Streamline Your Operations with Automated Sales Software for Quality Assurance

How to create outlook signature

[Music] hello everyone I'm Alan pop Cotter and you're listening to call talk for January 24th 2024 today's topic is getting the most out of AI and automation for quality assurance and contact centers if you're listening live we invite you to be part of the show and ask questions here's how you do it you can email me at calltalk benchmarkportal tocom I want to remind everyone that all of our shows are archived and available to listen to at Benchmark portal.com any time of the day and with that I would like to introduce the host of the show Bruce [Music] Bor thank you Ellen and welcome back to call talk everyone you know AI is being discussed everywhere in the contact center space and we've been leading the charge in getting through the hype and helping call center managers with useful information on how to approach how to approach AI for their centers uh one area of great interest we found is leveraging AI in the quality assurance area to improve customer satisfaction efficiency Effectiveness time to proficiency and uh sales success in the contact center so it's important to find the right blend of people processes and automation uh in the QA function to achieve improved outcomes and we've brought in an expert on this rather complex topic for you Chad Tatton welcome to the show Chad thank you happy to be here okay great well for those who don't know him Chad Tatton is a 24-year contact center veteran owning and operating multiple contact centers uh so he's had a lot of on the ground management experience in the contact center space and in uh 2012 he co-founded call criteria a quality assurance as a service company where he currently serves as coo uh Chad is really focused on using a blend of Advanced Technologies people and processes uh to attain the optimal quality for contact center operations so starting with human-based QA he's widened his work to include automation speech analytics and artificial intelligence uh helping companies of all sorts uh to find the right blend of AI and people-based solutions as part of their quality control processes so Chad really brings us a wealth of experience and insights and we're really delighted to have him here with us today so Chad this is a huge juicy topic and it might be helpful for our listeners uh if you could set a baseline about where many organizations are starting from their Journey to Automation and AI I'm sure many of our listeners will see themselves in the description you give yeah I think there's there's people all over the gamut so yeah I think the start me quality control in the call center from the beginning is revolved around finding out what happens on the calls so that you can make improvements and adjust the performance and the customer experience that are happening on your phone calls and the methods that have been used to gather this information in in the call centers has evolved you know it's evolved over time you know from the beginning when it all started you know certainly back when call criteria began it was all about just human QA right you listen to phone calls with with ears and fill out custom scorecards and then drive that information into the management teams however however you need to and then over time you know there was an introdiction of voice analytics and voice analytics well it's not AI is you know is changed where where voice analytics is is something where you just transcribe calls into text and then you use words and phrases that within the transcript to to to identify what's happened on the call but you know the third way that that is done in voice analytics has been around for 10 12 years it's not a new thing to folks but the you know the last one is really what this topic is and and more recently over the last several years more than more than ever in the last 12 years technology has been introduced that just makes it far more capable to use technology than simple voice analytics where you're looking at a transcription using generative AI or large language models such as chat gbt through open ey or and Tropic quad or or llama from meta or Gemini Google there's you know there's there's many others they're they're coming out all the time you know while they're still quite expensive they provide the ability for a far more comprehensive analysis of what's going on in the calls and just opens up this whole new world of capabilities for call center so you know to answer your question the QC industry is just evolving quickly for companies to as fast as possible leverage these new capabilities with the primary goal of really replacing these more expensive human type of interactions that they've always had to have are human reviews including voice analytics which it which which requires a lot of human engagement too so there's companies that are popping up all the time that they're claiming to provide you know to provide this ability and a range of solutions and call SC cers centers are just scrambling you know to get educated about these rapid you know advancing Technologies you know to sort out what's real what's not deploy the latest and the greatest Solutions and environment that's just really complex and varying with wildly very points of view so you know unfortunately many are jumping in without the proper understanding and the full picture of what it requires to make it work and then they're struggling to realize the results that they're seeking so you know I think that's kind of where we're at in relation to the technology piece and QC no that's a very good uh framework there and you know it gives a good reason why we need people like you uh and your company to be able to uh sort things out and make sure that there's uh some there's dep right dep to the analysis that goes on why the things you've mentioned to us too is the challenges with quote out of the box unquote Automation and AI implementations uh can you elaborate on that and what you are seeing I think our listeners would really like to hear that yeah sure you so as mentioned prior um you know keth QC at its core is just about learning what happened on the phone calls and there's really too many there's really two primary ways there's lots of subcategories but there's two primary ways that this is being done through technology and that's voice Analytics AI while both are uh certainly capable of providing really accurate insights you know if almost completely misunderstood what needs to go in to to to making it as accurate as it needs to be and the level of knowledge and human resources that need to go into it to get it to the point where where it's producing accurate enough results to to make an impact on your on your business you you can't just turn on a solution of any type and have a an acceptable level of accuracy right out of right out of the box you have to go through the process of of of trial and error you you have to learn what's working and then you use that information and you expand the syntax in the case of voice analytics so that it has a a larger search criteria or in the case of mL of llm scoring or generative AI you need to optimize the prompts that you're using to generate the results or or fine-tune the model that that is that is being used with with additional content that that you can aim at training the model to have have a greater understanding of the topic so that it's more accurate so we see companies that jump into a solution without this without this understanding you know it's an out-of-the-box solution that they expect they can just turn on they don't have the understanding and in the result is that they end up with a great deal of positive of false positives and false negatives and in many cases it makes it almost impossible for them to drive the behavior or improve their customer outcomes at all you know they need to have processes aimed at measuring the results that are coming out of whatever the analytics are calibrating them verifying the results so that you can have accurate findings so that you have what you need to do those optimizations make the changes that you need so that eventually it'll get where where it needs to be and also so that it has some value along the path right so you're not just spending money for months and months and months and months but it's just not quite good enough to get to get anywhere so that's happening a lot and you know and over time if you do these things and you do them right you can get to an accept acceptable level of action but you just can't get there without investing into all of these things so you know so back to your question I mean the biggest challenge that companies have is that they're not getting the results that they expect because they don't have the knowledge nor do they have all the resources that they need to get the the program to a point where where they thought it was right out of the gate yeah no so important so important one of the things that U we're uh doing as well is offering giving an offering for people who want to uh figure out what their maturity level is okay that in a in a technological sense obviously not in a personal sense uh in terms of being able to confront um Ai and so that how they can in fact make sure that when they get into something they do have a Long View of it and they don't get into the kind of problem that you're talking about where they have an un a feeling that you know just pull something out of the box plug it in and it's going to work so I think that's a really important uh message that you were just delivering there so and Chad maybe I could press down on the impact of unverified results from Automation and AI um if you could just talk about what the these unverified results are to for definitional purposes and are unverified results sometimes okay again I think our listeners would be interested to hear from you on this yeah you know I think the impact from inaccurate or or unverified results is is fairly self-explanatory you know on the agent side you need Buy in from your from your agents about the assessments that are being made on their calls you know if you're going to hold them accountable for errors that are not really errors you're going to lose credibility and it happens fast and then they're going to marginalize all of your efforts to correct action moving forward and it could be catastrophic to a QC program we've seen it over and over again and so you know if you're a third party call center and you have reporting responsibilities to your client you know the same logic applies but it can often have like far greater you know significant consequences so so that all being said you know inaccurate or unverified results are are results where you're trying to find out if something happened on the phone call and the the technology is saying this didn't happen but in reality it did because it's just in it's just not developed enough it's not optimized to get it right or it's telling you that something did did happen that didn't so that's what unverified results are but that being said it's you know unverified or inaccurate results but that being said that that doesn't mean that there's not an acceptable level of inacurate results they're kind of necessary to get to where you need to be but the great thing about automation is automation can often provide you a vehicle unlike before that allows you to get a really broad coverage of the call volume right so you can in a lot of cases you can assess 100% of the call volume and if you're scoring a large number of those calls and you're producing you know overall metrics that that aim at driving driving change sometimes it's okay if 25% of your of your QA answers that you're getting are wrong you know all of the calls are influenced the same way so when you're comparing One agent to another or one supervisor team to another campaign or or any other grouping for that matter the stat rank amongst the the folks that are in that same category that are doing the same calls is is is often times 100% accurate uh or not 100% accurate but it's just as accurate as it would be if you were scoring all of the calls 100% right so so it it certainly placs the value so if you're not looking at individual calls to drive your processes you can get away without going through all of the optimization and high and and all the investment that goes into extreme high levels of accuracy uh in which case even pure Automation and sometimes even out of the box Solutions you know back to that question might be good enough it's just a question of of what you're using the information right okay and really uh again just as a a background issue the whole thing about AI in Automation in this is that it brings in the possibility to have this huge amount of data that comes from listening to all these calls and then you know having the AI find patterns that are in some cases kind of obvious but in some cases are not obvious but can be really useful and that's where a lot of managers I think have their open to the power of uh what AI can offer here and sometimes those things are also found Through The Human Side of doing things so don't don't underestimate that as well I mean the loop that can go from uh human um human QA uh back to management in terms of uh items that can be good for training and for coaching and things like that uh but certainly with the AI with the huge amount of data that goes into that finding those patterns and then making them useful in terms of training and in terms of uh improving both the AI system itself and the agents who are interacting with it very exciting stuff very exciting stuff so uh did you want to I have another question for you here but did you want to elaborate on that any further Chad no I think that's I think that you you you nailed it on the on the on the head it's you know it it it certainly is something that that varies a lot there's a lot of variables that come into how accurate it can be and and and and what kinds of results that that you're getting and so you know there's there's a lot of other that we'll get into here that will that I think we we'll expand into that a little further Bingo okay so for those organizations that are looking for the most accurate outputs from their Automation and their AI uh what are some of the things that our audience should think about to achieve higher accuracy yeah that's a good question it's it's broad right there's a lot of things and I think that I can break it down into um into specific things that don't it's not just speaking to accuracy but but more importantly just overall success in deploying technology is part of your of your QA program as a whole right so the first thing and a really important thing is your transcription accuracy right because at the end of the day if you're using technology whether it's voice analytics or it's generative AI llm scoring of some sort at the end of the day it's analyzing the transcript and then it's coming with results based on the transcript so if your transcript is not accurate of your call recording you're kind of dead before you get started it doesn't have to be perfect but it needs to be at least at an acceptable level so when you're considering that starting point it's important to have good quality audio so that that is either good call good quality recordings because they're going to be P pushed into some sort of a process after the calls are done or if using some sort of live live QA um uh assessments you need to have really good clean audio so that the results can can accurate you also need to pick a correct transcription partner there's a lot of them out there right you know folks on this call may have heard of a bunch of them there's some really good ones there's some that aren't so good there's some that are really accurate but they have problems with speed there's some that are really fast but they're not that accurate there's some that have limited capacity they do a great job but they just can only do so much right and it's beyond you know you're beyond what they can even do so it's important that you have the right transcription partner in general and then also it's important it's important to have dual channel it's not necessary you can get away without having dual Channel audio but it's in it helps the whole process because you can easily tell who's talking doesn't really make sense to assess how your customer is doing on the phone call right so if you can break those out it makes it a lot better so so transcription is key the second thing I think that I would say is that you need to have the the right technology partner who lives in the world of of call assessment right because you need to have someone that knows what you don't know it's really complicated and it changes every single day I mean we get updates every day on wow this thing that we're doing today is different tomorrow we can do this where we couldn't do that so it's important to have someone that that knows what you don't know I think and that can really run run ahead of you in that in that area you need to have someone that can provide all the different types of scoring because there's you know there's multiple things we've talked about voice analytics and Ai and all the subcategories so it's important that you're working with some sort of a partner or a platform that has all the different types of scoring and not just one you don't want a software provider who will you know just hand you the keys and say figure it out they may have limited access to the best stuff that's available and or they're using just a a single a single llm or or or some other some other limit that causes you problem so so it's important that you have the right the right partner that can really give you the education and give you access to all the things that's there um the other thing is that's important is is is is choosing the correct method of scoring yep okay hello are you still there let's see Allan are we still there okay all right so uh we're gonna just continue on here and uh we're going to sort of think about some of the items that uh that Chad was talking about uh because at the end of the day he was talking about a lot of the challenges that contact center managers face every day and uh contact the me the challeng is also that the contact centers uh have to face with regard specifically to AI Automation and um what we're doing here with uh quality assurance one of the things that I think is really important to know is is how those challenges can be overcome how it is that you as contact center managers can actually face those challenges and figure out the best way to do whatever is necessary for your Center to get the best result possible so we do have uh situations where the uh way to learn about what your QA is like at the beginning is to do human QA sometimes outsourced so have people listen to that and take in that information get yourself a good base of information of um what QA can be done to help out also what you can do with that information to help out in terms of uh the training you give to your agents as well as the coaching you give to your agents and one of the things that we have seen is is that if your agents are able to be coached trained and coached with the assistance of video learning that that can help out a lot and why is that because the video learning is third party uh and it's also engaging and it helps to bring across the point that the trainer or the coach is trying to bring across in a more entertaining in a more forceful way so those are things that can help out with QA why because you're looking for a change of behavior with your QA function you want the uh agents to be able to uh actually come out of a session and act differently act better do a more a superior service for the customer and so these are this is one of the things that we found is that uh by having uh video training that complements the verbal training and the verbal coaching can really help out a lot so let's see Alan have we've been able to uh get Chad back if not we might just wrap up the show at this point and there's been a lot of really good information there that Chad gave to us we appreciate his his coming on the show we apologize to our listeners uh for the fact that we had this technical glitch and we look forward to having you on a future version uh okay maybe he is coming on right now ing to Allan so we will here there you are okay good all right well I was just talking about the uh the CH challenges uh you were talking about what are the best ways to overcome those challenges and I talked about a number of things in your absence but uh let's hear from you about what you have to say about how to overcome those challenges and then I think we'll probably be ready toh to wrap up the show sure so yeah I'm not sure where I cut off unfortunately so um I will uh yeah I think I I can just I'll assume I think I know about where it is but if you're talking about you know your question is is what what uh what have we've been able to do or what kind of things you need do to overcome those challenges that I was talking about right so exactly because Chad those those challenges are real challenges they're they're well stated on your part and so the question that I'm sure members of our audience would have is okay well how do Chad or how do other people uh confront those challenges overcome them and actually have the kinds of results that we're all looking for when we look to the QA function so mostly it boils down to just becoming educated it's probably the part that you didn't hear about the desired outcomes that you're looking to do that you're looking to get so you know what company can do is if they Ed educate themselves or for helping them and educate themselves they can identify what the root cause of the challeng is that that they're having right because that's the key you know the key is fixing what isn't working instead of trying to fix things that aren't working so oftentimes the challenge could be um could be because the transcription is bad other times it's that you're using voice analytics but it's just too complex of a subject matter or you're using the wrong large language model needs fine- tuned so the a big part of it is making sure that you that you understand all of the things that you're using and then you've measured you know the key is to measure the the success so you can address address them so those things are primarily in the area of Technology but the biggest thing that we've been able to help folks overcome at least and I think that it's where folks see the most success is that is that human intervention you know in the process if is discussed you need to understand what's working what's not so you can optimize and you can change the things that need to be changed to to get it right so if you don't have those in place you're really going to struggle to you know to get to where you need to be and so we've been able to do its on a validation on voice analytics and large language models for our clients and then their systems and processes internally that make that a lot more manageable and then lastly is just the human resources sometimes if you have a lot of volume you need a lot of Human Resources just to validate the results on thousands and thousands of calls and so you know that's another area where where where you can seek help to to get that done so that there's certainly a lot of other things um um along the path to creating a program that works using AI or voice analytics or any of those things but you know if you're doing all the things that that we mentioned you're paying attention to the right things and you identify the root cause of where your struggles are you know you're going to stand you you can stand a benefit a lot from from the technology that is introducing itself to the call centers and and and it's rapidly advancing capabilities for sure okay versen uh wow uh these are great insights um that are listeners can think about and apply we apologize to them for the uh temporary glitch that we had on the technical side but I want to thank you Chad and our listeners for a um a great session is are there any final words that you'd like to add Chad before we turn things over to Allan to wrap things up no no I I think that this uh I think we've done a pretty good job of of covering the topic yep absolutely okay well thanks again Chad and Allan over to you thanks again to Chad and to Bruce for your insightful discussion on today's show be sure to join us next month for another great show or look at our huge selection of archive shows on Hot Topics at Benchmark portal.com where you'll find over 15 seasons of this show from all of us at Benchmark portal keep those headset steady and your fingers ready this is Alan poter signing out have a great day [Music]