Conversational Data
Release Date:
Companies often get stuck between the data gathering and discovering meaningful insights into how they can better their customer’s experience. Especially when you’re talking about qualitative data – conversational, open-ended data that can be difficult to quantify. But there are tools that will allow a CX pros to utilize the vast amounts of “conversational data” that companies collect. Host Steve Walker welcomes Matt Dixon, chief product and research officer, and Ted McKenna, SVP for product at Tethr for a discussion on how customer experience leaders can take advantage of the vast amounts of qualitative data typically collected by companies.
Ted McKenna and Matt Dixon
Tethr
Connect with Ted
Connect with Matt
Highlights
Finding the hidden meaning
Ted: “…there’s a lot of hidden meaning in the exchange. So because we’re mining both sides of the conversation, both the agent and the customer, we find that give and take the back and forth the dialog actually unearths a whole bunch of meaning of I wonder why the customer was thinking that way. I wonder how they would react to that. And so we spent a lot of time trying to understand that those types of things.”
The call center’s renaissance
Matt: “Now the call center in progressive companies has become the sort of center of the listening enterprise. This is the place where the conversation happens and it’s really cast the call center in a different light, gotten them a seat at the table. I think in many respects for these higher level product strategy and value proposition kind of conversations happening that previously they weren’t invited to. But now they sit at the very center of it.”
Transcript
CX Leader Podcast: "Conversational Data": Audio automatically transcribed by Sonix
Download the “CX Leader Podcast: "Conversational Data" audio file directly.
CX Leader Podcast: "Conversational Data": this wav audio file was automatically transcribed by Sonix with the best speech-to-text algorithms. This transcript may contain errors.
Steve:
"This call may be recorded for quality and training purposes." How many times have we heard that all those recordings are untapped goldmines of customer insight?
Matt:
We've been using text analytics, right, to understand when the customer types in. You know, here's why I gave you that NPS score, and to give a little bit of color. That's a unidirectional kind of piece of information. We find that there's a lot of richness and a lot of ebb and flow in the experience that happens in that go between, between the brand and the customer.
Steve:
Using conversational data to reduce your customer effort on this episode of The CX Leader Podcast.
Announcer:
The CX Leader Podcast with Steve Walker is produced by Walker, an experience management firm that helps our clients accelerate their XM success. You can find out more at Walkerinfo.com.
Steve:
Hello, everyone. I'm Steve Walker, host of The CX Leader Podcast and thank you for listening. On The CX Leader Podcast we explore topics and themes to help leaders like you leverage all the benefits of your customer experience and help your customers and your prospects want to do more business with you. Your customers tell you what's right and wrong with your products every day. But those perfect gems of insights are hidden somewhere in your thousands of hours of recorded customer interactions. Companies often get stuck between the data gathering and discovering meaningful insights into how they can better serve their customers experiences. Especially when you're talking about qualitative data, conversational, open ended data that can be difficult to quantify. Well, hopefully my guests on this episode will help us understand the ways possible to break through that threshold of finding actionable data among the chaos. Matt Dixon is the chief product and research officer and Ted McKenna is the senior VP for product both at Tethr. That's T.E.T.H.R. It's an enterprise listening platform that surfaces insights from customer interactions. Matt, Ted, thanks for being guests on The CX Leader Podcast.
Matt:
Thanks Steve. It's great to be with you. Thank you for inviting us.
Ted:
Yeah. Looking forward to the discussion.
Steve:
I've been aware of Tethr now for several years and it's fascinating. And my history in the business is very much the traditional market research survey research. And, you know, as one of my younger colleagues who sort of said, you know, we may not have to be doing all these ticket transactional surveys forever because there's some pretty neat technology out there. But again, just for the context for our listeners, if you guys could just give us a brief overview of of what Tethr does and then also how you arrived at Tethr and just kind of set the context for our discussion today. I think that'd be helpful.
Matt:
Yeah, I sure am happy to see. You're basically in a nutshell, what Tethr does is we take unstructured data and we help companies make sense of it. So as you said, Steve, you know, historically, the way we we've done market research or understood the customer experience has classically been through surveys. What we do is we we help companies supplement that. And as some use cases even replace the need for surveys by going to all that unstructured data, which, you know, truthfully is I think, as you kind of suggested here, has been piling up around companies for years and years and years. These are phone call recordings that we have from our call center, increasingly chat interactions that our customers might have with our folks at email exchanges, data in ticketing systems, our case manager systems like Service Cloud or Zendesk, SMS exchanges, Facebook messenger exchanges, you name it. It's we really specialize in conversational data. And so what we do, we help companies get that. They get their hands around that conversational data. We take it over to Tethr. Now, if it's an audio piece of data, like a phone call, we transcribe it. So we turn it from unstructured audio into unstructured text. Then we apply machine learning to that to that unstructured data to try to make sense of it. And what that means for for the layperson out there is we we built and we allow our customers to build what's called machine learning category. So imagine you as a company are trying to understand maybe something really simple, like how many of my customers are calling about X, X product or Y promotional offer? Well, we can teach the machine to go find and spot those things in that conversational data so we know what volume of our phone calls are about. We can also we can teach the machine to spot different types of behaviors going on. So let's imagine in a call center setting, we might want to understand which of our call center reps are using behaviors we don't want them using, like what we might call powerless to help. We've all had this experience. The call center rep says, hey, Steve, I'm sorry, there's nothing I can do to help you with that or the policy prevents me from doing the thing you're asking me to do. I'm hiding behind policy or conversely, our reps using things like advocacy. They're saying, you know, Steve, let me see if we can figure this out together and see if we get to a positive resolution here, get you back up and running. Those are the things we want our reps doing. Now, there are many different ways a person could express those things in the same ways that there are many different things a customer could express their frustration or their confusion or their dissatisfaction. And so we've taught the machine to understand those things and spot them in that unstructured text. Ultimately, we're trying to do a serve that data up in a way that is usable for CX leaders, for product leaders, for marketing leaders. And how do we take this messy thing of unstructured data but serve it up in a really structured way so that a leader or a practitioner can do something with that data and they can drive actions around it, which is Tethr and Walker both work with Qualtrics and that's a big focus for them. How do we actually help practitioners do something about those insights that they're finding either in their survey data or in their unstructured data.
Ted:
So I think what Matt's hitting on here is that you can use this type of technology to get to the types of data that you would hope to get you for surveys. But because of limitations associated with sample size or bias or things like that, it just makes it harder. There are other related use cases that get into, you know, that's something I never would have thought to survey about but I still get the insight because I can mine it in such things customers are already talking about. But there's a more fundamental thing that Matt's hitting on, which is that there's a lot of hidden meaning in the exchange. So because we're mining both sides of the conversation, both the agent and the customer, we find that give and take the back and forth the dialog actually unearths a whole bunch of meaning of I wonder why the customer was thinking that way. I wonder how they would react to that. And so we spent a lot of time trying to understand that those types of things.
Steve:
Yeah, that's really powerful. So that's what you've used the term a couple of times, I want to define it for our listeners so that they can add it to their vocabulary. But this is "conversational data."
Matt:
Yeah, that's right. That's right. This sort of the two way exchanges that happen. Right. And as Ted said, it's sort of there's a lot that happens kind of in the corners and in between in those interactions between reps and customers. You know what's interesting, Steve, is we know in in the field of kind of survey research and think about and see actually we've we've been using for a long time text analytics. Right. To understand when the customer types in, you know, here's why I gave you that NPS score. Here's why I said that my interaction with the rep wasn't satisfactory and to give a little bit of color, and that's super valuable to us as practitioners understand what we can do to get better. Right. But that's what we focus on, that's a unidirectional kind of piece of information. It's just the customer typing something in. But as Ted said, we find that there's a lot of richness and a lot of ebb and flow in the experience that happens in that go between between the rep and the customer.
Steve:
You know, I'm glad we're doing this podcast because I learned something here today already that, you know, 40 years in the research business. But we never really thought to kind of listen in to the other side of the conversation because, you know, when you go to take action, that's what good practitioners would do. They're like, why does this, you know, division have better CX results than this division? Well, it has something to do with the way they're training and interacting. So you kind of get that ethnographic type of analysis going when you have both sides of the conversation. So…
Matt:
…yeah
Steve:
…again, just because I'm still learning about Tethr and you guys are opening my minds to think about this differently, if I'm a CX pro that has not really thought about conversational data in and effort drivers, where would I start? You know, kind of how would I get going on something like this?
Ted:
Yeah, so TI's predictive algorithm we've built that works on every interaction. And what he's looking to do is to predict the customer's perceived level of effort. When we think about effort, it's pretty expansive in nature. So it is the types of things, the actions the customer has had to do along the customer journey. I'm trying to execute versions of self-service and that failed. And so now I'm talking to the agent. There are elements of the agent experience itself, which is often overlooked aspect of what can make the the experience itself easier or more difficult or literally the words the customer or the agents using and the language patterns of using and and so forth. And then there's a series of kind of emotions, you know, those of you that are familiar with the effortless experience, original research, we know that up to two thirds of effort is really about perception. And so the emotions the customer brings to that conversation and then leaves the conversation with is a very, very big component of kind of the residue left, if you will, of that. And so when you take all those things together, journey related things, baggage to bring into conversation the types of things that agents are doing and then boil it into a single score, we have a separate one that predicts the agent's contribution to that same that same notion. But it gives you a very crisp sense for an ongoing basis. You know, what's the damage done here? So if you think about the Tethr effort index across every one of our interactions, you're going to get this range of scores. Right. But if you start to think about that at scale and having that for every single interaction and combine that with all the ways in which you're looking at that effort, it becomes really powerful to be able to say, you know, for this segment of customers or for this part of my team, you know, the thing they really suffer from is digital experiences that are driving them into the call center or agents and how they're handling it. So it becomes sort of a diagnostic approach to finding those those biggest sources of effort in which you can prove for the foreseeable future.
Matt:
The one… The thing I for the CX leader who's not familiar with this technology and what it can do in the potential you of you've grown up in the world of, hey, I'm responsible for the loyalty survey of the survey. And that's kind of, you know, among other things, that's one of my main jobs here in my organization. And I've never contemplated, like I've always had in my head, that, yeah, they they do speech analytic stuff in the call center, but I never knew as a CX leader that would be something I would, you know, take advantage of. The effort index is usually, as I says, the pointy end of the spear, because what it really is, is a predicting the survey score. So we built it as a effectively a predictive customer effort score. So imagine every single phone call, every single chat interaction getting a score, which effectively is the machine saying here's the level of effort, the customer experience. But without you having to rely on the customer filling out a survey to tell you what the level of effort was. So bigger sample, as Ted said, really are able to get quite a depth around what was driving that level of effort. So the practitioner would do something about it. But I find that that's the the moment where the CX leader gets it because they get the survey. And so we've kind of made this as, I imagine, a certain machine predicting what the survey score would have been giving you a way bigger sample, a much richer data to go drive action together.
Ted:
There was a moment in time when Matt tried to call the survey-less survey but…
Matt:
This is true…
Ted:
…dispensed with that.
Matt:
By… The branding folks killed that one. [laughing]
Steve:
Well, you know, we for years have talked about how the customer relationship survey would be a predictor of future financial performance. And over the years, we've proven that in many cases. But really, you've got even a, you know, an earlier warning system there with this kind of technique. And you're right, you're not limited by sample size constraints at all. And, you know, another kind of old saying that pops into my head, Demming told us to love variation in our data and hated in our process. And, you know, when you can aggregate all that conversational data, you know, you're talking about huge sample sizes. So even, you know, small segments at each end of the spectrum on the bell curve are still going to be, you know, significant for looking at that that set.
Matt:
You know, in our shared customers that we work with together at Qualtrics. They you know, I think one of the most powerful things…. And this is, you know, for us at Tethr is exciting because we went from kind of understanding this data and being able to dashboard and report it. But as you know, the folks at Qualtrics rightly say everyone's got dashboards and reports and scorecards now, but how do we help people actually do something about it? So one of the most exciting use cases there is this Tethr effort index. The TEI scores a zero to 10 scale. And we we found a very high correlation in the bad end, especially that's the scores below a four, that's the there's the really bad ones is the high effort, lots of friction, lots of customer frustration. And so what if we as a practitioner, without us having to do or touch anything, we could create a loop, a workflow whereby every Tethr detects a sub-four, below a four Tethr effort index scored interaction, that that customer gets an automatic survey or outreach from Qualtrics that's tailored to what specifically the customer was complaining about. So could be, hey, Mr. Walker, I understand you had a good interaction with our call center earlier today. You were talking about your frustration with our app or our website. At Acme Company we really aspire to be a leader in self-service, in the digital experience we deliver. Clearly, we let you down. Would you mind filling out a short story and telling us what we can do to improve here? Because we really want to make sure we hear you. But that survey gets a very high response rate and really rich for data. Right? So that's a really great use for that survey outreach and a really nice closed loop kind of action that we can enable using that algorithm on the front end.
Steve:
And I know you guys work across a bunch of industries and there's probably nuances around industries, but are there some sort of telltale signs or are there things that you could even just look at to say, you know, this is a big detractor? This is where it takes way too much effort or maybe conversely, you know, what are the characteristics of an organization that's already driving down that the customer effort?
Matt:
You know, so I think that one of the first places we go to and Ted, you can you can speak about this in more detail, but is the the agent side of things. So Ted said, before you know, the original research, we found that, oddly enough, you know, sometimes when I say reduce customer effort to a practitioner or CX leader, they immediately run to, you know, technology investments and Six Sigma process engineering and big, big, heavy duty projects. But it turns out that most of that effort in the eyes of the customer is efforts in the eye of the beholder. So it's a lot about not what I had to do. It's more about how I felt, about what I had to do in so much of managing that. That side, the perceptual side of effort comes down to the words that representatives use when engaging with customers. And so we find a very clear split between language techniques and behaviors used by those companies that are guilty of high effort, high friction experiences, things like silence, time in interaction, where that, interestingly, is suggestive of the fact that the ref doesn't know what they're doing, right, and so so it really starts to exhaust the customer and it sends the customer the message that, or gives the customer sort of a lack of confidence in the outcome they're ultimately going to get. Conversely, you know, active engagement. So where we see ironically, we wrote about this just recently on things like if silence time is bad, what's really interesting is a positive is really active engagement. Now, a lot of other analytics companies out there have talked about well, in the conventional wisdom, I think would be never interrupt or talk over your customer. Right? Never. No overtalk, no interruptions. It's impolite. Just kind of basic common sense. What we find is that's actually a good thing around customer service because it's suggestive of a rep who is actively engaged with that customer. They're not going through the motions. They're in their in their demonstrating their subject matter, expertize which in prior research we found is the kind of experience a customer wants when they're deeply frustrated, they have a problem. They want to talk to us. They don't talk to somebody is going apologize and they want to talk to somebody who's smarter than they are about the issue they're experiencing, who takes control of that experience. And so markers of that are not hiding behind policy, using things like advocacy, like, you know, I can't do what you asked me do. But I tell you what, we are going to get you back up and running. And I have another idea. So bear with me. Let's let's figure this out together. And then just being actively engaged in the conversation really instills confidence in lowers that an entry level.
Ted:
One of the distinguishing factors of of companies that do this really well, especially from their agent side, is that going that extra mile? So so we call proactive guidance, but it is something we see time and time and time again is the best organizations and the best reps are looking for that next possible suggestion or the next possible action, which is the call center has the added benefit of potentially preventing the call back. If you're proactively solving something is downstream that saves us money, makes the customer feel better and so forth. But that going the extra mile is something that you just don't see as often as you probably should. One or two other points I was going to make is we look at the things that that seem to be quite broken. And Steve, you mentioned that high degree of variability. We tend to see that actually in digital experiences. So, you know, people underestimate how much you learn when the customer calls in to talk about what's wrong, because often they're bringing baggage in. Baggage comes from other parts of the experience. So if they were trying to log in and change a password or pay a bill or do other things on a website, very specifically, we see high amounts of variability there for organizations. And when it goes bad, it's generally about confusion. Very specifically, confusion is an emotion for these these customers. They just they're confused as to how to execute this. And that adds to additional frustration that's unnecessary. Now they're having to talk to the agent and handle something they probably should have been able to self serve on. And then Matt hit on another one, which is on the sales side. So so prior to them becoming customers, it's more about uncertainty and indecision and dealing with those types of of things. The winners we find are much better at dealing with the indecision in ways that others aren't.
Steve:
I want to take a break here and tell you about Walker's newest report, "Deliver More Value with XM Data," which provides a practical framework for integrating experience data and operational data to drive better decisions. You can download the report for free at cxleaderpodcast.com/xoreport.
Steve:
Hey, my guests on the podcast this week, or Ted McKenna and Matt Dixon, both execs at Tethr, a company that provides listening platform that surfaces insights from customer interactions, having a fascinating discussion about new ways to think about how you capture and use the data that comes out of your call center or really any of your customer interactions. These gentlemen refer to it as "conversational data." We were talking a little bit offline, but it's one thing that kind of repeats itself often on The CX Leader Podcast: this is a great time to be in the customer experience business generally. You know, the appetite from management is just so all about understanding the customer, delivering superior value, creating great experiences. So you've really kind of created a new source of information, not just for the call center folks or not just for CX, but this is really becoming essential management information in today's economy. You talked just a little bit more about kind of this widespread use of this this data that most companies have had for a long time. But just now we're able to really, really utilize it.
Matt:
Yeah, absolutely. So, you know, this what was once called, I would say go back maybe about 10 or 15 years ago was speech analytics. And that's kind of where this this space, kind of where this industry grew out of that. What speech analytics was was very specifically a technology that was geared toward call center leaders. And a specific use case was, hey, right now we've got a team of people who listen to a sample of calls from our call center agents, and they score them using a checklist like did Steve say the customer's name three times to be thank the customer for her loyalty? And he smiled through the phone. Did he say anything that's going to land us on the front page of The New York Times? Or get us a nasty, nasty gram from the FCC or anything like that. And so they would they would QA or quality, do quality assurance on sample calls. And the original use case was, hey, it doesn't make a lot of sense to have people doing this. We should have a machine do this and then we can free up our Q&A team to do more impactful things like coaching and process redesign and things like that. So that that technology and that that approach has been around probably say at least for a decade now. What we found, though, and as we talked about in the discussion today, suddenly you've got CX leaders who want to get access to data. And what they care about is understanding the customer experience, supplementing or even replacing the survey and really surfacing at scale these CX improvement opportunities, jury friction removal opportunities. Well, you know, if we look across the customers that we're talking to, a Tethr, what's what's interesting is what started as a call center and then grew into CX. Now we have we have customers at Tethr who are coming from the digital team. So as Ted talked about before, these are these are digital leaders who want to understand, in the words of customers, what's getting in the way of them having a successful digital experience, what's forcing them to have to call. And it turns out listening to your calls, using machine learning is a great way to find out what happened before they picked up the phone and what went wrong. But even fast forward to market research and competitive intelligence compliance. Ted talked about sales before. These are all places in the company where people want to get their hands into this data, because now, as you said, Steve, the technology exists to make sense of it all. And so what's really interesting is it's now thrust the call center into a really kind of bright light, if you will, in those companies, whereas once it was managed aggressively for cost and it was kind of this red headed stepchild function instead. Now the call center in progressive companies has become the sort of center of the listening enterprise. Right. This is the place where the conversation happens and it's really cast the call center in a different light, gotten them a seat at the table. I think in many respects for these higher level product strategy and value proposition kind of conversations happening that previously they weren't invited to. But now they sit at the very center of it.
Steve:
Gentlemen, we've come to that part of our show where we always ask our guests to give take home value. This is kind of one key point or one takeaway that our listeners could sort of take out of this podcast and apply tomorrow or next week and improve their own customer experience efforts that they're engaged in. So I'll give each of you guys a shot. Ted, you want to go first on this one?
Ted:
Sure. Yeah. So one nugget just to think about. We spent a lot of time talking about, you know, how do you mix in with your traditional quality assurance measures, the customer's perspective on the way in which your agents are engaging? And so that's where I'd push organizations to think hard about: does… Is there a place on my scorecard where the customer has a voice and they're actually have perspective on the agent, what the agent's doing? And because of what I suspect you'll find is that there are a lot of places that hit what we mentioned earlier where you just kind of go through the motions and they're not kind of taking that extra step. They might not be doing bad things, but that doesn't mean they're doing the good thing. As a way to get us to the next step, so just a thought for 2021.
Steve:
Great. And Matt?
Matt:
I would add on that was a great one. I think I would just hit on an idea I've mentioned before or talked around before, which is what's the stop start around agent language. So how do we get our reps? Like a quick win is stop hiding behind policy. Even if the customer asks for something that, you know, if you do the thing the customer is asking for, you'll get fired in order to like the company will be fined. But there's a really easy of finding place there. Instead of saying, here's I can't do that, said lead the response with here's what I can do. Right. I'd think about times where you're using negative language like "you can't," "won't," you know, these kinds of things in what's the positive language alternative. I think about an example, one of the companies we might of an example, our story from a company wrote about this in the effortless experience. But to actually go out and find out what are those call types, those things that we say to our customers that really drive low NPS, low CSAT, high levels of effort. For them, it was stock outs. That ended up being a real source of frustration when the customer calls in, wants a place in order for a certain number of lighting systems. These guys are lighting systems to builders and they place that order in. The rep says, sorry, we're out of stock. Well, that just causes everything to go downhill. Customers upset because now I got the whole project on hold. Instead of saying that, saying, well, we're excited to say we're going to have this back in stock two weeks and send them right to your construction site. How many do you want? That kind of thing. Even for a commercial real estate developer running a tight ship with that kind of just slight change in posture, in phrasing makes all the difference in terms of lowering the perception of effort. So, again, finding those those things we know are pain points and giving our reps language alternatives to replace them with is a kind of a just do it Monday morning type of thing.
Steve:
Matt Dixon is the chief product and research officer and Ted McKenna is the senior vice president for product, both at Tethr. Fascinating company. If you don't know about them, you should check them out. Matt, Ted, thanks for thanks for being a guests on the podcast. Really enjoyed our dialog. And I learned a lot and I hope our listeners did, too.
Matt:
Thanks Steve. This is great. Thanks for having us.
Ted:
Thanks for having us.
Steve:
Hey, if anybody would want to continue the conversation, can you just give us a little bit of contact information, you know, LinkedIn or maybe your company website or so?
Matt:
Yeah, sure. Absolutely. Invite all the listeners to connect with Ted and myself on LinkedIn. We're pretty active on LinkedIn, so look us up there or if you're interested, learn more about whether it's tethr.com. Because we're a startup, we spell it differently. So it's TETHR. You know, we're we run a tight ship, so we couldn't afford the vowell. So it's tethr.com.
Steve:
It's really great to you guys to come on. And like I said, I'm a relatively new convert to this kind of technology. But after spending a little time with you, I think there's great opportunities for CX to really leverage all of this cool technology that you guys have. So thanks again for being on.
Matt:
Thanks Steve.
Ted:
Yeah. Good luck to you.
Steve:
Well, if you want to talk about anything you heard on this podcast or about how Walker can help your business customer experience, feel free to email me at podcast@walkerinfo.com. Be sure to check out our website, cxleaderpodcast.com, to subscribe to the show, find all of our previous episodes. We organize them by series, by content and category, and we also have some contact information on there. You can drop us a note, let us know how we're doing or give us an idea for a future podcast. The CX Leader Podcast is a production of Walker. We're an experience management firm that helps companies accelerate their XM success. You can read more about us at Walkerinfo.com. Thank you for listening and we'll see you again next time.
Sonix is the world’s most advanced automated transcription, translation, and subtitling platform. Fast, accurate, and affordable.
Automatically convert your wav files to text (txt file), Microsoft Word (docx file), and SubRip Subtitle (srt file) in minutes.
Sonix has many features that you’d love including collaboration tools, share transcripts, transcribe multiple languages, powerful integrations and APIs, and easily transcribe your Zoom meetings. Try Sonix for free today.
Tags: Matt Dixon Ted KcKenna Tethr qualitative data Steve Walker call center