Episode Transcript
[00:00:23] Speaker A: Welcome everybody to our latest Metrosite episode. Today we're going to talk a little bit about the contact center space and management of the contact center. And you know, certainly one of the things that I see is that contact leaders and technology support teams as well have to really work together to optimize technology behind customer interactions to ensure that things like voice and video quality exceed expectations. And sometimes they don't. Let's be honest, when customers have a poor experience, it definitely impacts company's ability to serve them, it impacts their brand, it impacts their reputation, even revenue or employee satisfaction. There's a lot of negative that can come when you don't have a good customer experience, as we all know. So today I am delighted to be joined by Ross Williams, who is the chief product officer at Versailles. And we're going to talk about how management platforms can help improve the quality of customer interactions. So, Ross, welcome.
[00:01:24] Speaker B: Thanks, Robin. It's great to talk to you again.
[00:01:28] Speaker A: Yes. So to start, just give us all some background on yourself and your role at Versailles.
[00:01:36] Speaker B: Right, so, chief product officer and co founder of the business. So Versailles been around almost a decade now and we are a spin out or spin off of an integrator here in the Australia New Zealand marketplace. So engineer at heart, I guess. And that's my background.
And yes, just recently, in the last 18 months or so, chief product officer.
[00:02:03] Speaker A: So always innovating on new products and new ways to basically use technology to help solve business problems, maybe create some new opportunities as well.
So to me, we've all seen that it's pretty clear that agents are going to continue working from their home offices, either full or part time. So let's start there where I certainly hear quite a bit that CX leaders are struggling a bit to support these remote workers from a performance standpoint.
Is your headset working? Is your network working? Do you have too many apps running in the background? So quality is suffering? There could be all sorts of reasons that companies have these remote agents who just don't have the best performance as far as their technology goes. So I want to understand what you're seeing with your customers, your prospective customers, even as to why that's happening. What are some of the biggest causes that you're seeing based on conversations as well as your own data?
[00:03:05] Speaker B: Yeah, as you say, the work from home phenomenon is here to stay.
I saw a study on the weekend where two thirds of agents would actually leave a business if they didn't have a work from home as part of the business.
Lots of good reasons. They're saving 4 hours a week, which they are able to spend with friends and family. And there's the monetary side as well. So I was interested to see that agents are saving on average $13,000 a year by staying home, so not having to travel, pay for parking and eating in. So lots of good reasons why people want to do that and are going to continue.
The tough thing is, from a technology perspective is we know that contact centers have always been really metrics driven. So the average speed of answer, average talk time, first call resolution, those sorts of stats are there. And when agents moved home, that sort of visibility remained because it was part of the contact center application. But the part that went missing was whether it teams ran the networks, you know, managed everything, they had dashboards and lots of a practice around the visibility of the technology. So they ensured that the customer interaction was of a high standard. When the agents moved home, the IT teams lost visibility of that network, of course, because the enterprise network is largely no longer used and it becomes the home network, the home setup.
And we did a bit of a study when I first started in this role as to the top ten causes of poor quality interactions for at home workers.
All ten of them are outside the visibility of most IT teams, which is kind of ironic.
So, you know, new tool sets, new management capability is required to be able to see that full end to end interaction, to see what's going on, rather than relying on the agent to do their own troubleshooting and fix their own networks.
[00:05:09] Speaker A: So what are some of the ten, I say all ten of them. What are some of the things that you saw in that study? I'm just curious.
[00:05:17] Speaker B: The performance of the local workstation itself, process memory disk applications are running, is there, resources are up for contention, is there problems there? And of course, the ISP itself, not only up and download speed, but latency of the ISP and how well that network performs. You know, there's a bunch of them that all kind of related to that sort of stuff.
[00:05:44] Speaker A: I know I used to have problems with my, you know, just my performance on, and I always thought it was, I always thought it was my Internet access, and I learned that it was actually a statistical application I had running in the background. Whenever I had that running, I get the choppy video, and when I turn that application off, oh, miraculously, my Internet improved, my Internet performance improved, you know, so it really wasn't my Internet provider, was the fact that I was running this app, and it was just happenstance that I noticed it. Otherwise I probably would still be thinking an Internet issue. So that type of thing, you just can't tell.
I've talked to you in the past about how you guys are now integrating with Jabra's headsets.
Jabra's headset can integrate with the Versailles service management platform.
Talk to us a little bit about how and what that does, because I would think that if you can integrate with what's happening all the way to the agent's headset, you can probably get some insight as to what's causing some problems might be happening.
[00:06:48] Speaker B: Absolutely. And that headset integration that gave us like the last yard, if you like, in the whole sort of end to end picture that we're now capturing.
Agents prefer to have to be untethered from their workstation if they can. So Bluetooth headsets are really popular.
Ironically, Bluetooth runs in the 2.4 GHz spectrum, so that's the same as Wi Fi. So impacts to voice quality or video quality tend to be cumulative. So you get the effects of the ISP, you get the effects of perhaps running an application like you have that's, you know, taking up that resources on the hardware itself, you know, and then you add to that the Bluetooth link quality, which we're now capturing, you know, so that full end to end picture is really, really important. And that last yard, you know, with our friends at Jabra, has enabled us to capture a whole lot of analytics that, you know, frankly went beyond what we were thinking, you know, even to the point of boom arm position, you know, and which is a classic one for agents, you've got to have that boom positioned correctly. That sort of thing is now on visibility for it and management teams.
[00:08:04] Speaker A: So how does that play out in your tool then? I mean, is it just integrated in the same, you know, screen? Basically.
Does it proactively give recommendations like, hey, sound quality is bad here, do X, Y or z? Or how does that, how does it play into the whole platform?
[00:08:22] Speaker B: Yes, it's good question.
We try to tie everything back to it in a particular interaction. So for the likes of a Conexcenter platform, say, for example, Genesis Cloud, we will capture the interaction itself out of the Genesis cloud, but we will add on to that sort of all the trace information that you might get from. If you're using BYOC, you bring your own carrier and you've got your own session board or controller. We bring that in, tie it together with workstation performance and the audio device itself. So when you use VSM and you look at an interaction, you see the whole thing all the way from the microphone and speaker on the headset, all the way through the cloud provider, out to the PS ten with a call, exits. And that's all tied together in a single interaction. So at the time, you can see if the boom arm was mispositioned.
[00:09:20] Speaker A: Yeah.
And as you see customers using this, are they really acting on it? They're seeing this data, are they actually making changes based on it?
[00:09:30] Speaker B: Yeah, absolutely. Absolutely. It was a classic example. I was involved in one recently, similar along the lines you just mentioned, Robyn, where the user thought their Internet connection was the problem, and they'd suffered, suffered in silence for a long period of time. And after they had deployed VSM everywhere, which gives them the insight into that end user workstation, the problem turned out to be that they were running applications, they were running out of memory, so they're running memory at too high level. And that simple change was addition of memory to that device. And that problem is now solved permanently.
[00:10:09] Speaker A: Oh, that's awesome.
So when you think about what you do as chief product officer, you're looking at all different ways you can improve this performance.
How do you find the companies can see that ultimate 360 degree view, that 360 degree visibility? I guess I would say with applications and network devices and rooms, there's all these different pieces you've got to pull in where you ultimately really see measurable improvements in CX.
[00:10:39] Speaker B: The key is to have all of the devices along the path under management. And a key development for us during COVID was to create VSM everywhere, which gives visibility into those endpoints. So the end user workstation, the rooms devices themselves, and then tying that back, as I said before, tying that back to an interaction. So an agent would typically report a problem, they'd say who they were talking to in the date and time.
And while it's technically possible for skilled engineers to piece that whole picture together and go through logs and pull all the files out and look at them, and a lot of manual labour from a very skilled person to piece together that full end to end interaction. We've automated that whole thing with VSM and VSM everywhere, so a regular tier one, tier two, service desk person can now act like a tier three. So all the information is brought forward and surfaced, and using our machine learning, learning algorithms, we surface the, the likely causes of a poor interaction. So sometimes it's obvious, you know, in which case we'd have like a red triangle with an explanation mark on it saying, hey, this is almost certainly the cause of your problem, and sometimes it might be amber, but we'll highlight those you know, those potential causes just to simplify it and to reduce the complexity and the skill level required to troubleshoot these problems.
[00:12:05] Speaker A: I mean, do you think that, do you already, or do you think that at some point you'll be moving into like a self service mode for some of this at least, like some of the simpler, you know, maybe causes for some of the problems?
[00:12:16] Speaker B: Yeah, absolutely. So while making that information available to IT teams and then sort of summarized rolled up information available to management, really important that you surface that information for the end users, as you say. So that's an area we're working on right now. So those ten most common causes, if you like, will surface them for the agent. So, you know, they can tell ahead of time or in real time when they're on a call, if the boom arm's not positioned well, if they're getting interference on their Bluetooth connection, you know, that, you know, they might be able to hear the external customer perfectly well, but it's not the same in reverse. So those sorts of things will be surfaced for the end user so they can proactively take steps themselves.
[00:12:59] Speaker A: Yeah, and that really does make a difference from a customer perspective because their overall interaction experience is going to be better at that point.
So we certainly hear a lot about non voice digital channels all the time. I mean, one of the big things in the news lately is how, you know, frontier airlines, like, got rid of all voice. I mean, I'm holding firm that these digital channels will never overtake voice and video. I think those channels are here to stay.
My research tells me that I see that not only do voice and video typically result in higher CSAT, they also produce kind of a crazy high number of data points, just sort of flying under the radar data points about the performance of the call. Now, AI obviously has to be a part of that analysis in order to just even determine what the insights are and to automate what you can do with them. Automate workflows that can come out of that. And ultimately, I think you can improve things like your agent or employee productivity, even some security issues. I wanted to ask you how Versailles helps in this area, because I know you're doing a lot in the area of AI and just trying to help in what I mentioned with the voice in digital channels and all you can do there.
[00:14:16] Speaker B: Yeah, and I agree with you completely, Robin. Voice and video is here to stay. There's no question about that. And the conversations, let's face it, the conversations that really matter when someone calls customer service because something's gone wrong or they call their bank because there's an issue. You can't have a complex interaction with your provider via anything other than voice unless you've got, like, two years to solve your problem.
So those conversations, those voice and video conversations, they really matter.
And if the interaction with the agent is good, you build brand reputation, you build loyalty from that customer. They have a great experience. Their problem's resolved.
If it's not, if that interaction isn't great, it can have exactly the opposite effect. These voice interactions are really important, really important.
And you did write about the amount of data that's generated by these platforms. Now the cloud platforms, session border controllers, and now the end user workstations, the headsets themselves, intelligent headsets, they're all generating huge amounts of data. So what we do is we consume all of those.
We're really hungry for all of that. Any new data source that we can get our hands on, we love to grab it and bring it in, put it into the mix, tie it all together so it's related back to interactions.
But the sheer volumes of that we've seen a typical interaction could generate 15,000 to 20,000 data points.
And that engineering background I've come from for myself. I mean, I've had poured over logs. You know, I'm old enough that they were, they were in hexadecimal when I started, you know, poured over that sort of thing, looking for patterns. And it's really a simple matter of, for engineers is to find a pattern. What pattern of data points equals a good interaction, and what pattern of data points equals a bad one? And when you can do that across hundreds of millions of transactions, and we're looking at around about 600 million transactions a month on our platform, those patterns, those patterns around what constitutes good and bad, they start to pop out. And we've got some tremendous insight around those common causes, and we use our Microsoft neural networks to do this. There's no rocket science behind it. It's really just a simple matter of surfacing that and making the information available for those who need to consume it. So engineers, or even, you know, if they're troubleshooting something, or even better still, is proactively alerting, saying, hey, you know, we're noticing these patents. There is going to be a problem emerging for this and getting, getting on the front foot.
[00:17:08] Speaker A: Yeah. So it really does. I think more than anything, it helps people be more proactive to avoid those problems from happening. Like, if, you know, ahead of time, okay, this set of criteria or this set of issues or circumstances is going to create a good customer experience versus bad. You can kind of try and proactively make sure that what you're doing is sitting over here in that good category rather than the bad.
Maybe. I don't know if you do or don't. I know you're looking at so much data, but do you have any specific examples of, you know, how companies have used this capability that you provide by looking over all these data points to make a change or make a difference.
[00:17:45] Speaker B: Quality of their interaction across those hundreds of millions of data points. And it's very common that people use Wifi in their homes. Not a lot of homes have the workstation right next to the router, so you can plug in on a wide connection. A wide connection is always going to be better, and we all recommend that if you can. The reality is, for most people, they can't.
We've got some really good examples around patents we picked up, and there is a distinct pattern between the signal to noise ratio or link quality that you see on a Wi Fi system and the received level of the wifi signal itself. So just those two in combination, in isolation, they can cause a problem, but at a particular point, they definitely cause a problem. So, you know, we've had companies that use their insight just to coach, you know, coach people how to tune their home environment a little bit better. And believe it or not, a simple, simple remedy for a lot of places is to put your router as high as you can in your house, you know, so if that's up high, it's less likely to, you know, sort of have stuff in the way that that's going to interfere. And that insight we've developed just by seeing those hundreds of millions of interactions, seeing what the pattern is around that. And it is very, very clear. That one is very clear.
[00:19:15] Speaker A: I'm sure you're learning new things every day as you look at all this data, new patterns that emerge. So just really cool. And I think really great value to companies who are using your platform.
I appreciate you taking up the time. I could talk about this for a long time because I'm a data geek as well, and I just love looking at data and just the, it's almost like a puzzle or like artwork in some ways, like the stories that develop out of just looking at data and all you can glean from that. So I really thank you very much for taking out the time today. I thought this was a great discussion. And just thanks for sharing your knowledge on CX management insights and how you're using AI. I think that companies who are having issues in this area with their CX performance should definitely give, give you a call, talk to you.
You can tell them all the things that are happening during that call and what's going right and wrong.
[00:20:09] Speaker B: Right. Yeah. It's a fascinating area. I mean, and to me, Robin, you got a lot of data and complex stuff, but the resolutions are often really simple. It gets boiled back to something really simple, like the location of your router, you know, and that gives me a lot of pleasure, actually, when you see such a simple answer to a complex problem that's been a mystery for a long time.
[00:20:33] Speaker A: Absolutely.
All right. Well, I want to thank all of our viewers and listeners today, and thanks again to Ross for joining us. And I hope everybody has a great day.
