some additional supporting dashboards that give insights into the evolution organizational burnout risk over time that you can see on the right. And there’s also filters that you can drill down to the, business unit manager, an individual agent level, to focus on the preferred level of detail there. There’s also finally some additional dashboards that give insight into individual agent performance with a focus on HD and occupancy. So, again, very, very excited to announce and share some of these details with you about Intidium’s burnout and attrition indicator product. It’s one example of utilizing data from contact center technology integrations to yield innovative results. And, I’m gonna turn it back over to Ted to talk about a few other examples. Thanks, Chris. Again, I’m I’m really excited like Chris to announce this burnout nutrition indicator. It’s one example of utilizing AI and the massive amount of data that call centers produce to address a long standing challenge across our industry So how are we doing this? Well, if you’re not familiar with our core technology, InterDium’s contact center automation is a critical component of the contact center ecosystem, reading ACDQ conditions and agent phone states in real time while simultaneously communicating with your WFM platform. Our position allows for us to create a wide range of real time automated solutions In the simplest sense, we find time, time to deliver a wide range of activities, time that without automation would otherwise be lost. It’s really best illustrated through examples. I’ll share two for those of who you who aren’t familiar with our technology. With a direct integration to the ACD and WFM, we can write rules like what you see on the screen. Longest call in queue is less than x. Total calls in queue less than y offer a surprise micro break because we also write back to the WFM system automatically, there’s no extra work for having to manually go and code this activity. Wellness breaks or other activities can be delivered directly to the agent’s desktop and provide a real time right back to the agent’s schedule. A second example is coaching. I coaching is a critical, aspect of agent development and performance in the contact center. We all know, though, a lot of times we try to preschedule coaching gets canceled in the name of service level. At the end of the day, we just don’t deliver nearly as much coaching as we’d like to. With intraday automation, supervisors can easily prioritize agents for coaching sessions based on predefined rule conditions such as their availability. The automation then sends a desktop prompt to the agent to begin the coaching session and the WFM schedule is automatically updated for that coaching segment. This streamlines the coaching process making it more efficient and effective contact centers can improve agent performance and ultimately enhance customer experience by using these types of automations. Our automation solutions already provide our customers, tremendous opportunities to save money significantly enhance both customer and employee experience, but by integrating AI into this automation framework, we can harness the power of data analytics to proactively identify agents at risk of burnout. In doing so, we’re not optimizing your operation, we’re really investing in the well-being of your most valuable asset, your people. We value your insights. We regularly leverage feedback from our customers and the community to show help shape our products and services. So we’d love to now take a few minutes and and, field any questions that you may have around what we shared again. In chat. If you wanna just drop a question or a comment, we’ll take whatever time we have left and as many questions as we have. To try to expand on any of these concepts that Chris and I have shared today. I’m going over to the chat to see if we’ve got, we’ve got a Yeah. There’s a couple question or actually, there’s one question that came in from Ricky around. Is this ranking of burnout risks set based on averages for the business? Or a baseline manually established or just averages across all the business units. What is the baseline establishing based on to see where the agents rank? And so My response in the chat was it’s it is based on baseline data for each individual customer at the business unit manager group level, so it’s not just an average across all business units. It is not yet taking call type into account. That’s one of the additional data points that we’re looking to add. But hopefully that answers that question for you, Ricky. And we’ve got another question, Chris. Would you differentiate burn rate by line of business or each program? It’s just differentiated by if I understand the question, it’s just differentiated by line of business right now or business unit. If if program aligns to call type in the con in this context, it’s not currently taken into account in the model, but it’s something that we’re looking Other questions out there, feel free to drop them into chat. We’re pretty excited about this product. It’s evolving. We’re just releasing it now. And as Chris said, lots of things that we wanna take feedback on to enhance it. Let’s see. Mark has a comment, org is currently looking to move away from our current on prem telecom stack to move to a CCAS solution in the next year or so. Will your tool support multiple CCAS solution since we don’t know who will be selecting yet yet. And I think the answer to that, yes, but Chris, I’ll let you chime in as well. Yeah. That’s exactly right, Ted. So That’s one of the things we love about this is we’re already connected to the vast majority of CCAS environments. We’ve spent the last two to three years making sure that our pro our product because we integrate to workforce management and ACD, so, so heavily, is able to support the broad customer move from premise based technology to contact center outs as a service technology. So we have, again, set up almost all of the major integrations there or maybe all of them, probably all of the major ones. There are a few minor ones that we still have left. But, but, yes, we we support a vast number of CCAS integrations with this product. Alright. Thanks, Chris. Another question, is there reporting over time that summarizes, identifies the burn rate by each leader, in other words, does this identify poor leaders? The the intent of one of the dashboards is is really to group the burnout risk by manager group. It is a daily snapshot on the dashboard, but there is there are export capabilities from the report so that you can manage and you can export that data, either on a daily or weekly basis to see how that data related to the manager, grouping is changing over time. I love this next question. Would it also look at voice modulations to assess stress levels and I I’ve got some thoughts on that, but go ahead, Chris. It doesn’t today. However, one of the areas that we’re looking at it, you know, integrating to additional technology is speech analytics. So once we make some progress, make, you know, integrating to those speech analytics systems and use we could use that data as a, you know, that additional data that I mentioned earlier and augmenting the model just to make it more predictive. And in in that case, we could bring that in, but it’s not something that we do today. Right. And then those types of things are what we wanna hear from the community and our customers about. It’s what other interesting theories do you have that we potentially ingest into the AI models to further refine them. Great question. Next question is the model taking into consideration any tenure component from the agent population or purely based on data points described through the slide. It’s a great question. I’m glad this was asked because it gives me an opportunity to talk about this being a very customer, a kind of a partner based customer solution So each of the customers that we’ve engaged with are providing us with additional data, things like reason codes for attrition, ten year, CCAS scores additional data that will allow us to, make the model more predictive. So while the baseline model itself doesn’t doesn’t take that as input, we can really bring in any piece of data that a customer is able to share with us, and then run it through the model under and if it’s increasing the predictive abilities of the model and then and then set up data pipelines to make sure we’re continuously getting that and feeding it into the model so that it is more accurate or more predictive going forward. So I think any most any type of data, we can look at and and establish whether whether or not it’s helpful or not. Things like tenure? Absolutely very helpful. And I just pulled back the slide in case people have a lot of the questions are kind of drilling down into this, which is great. Next question, as a comment, we have Genesis cloud CX platform. I’m assuming we’re good with that. We’re good with that. We’ve got the ACD and workforce management are actually WEM components of that. Yep. How do you address, a contact center’s agent concern about the level of monitoring required with this tool? That’s a great question. That is a great question. We have always positioned our solutions as agent assistance. And so, I think we would continue here. I think one of the things is that the the actions as a result of this either, you know, the automated actions typically will go to the supervisor, and the and the agent’s not gonna be receiving any automated actions outside of things like surprise breaks or things that they would view as positive. So I think it’s it’s most of the most of that will be handled by conversations between a supervisor and agent. And I don’t think, agents will typically look at this as being something that’s invasive or or big brother type. Right. And that’s been one of our philosophies is really technology in support of the eight first, and that I was gonna say basically the same thing. I won’t won’t repeat myself there. Next question, will it track by scheduling, time of day, day of week, schedule components in in the model itself, Chris? Yeah. Great question. We do we do really believe that WFM schedules will provide additional insight and it’ll help, you know, allow us to augment the model. The current baseline model doesn’t doesn’t have workforce management schedules as an input, but the team is actually working right now on on that. And so what we’re doing is working what we’ve done is created twenty two standardized workforce management schedule categories. And then what my team does is work with, the customers that we’re in pilot with to, map all of their segment codes into those twenty two normalized categories so that the model has normalized data that they can work with. So no, actual results yet based on that data, but we feel very strongly that it will really be, huge to the model and and its ability to increase predictive abilities. Great. Thanks, Chris. Next question. Just wanting to understand the ranking a little more we looking at occupancy and indicators like hold and talk time to see how much see see how much the agent over time is taking control of the day, I e, making breaks, withhold, and slowing down volume with long talk times. Yeah. We’re looking at those components handle time. And just to give you a little bit more insight into how the machine model is, machine learning model is trained and how it works, we separate the data into two sets. There’s a training set in a testing set. And so, in the training set, we’re essentially telling the model which agents are going to a trip, and it’s looking for patterns in all the all the input features that we discussed, all the components of handle time, the averages of time on call, and averages of, average handle time, that we discussed, you know, over different periods of time, it’s looking at all of that data and then making predictions in the testing data set And that’s where we’re able to understand how accurate we are in predicting attrition. But it’s it’s a little bit more complicated than just saying, okay, we’re looking you know, discreetly at occupancy and HD. It’s really it’s really truly machine learning and artificial intelligence that’s actually happening that’s allowing us to bubble these highest risk agents to the top of this table. I hope that answers the question. If it doesn’t, I can please clarify, and I’ll be happy to to try again. Thanks, Chris. Is, our use case in deployment being piloted. What are our findings from these so far? Yeah. We’re in pilot with a couple customers. We haven’t actually kicked off those pilots yet, but looking to do that at the end of this month. So I don’t have any results the pilots yet, but we’re really excited, that we’ve been able to partner with a couple of customers to, to roll this out. Do agents have access to the tool’s data? Are they able to see how they’re categorized on the burnout scale? Agents do not have access to this data. This is just supervisors and above. Alrighty. Next question. How long are the wellness breaks and does it only happen to those who are that are identified? Good question. That’s actually, a business decision that would be made as part of, you know, designing and rolling out the, this use case. A wellness break. It it depends on if if you have an in house tool that you’re using for wellness breaks and just having a tritium deliver that, or if if your if you don’t have a wellness break tool, something like Thrive, then you can still deliver a break, you know, a timed break with inter DM just wouldn’t be that same kind of meditation or wellness component to it. So it really depends on, what tools that you have at your at your organization’s disposal and what business decisions are made as far as how long you want to give those breaks to agents, but it’s all totally customizable with an intridium. Next question is on the the prediction, side of things. Does it the eighty percent success and prediction mean that eighty percent of the agents in the high risk or critical risk category do actually a threat? Yeah. It’s a good question. It What it really means, if we go back to, the conversation around testing and training, the machine learning model, after we after we trained the model, what it means is it’s successfully predicted with eighty percent accuracy, agents who ended up quitting or a trading. So it’s there’s a little bit of nuance there, because does that mean they were in high risk or critical risk, not really? It just means we were percent successful in predicting that they were going to a trade, and they did a trip. Thanks, Chris. Currently, this is This is generally for phone agents, but, nothing in back office, chat, etcetera? Correct. This is this is for phone agents currently. Here’s a cost question. How will this be introduced? Additional costs are included in an upgrade, for those existing customers of ours? This is an it’s gonna be priced as an additional solution. It’s not something that’s included in in a platform upgrade. And the questions keep rolling in. Does the model have the ability to fold in other metrics KPIs besides traditional productivity, occupancy, HT data, things like FCR, CSAD MPS, indicators tracked at the agent level, theory being agent frustration at inability to adequately solve service customers causes frustration, attrition, etcetera. One hundred percent. The model as it exist the baseline model as it exists today does not take those into consideration, but as I mentioned, a little bit ago, When we work with a customer to embark on the sol unrolling out the solution, any types any of these types of data that that are able to be provided to intridium, we will augment the model to include them, do another data analysis exercise to make sure that we optimize the model using the data that we currently use and this new data that’s coming over. And in most cases, that’s gonna make it a better model. It’s gonna have a higher accuracy of of of predicting attrition. Thanks, Chris. What’s the I know we’ve been talking about this, the lead time lead time for a trading? How much data per agents needed to make a prediction. And I think we also might wanna just comment on, you know, expanding that lead time because we know at the end of the day, if we say, hey, they’re ready to a a trip, and there’s only three days. That’s probably not gonna be good enough to avert it. So comment on that. Yeah. I think there’s really two components to this question. As as far as the amount of data that we need to make a prediction, we typically need three months of data in order for the model to be trained well enough to make those eighty percent accurate predictions. And what we’ve seen is we’re eighty percent accurate, predicting attrition fifteen days in advance. So, to Ted’s point, one of the things that we’re looking at with adding in the workforce management schedule capabilities and then working with customers to bring in additional data is really to get that fifteen days, extend extend that runway to a month plus because we know that a couple weeks is probably not enough, to start taking meaningful action on some of the agents who are really in that critical category. And I know this is another topic that we’ve been talking a lot about. Have you tracked the success of the burnout interventions, on deflecting attrition or identity or the identified associates with the attrition risk. Sort of the Yeah. That’s that’s a great question. That is since we’re we’re about to embark on pilot a couple pilots. One of the things that we are doing as a component of those pilots is is providing a kind of a lightweight form based applications for supervisors to document the actions that they’re taking What agents are they interacting with? What burnout risk category? Is that agent in? Any additional information about the actions that they’re taking then we’re dumping all that data into a repository that will allow us to go and and look at that data historically over time to understand, okay, which of these actions that have been taken as part of the pilot are moving the needle on, you know, moving burnout risk in the right direction, from critical back back down to low. So that’s, I don’t have any out any actual data or outputs yet, but that’s something that we’re really focused on in the pilot. Cause we wanna make sure that we’re recommending the right actions to take at the right point in an agent’s burnout risk, journey. Great. Thanks, Chris. Is this available now when will this be ready? Looks like one of our final questions here. Yeah. This will be ready starting next year. Oh, we got one more question. Ethics’s questions in in certain types of insurance like auto medical age gender, for instance, are are used to inform risk. I’m sure there have been studies that show impacts of those more personal factors on attrition as you continue to improve the product, where will you draw the line? Is there boundaries squarely at using performance KPIs and not individual agent data? Yeah. That’s a great question. I, I don’t know where we’re gonna draw the line. I think probably, again, as we work in partnership with our customers, and and every time we work with a customer on this, it involves their human resources organization. I think we’re we will probably make the right decisions in partnership as to what if there’s any, you know, ethical, any ethical issues with the data that they’re providing us. I don’t think they provide it to us. So I think it’s probably customers that will end up drawing that line, in partnership with us. So I’m not super concerned. There are certain personal piece of data like tenure that have come up, that I think will be important inputs into the model, but certainly some of these are are probably not going to make it into the model just based on an agreement between customers and us. Alright. And and I I would second that. I mean, it’s a fantastic question ethics and AI, regulation that still isn’t really developed in our industry around AI is gonna be an important topic, but I think we we solve that together. You know, companies companies are well aware of where that line is, and we’ll partner with them as well. One other question that popped up in the Q and A window, is there an element of country, slash location of the associated fact of the associate factored in any way. Local markets may also, drive higher or lower attrition. We don’t, we don’t currently have that as part of the model, unless that’s naturally mapped into, like, a business unit somewhere. But, no, I we don’t separately have location as as a component or as a input feature to the model. Alright. We got one more question that just popped in. Could the behavior at two weeks, pre attrit be driven by associates already decided to with a k a short timer mentality and not necessarily, risk that can be mitigated? I think certainly possibly. For sure. I think that’s a great question and a great point. I think that’s there there’s no way to know, and there’s no way for me to know about an individual agent at this point, but I think that certainly could be a possibility. Agree. Alright. I’m not seeing any other questions. So, oh, we took a lot of them. Maybe we’ve got one more coming in. Maybe we don’t. You got it. Well Fielded a lot of questions. We we have. And and we really value the insights. And as I mentioned, we regularly leverage feedback from our customers and the community as a whole. At the end of the day, we think this is, obviously, a long standing, problem that we would love to help the industry solve, and it’ll come through partnership. But we’re on to something that, we think can really meaningful have an impact, a meaningful impact on attrition. Chris, any final thoughts from your your your end? No. I agree. I really appreciate the, the the time and the questions and insight and, really look hopefully, look forward to working with a lot of you on this. And hopefully, if you come to SWPP in March, you’ll be able to see Entredeem and, actually see the new product. Right? That’s right, Vicky. We thank you for hosting us, and, we definitely love to feel more questions at SWPP in March. Or feel free to you’ve got our links on here, and you can reach out to anyone At Interdium, we’d be happy to to share more with anyone who’s interested in learning more about this exciting product. Thank you, Vicky. Alright. And, again, this has been recorded, and we will be sending it out to all of you. So, somebody’s asking will you be sending this slides out. That’s an interim question. I’ll be sending the recording out. Probably not be sending the slides I mean, Ted, we can talk about a, you know, a subset of them if you want, but I pro I don’t think I wanna send out the slides with the dashboards. Yeah. And maybe a couple other ones. So you’ll but you will have the recording and You’ll have the recording. So you’ll have you’ll have access to see what the what the slides actually looks like. Alright. Well, thank you so much, everybody. We’ll give you back a couple of extra minutes in your day today, but thanks so much to Ted and Chris. And if you do have other questions, make sure you reach out to them. Got QR codes there on the screen, and we’ll sending out the information from the recording later. So everybody have a great day, and we’ll see you soon. Thank you. Thank you.