#H2HChat The future of artificial intelligence and customer service with Scott Carr

Here we are. We are live. We are back for another H2H chat. So excited to be here today. We are going to have a lot of fun talking about some really, really intriguing stuff. I know you're going to have a good time. I know I was just talking with Scott here off air and we were talking about some really cool techno stuff. If you are a geek by nature or you're just fascinated by what's going on in the artificial intelligence world, or you have an interest or are working within the customer service world, you're in the right place. This is going to be an incredible, incredible conversation. I can guarantee it. I guarantee it. I just spent some time with George Zimmer last week, so now I can say that. So, so we are here with Scott Carr. Scott, welcome. How are you doing today?

Doing great, Bryan. Thanks so much.

Yeah, man. So, so happy to have you. Just so that everybody knows kind of a little bit about you, I'm going to give a little bit of background and then, and then, and then maybe you can tell us what, what's not on this page. Some things that we, we, we aren't reading here that we should know about you. The first thing that, that I'll say is Scott is and has been in more than just one business. I believe, Scott, you were saying this, this is how many startups for you?

This is number 5, Bryan.

Number 5. So 5th startup, incredible startup. And it's not the first successful startup. His passion is in building companies. He's currently the CEO of Modria. Hopefully I said that right. The leading online dispute resolution platform where they're helping businesses and government agencies achieve faster and fair resolution to disputes of all kinds. How he does that and how they do that, we're going to dive into that. But previously, he was CEO of Yodamark, building Harvestmark business. Shoppers would use Harvestmark mobile apps to trace their food to the source while producers and grocers used big data analytics to reduce waste and drive sales. Pretty awesome stuff. So they, they were working with companies like Driscoll's, Whole Foods, and Kroger. Some pretty cool, cool places. Prior to that, he was with Yodamark. He spent 11 years at Digimark in 1996, was the 4th employee and pre- revenue and was part of the leadership team that took the business public and grew it to over $100 million in revenue. Before that, he led InCube in the early days of video and on demand. He's a customer-focused sales guy by nature, but loves closing deals that solves customer problems and builds customers for life, which we'll talk about because we were talking about how to create value and where that value is created. And I'm sure that's where that, that part of him comes from. You know, I could keep going on. The last thing I'll say is that he is here in the Bay Area as well as myself. He's a lead, leads a highly active life. He's a dad, a husband, married in a blended family of 5 children, as am I. So we have that in common. Not 5 children, though. So that's pretty awesome. And he has a passion for international travel, gourmet cooking, music, cycling, and being on the water. OK, so, Scott, welcome. And tell us, what did I not read? What would we not know about you that's typically on your bio?

Yeah, thanks, Brian, and thanks so much for the introduction. You know, this idea of building a company is thrilling, but I think the delight for me is when we do something for a customer that helps them serve their customer. And that's why Modria has been such a fit for what I love, because our whole goal is to help our customers help their customers when there's a problem with a transaction. And so it not only rings the entrepreneurial bell in me as we build this company from early stage to scale, But I just love going out there and making everybody's sort of commercial lives better.

So, okay, so now you're in this great place where you've literally— you're making lives easier. You're making human lives easier. And by the way, I want to let you know that anytime you use the word human, this is going to happen.

I love it.

So this is called the human-to-human chat. And so anytime you use the word human, I don't want to scare you. So I'll let you know that that's going to happen ahead of time. So tell me, tell me a little bit about, about your business. It's pretty fascinating stuff. And I love— I'd love for you to share with everybody else. What are you— what are you trying to achieve?

Yeah, let me, let me get there by giving you just a tiny bit of history, which is we're a spinoff from eBay and PayPal. And in 2003, the eBay marketplace was stuck. It couldn't scale because they couldn't efficiently resolve problems between buyers and sellers. So they reached out to Modria's founder, who had written a book called Online Dispute Resolution for Business. They brought him in in 2003 as the first director of dispute resolution, and Colin went in over the course of the next 8 years and built the software platform first for eBay, then for PayPal, that scaled up to resolve 60 million problem transactions a year without human intervention. So 90% of those got resolved without a customer service touch. And what that meant is that participants in those marketplaces could transact rapidly but also resolve issues at the same pace. In 2011, he spun out and formed Modria, and our business is about taking that concept to businesses of all kinds. So we help e-commerce companies scale the post-transaction journey at the same rate that they can scale the front end of their business. And that to me is the exciting point. You know, all of us have shopped online, and when you go out and you've like bought that perfect birthday present and it's supposed to be here on time, the date's approaching and it doesn't arrive, you're stuck, right? And you don't want to get on the phone and talk to a customer service rep. While that's a human-to-human experience, it doesn't delight us as shoppers. And so how do we balance the role of software and the role of people in that post-sale journey, and that's what we're about.

Okay, so where is the line between human and machine? Where is that going to exist in our realm of customer service? Because everybody does need, I believe, an interaction from a human at some level. How do you make room for that?

Yeah, that's a great question. The first rule of these resolution journeys that we design is you should always be able to escalate to a human, right? So when I call United Airlines and I know that I want to Talk to an agent about an upgrade, I don't want to go through the interactive voice response system. I want to say agent, and I want a person on the other end of the phone. At the same time, if my order's late, or if I want to cancel my order, or if what I have I just need to return, I don't need to talk to somebody. I want to do that at 10 PM in my pajamas on my iPad. I just want to go, hey, I paid for next-day shipping, it's not here. And I want the software to say, I'm so sorry, actually it's on its way, it's going to be delivered tomorrow, and here's a $25 gift certificate for your trouble. I don't need to talk to a person. So I think the issue is, how emotional is it? How bad is the situation? Or how much does it need human judgment to resolve? And that's what we try and design in these journeys.

Okay, sorry, you did say human. Okay, so, so let's dive in a little bit into customer support, customer service. Where, where, where have we been and where can we— what's, what's the— what is the ultimate version of the sweet spot, the best customer service situation where everything really is working well through the process? It's the best place we can possibly be. How— what does that look like? Give us a vision.

Yeah, so I think the— let me tell you the problem and the solution. The problem, I think, is that when we build these businesses, we design for the 97% where transactions don't go bad, and we focus on scaling the top line, and we really build this customer experience journey around purchasing. What category? How do I select the right thing? How is payment and checkout effortless? But we don't get to the post-sale journey. Until we start to realize that it's costing us customers, or it's affecting our brand, or it's getting expensive. So to me, that endpoint that you're talking about is that I can resolve issues as fast as I can transact, and it's as effortless for me to sort out a problem as it is for me to click the buy button and have it delivered the next day. That's the journey we're trying to create.

That's great. How do you create a journey?

So I think, you know, the pieces of the journey are understanding what can go wrong. That's the first thing. So if I'm a services marketplace and I'm going to hire a freelancer, the types of transaction issues are different than if I'm buying shoes online or if I'm buying electronics online. So we build these journeys based on the category of product or service, the size of the purchase— is it tiny, is it big? and the volume of these transactions. You know, if I go buy a car online and I'm going to spend 10 grand for a used automobile, I expect a different level of service than if I'm going to buy replacement razor blades.

Hmm.

And so I think what's important is to understand those types. And now what's powerful about Modria is we came out of eBay, and we really saw all of these different categories, and we had to figure out— the team did— how do you build resolutions for everything from Collectibles to electronics, to fashion, to motors.

And

I think that's what we bring into this equation is we've done hundreds of millions of these. And so you learn along the way, right?

Yeah, there's nothing like experience. Is there a point where you won't be able to tell whether it's AI or a human?

I think there's no question that we're going to get there. I think the real question is one of those technology diffusion You know, how long does it take before I don't realize I'm talking to Siri and I'm just talking naturally? How long does it take before I post on Facebook Messenger and I really don't know if that response was from a bot or a person? And, you know, we know that machine learning and AI are advancing at incredible paces. So do we get there in 5 years? Do we get there in 15 years? We get there, and I just— it's a question of time.

We have Alan Berkson, who is heading up or leading the marketing team at Freshdesk, just asked, what is Scott's company? It is Modria. And actually, interestingly enough, your company is working with companies like Freshdesk, directly with Freshdesk, to help bring artificial intelligence— is that correct— into customer service platforms?

Yeah, so we bring resolution journeys into the customer service experience, and we— the technology under the hood is whatever's appropriate for that type of journey. Oftentimes we solve this problem with rule-based systems, with policies. You know, oftentimes our customers say, look, if Scott's lifetime value is more than $3,000, this is his first dispute, and the transaction is less than $50, give him a full refund. I don't need AI to make that decision, right? Mm-hmm. Now the power of a Freshdesk, a Zendesk, a Salesforce support platform is managing the tickets and queues for the humans that serve all the tickets that we can't resolve ourselves. And is we front-end that experience. So you come into an e-commerce site, you've had an issue, you're going to see a Modria resolution button on your accounts page. And you click that button and we're going to ask you, gosh, tell me what's the nature of your problem? And you're going to say, well, I didn't get my item. So we're going to walk you through that diagnosis. Now, if we can't solve the problem and a person needs to get involved, we'll open a ticket in Freshdesk and then the customer support team can manage that process as it unfolds.

Got it, got it. Okay, that makes sense. So agnostic to the platform, it's really the underlying intelligence of the conversation That happens, so you're focused on the conversation. That makes a lot of sense. Hopefully that helps Alan to understand as well. And and so we're going to be taking questions for everybody who just just joined us, or maybe I see there's a couple people on here that are here for the first time. Make sure that you use the hashtag H2H chat. There's a Twitter conversation going on, and and it's a great conversation. There's all kinds of comments from people like Gary and. Larry, Gary and Larry. It's the Gary and Larry Show. Alan and Randy and Loic and Terry, Jim, Molly, Roxanne, whole whole bunch of people, Jason. So welcome to H2H Chat. If you're just joining us right now, make sure you're over on Twitter using the hashtag H2H Chat and make sure you're asking questions because the second half of this I'm going to take your questions and and repose those to Scott so that he can answer your questions directly. I've already got a couple down, so great stuff. So, let's first talk about the chatbot industry and what's happening right there, because it appears that chatbots are trying to do what you're talking about. And I put the emphasis on trying, because I've tried to use them and I don't think they're there yet. But so, what can we expect from chatbots?

Yeah, to me there's a really interesting phenomenon that's happening around us, and to me it's a user experience transformation. We're going from clicking to talking. We're going into a conversational interface with software, and whether that's being fronted by Apple, or Google, or Facebook, we're suddenly able to talk to our systems Star Trek-like and have it do something on the other side. So the first thing I see with chatbots is we're now, whether it's text or voice, the interface has shifted. Now behind the chatbot, the question is how much of that intelligence is actually built into the chatbot, or is being backed up by rule-driven software, or is being backed up by people. And I think what we're seeing is companies experiment with whether the chatbot plus AI makes decisions, or whether the chatbot is an aggregator of the conversation that happens everywhere so that people can respond to the questions.

So, There are a lot of different situations. And when I say a lot, I mean, everyone is unique. Every person has their own way of receiving information.

Right.

Some of us, it depends on the emotions. When we buy something, when we purchase something, a certain level of emotion goes into that purchase. I oftentimes think that we don't actually buy things, we buy emotions. Because that's really what we're— what's giving us that gratitude or that happy moment. That feeling that we get when we purchase something creates an emotion when something goes wrong because we expect it to be awesome. Right. We expect it to be outstanding. And yet it's very rare for a company to be able to deliver on that. You hear the times when Nordstrom used to take back tires, which I highly doubt they actually did, but it still is a good story to tell. Amazon takes back almost anything, but yet I go to a local camera store and they won't take my 24-hour return purchase with my receipt because I've opened the box. How are we going to eventually make this something where the emotion is better when you have a problem than when you first bought it?

You know, it's an awesome question, and I got to tell you, we studied millions of buyers, millions of buyers. And we looked at their behavior for those who had a problem and went through this resolution journey that I'm telling you about against those that didn't have a problem at all. And what we saw is in that loyalty moment, if you deliver a fast and fair resolution, you can return your camera within that 24-hour window, no problem. When you deliver a fast and fair resolution, you get an 18% bump in reactivation. Customers vote with their loyalty when you take care of them when it counts. And that just, you know, we can prove it with data, we can prove it with personal experience. But that doesn't mean that the answer is always refund your money, because what customers want is for you to be fair, for you to be transparent, but you to also be empathetic. So if I ordered a gift for my wife and it hasn't shown up yet, and you tell me— or better yet, I order a gift and I get the wrong thing. And you tell me just return it, you're going to make me go shop again. I don't want to shop again. I already picked the thing I want. I don't want to return it to you. I need you to make it right. So coming up with a journey that's fast and fair, which can also mean, look, you read on the website, no returns after 30 days. It's been 60 days. You can't return it. As long as you can give them that answer, what we know is that consumers would rather Lose fast than win slow. Creating this fast and fair resolution, this journey that allows me to delight you, delight you as a human at that point of transaction trouble, that's how we win.

Delighting a customer sometimes means that you have to interact with them. One time, I remember Amazon. Did this incredible, um, response to someone who had, uh, had something happen to his, um, King Arthur sword. As funny as this sounds, that this is all about a King Arthur sword, the irony of the whole thing is that the customer service rep on the Amazon side responded in the same time period speak and said, art thou sword is lost? We shall find it for you. Please bear with us while we contact King Arthur to make sure he can get his, his men on it and continues on. And I'm like, this is epic. How cool is this customer service moment? I just want to order more things from Amazon now.

Right.

How do— and that's the delightful moment that you're talking about where you don't have to actually give something free. Sometimes you can just, just be there in the moment with the person and surprise and delight them. Totally agree. How do you do that with artificial intelligence?

And I think the question is, when is artificial intelligence at a point— Yes, thank you. —at that level of emotional intelligence? It's not there today.

Yeah.

When I say, hey Siri, call Caten Shaw, she says, do you mean Eric Shaw? And the first thing I feel is frustration. No, I didn't say Eric Shaw, I said Caten Shaw. Now, if you do that to me in a transaction problem, I'm going to— my emotion is going to go straight up. You didn't defuse it. You weren't empathetic. Didn't make me feel better. So I think really, to me, this is parsing out— there are customer service issues that AI can address, and that, that set is going to get broader and broader as AI gets bigger and bigger. And what we have to do as, as humans is be careful about expanding that line only when we know it's to the benefit of our customers. Because if we're only doing it for the benefit of ourselves, we risk pushing our customers away. And so I think this becomes a delicate design balance. I was reading an article the other day that KLM is using artificial intelligence to power its social media support reps, and they'll get a question in on social media and the AI will propose a response. Now, 90% of the time, the reps edit the response. But it makes their job easier because they have a starting point. And the question is, when does that needle move to 90% of the time the response is right? And that's what we have to watch carefully. This is, this is an evolution, right?

Um, it's totally an evolution. And it's, and it's, um, it's, it, it strikes me that this is more marketing than the actual marketing that took place to get you to buy the product.

That's right. That's right. Yeah, and you know, the, the— I kind of go back to my airline experience. I'll go online and book a flight, no problem. But if I'm trying to make sure that I can get an upgrade over the Atlantic on the way to Europe, I know the agent on the other side of the phone can work this system better than I can. So I'll pick the phone up and call. But if I just want a direct flight from San Francisco to New York, I'm going to do it online. So how do we provide an effortless experience for our customers, but still let them get to a human when stuff gets messy. I think that's the design center.

Yeah, that's right. So, okay, so what you're talking about is 1,000 if this, then this.

That's right.

Machines are input only, and they only will respond if you've given them that input to respond in that way. Is there any chance they can go beyond that?

That they can begin to reason, that they can do their own inference?

That they can take something and make a decision based upon something that happens? Yeah, can they make an inference on something?

Yeah, I think that, to me, is the difference between machine learning and artificial intelligence. Machine learning is that I can train myself on hundreds of thousands or millions of instances And I can make the same judgment you would have made given all of that data to train on. And artificial intelligence is I can take what I know and I can project something I didn't know. And I think we see IBM Watson maybe in that box of really being able to move forward. And we have prospective customers who are doing innovation with IBM and trying to bring Watson into their customer service centers, and they're the leading-edge innovators who really see the potential here. But I think it's still R&D. I don't think it's— it's not yet time to turn over our customer service experience to an AI.

Yeah,

and the other thing I see is that the data that we're using to train on is full of noise. When we look at all the transcripts from a customer service agent's sets of experiences, how do we know which ones produced good outcomes? So one of the things that Modria does in these experiences is we put metadata around every journey. We know what the outcome was, how much it was, and we can also run NPS surveys on the back to help us train these machine learning algorithms so that they're better and better at creating good journeys.

Oh man, this is such good stuff. We have just 5 minutes left before we're going to start taking— I'm going to start doing the questions, and quite frankly, we're not lacking questions. Awesome. So even if you keep your questions coming, don't worry about it. We'll get them out to Scott and his team so they can answer it for you. But so keep the questions coming. But everybody is definitely enjoying this conversation, as I can see the conversation's going pretty quick on Twitter here. So thank you all for your comments. Now I want to dive into the timeline that we're looking at. What does it look like in the next year, 5 years? How are things going to change based upon what you're looking at and what's possible?

So I think we get to a point within 5 years where somewhere between, say, 75% and 90% of your customer service journey problems are addressed in software. Now, whether that's addressed by an AI, a machine learning algorithm, or rules, I think we're going to find that the majority of our customer service issues are solved in software, not with people. I think that's the 5-year horizon. I feel like it's more like 10 years before we see AI using emotional judgment to help solve the really messy problems. That feels to me like we're still over an R&D hurdle, because I feel like right now we're in the convert data to action Not convert data to inference. And really, what we're talking about is the democratization of artificial intelligence. Because if you kind of step back, what I see is that Apple, Google, and Facebook have created AI interfaces to the rest of the world. When I say, hey Siri, book me a restaurant reservation, Siri's doing the AI, and OpenTable is actually—

Did Siri just answer

you?

The ultimate irony just happened. And

then she asked me which restaurant, which was

great.

But what I could tell you is, had I said book a table, she would have called an API at OpenTable. OpenTable didn't do the AI.

Right.

So how long does it take for that artificial intelligence to diffuse into the corporations that are providing these services?

Yeah.

And to me, that still feels like a decade.

Yeah. OK, so just a couple more minutes. I'm going to— before we dive into questions, keep your questions coming, #H2HChat. Okay, so, so part of where you guys have also been working is just even inside of the government, right? And obviously, I'm sure there's some things you can talk about and some things you might not be able to. But we're also talking about disputes between 2 people, 2 humans, 2 people maybe that are getting a divorce or that are arguing over a car sale and who owns the car, you know, 1,000 things. How can a I don't want to say a robot. How can artificial intelligence dispute 2 separate sides, 2 emotions, and bring them together and solve a problem?

Let me tell you that, in our view, there are 4 stages to resolving a dispute. The first stage is diagnosing the problem. You need to tell me what your issue is. You need to describe what's going on. Hey, I got my property tax bill. You're way out of line on the property valuation. My taxes are too high. And I want to complain, and I'm going to ask you some information to diagnose that problem. Why do you think that's true? Now, at that point in time, a piece of software could make a decision and say, you're right, Scott, we overvalued your house. We look at comps, we agree with you, and I'm going to reduce your valuation. And then we're done. Software did the work. But if I don't agree with that process, I want to negotiate. So I go from diagnosis to negotiation, and now we're human to human. We're going to negotiate with one another and we're going to try and sort out that problem. Now, if we can work it out, that's awesome. If we can't work it out, we might want to bring a mediator into the problem. Mediator's job is to be neutral and facilitate our negotiation. If the mediator can't get us through that, then we bring in an arbitrator. Now, an arbitrator today is a person who is skilled in that particular art, who knows the law, who listens to both sides and makes a judgment. Over time, that arbitrator could become a digital judge who could be trained on all of the related case law, on all of the laws of the land, and make a judgment that's based not only on the data of the case, but on more data than a human can probably process. To me, that's these intersection points of where software intersects with, intervenes in the resolution process.

It sounds like— Like the emotion then will be taken out of out of it because then it becomes a a question of what you put in will drive the decision the the the the the ability for this AI I don't know robot or or or decision-making system to to go based upon sheer fact. There is no opinion that it can go on. Like, if a judge were to make a decision, yes, they're going to go based on law, but there is a certain part where there's also emotion. You're removing the emotion and putting in sheer decision-making based upon fact. Is that right?

No, it is and it isn't. I think there are cases that are pure fact-based that we can do just with data. But there are also cases that the emotional impact, the emotional harm, and the emotional judgment are really the basis for decision. The first thing I'll tell you is that when we resolve things online, we can calm our emotions. If you and I are in the middle of a dispute, we each can take a little extra time to frame the problem, and that allows us to take the unnecessary emotion out but still be heard. It's critical. In these kinds of problems that you can be heard. So that has to be part of the journey. But there are also going to be situations where, you know, for example, we in one of the early Modria implementations, we created something called Community Court. It was for a classified site in Europe where a buyer and seller had a problem and they needed to work it out. And so they actually convene a jury out of the classifieds community online. They each tell their story. The jurors vote and the majority of the jury carries the decision.

Now,

that's emotional.

Mm-hmm.

But it's also a jury of my peers. So we can't lose the basic design of our justice system and throw it over to some cold fact-based judgment only. We have to divide up the different types of issues and make sure that we maintain the richness of the justice system, but bringing the technologies of the online world into that environment.

Awesome. Oh my God, mind blown. Okay, so congratulations, Scott, and to everyone who is on H2H Chat right now. Everybody has helped to make this trend on Twitter. I couldn't be happier for this topic and for everything that you're contributing here, Scott. Thank you so much. We're going to move into the— and we're trending, by the way, nationally. So this is, this is pretty important stuff. Great stuff. And thank you again, everybody, for, for all of your comments. I am going to attempt to get through all the questions, which means that we're going to have to— we won't— we don't have to go too fast, but we probably will take a couple of minutes per question.

Yeah.

OK. Let's start with— sorry if I get your name wrong— it's Rash Chan. He said, can AI make customers feel special? We humans failed miserably at making people feel special.

I think this is the area where AI is trying to bring emotion into the situation. If there are patterns in a dataset where we found ways to make the customer special, I think we have a shot. Personally, I'm not ready to turn my customers over to an AI to let them feel special. I feel like the AI needs to help humans make humans feel special. And I think, over time, perhaps it matures, but I still feel like AI is a tool, not a replacement for the human-to-human interaction.

Okay. Well, I have to do it this time. I've been holding off a little bit here and there, but this one deserved it. All right, here we go. Next question is Loïc Coutier, I believe. What are some of the pitfalls of AI in customer service?

So, you know, you have this potential that everything kind of regresses to the mean and you lose all of the nuance, and if you just flip the conversation over to an AI, you could get a very cold response that only feels rule-based, where your customer isn't heard. So I tell you, you can't return your item, and you tell me, but wait, yes I can, you don't understand. So I think one of the pitfalls is to view AI as having replaced the humans that deal with that part of the problem set that software can't address. And that to me is the risk, where we race too fast to a new technology and forget we've been serving customers You would think,

right? Okay, great stuff, great stuff. I'm going to keep it going just for purposes of time here. Jason Martin— love Jason, he's always a great H2H chatter— said, when AI goes awry, how do you save the user experience from drowning before the customer leaves? How do you hide the malfunctions from them?

So we, we track metrics on the interactions, and I think the key is watch for problems. For example, how many times are we going back and forth in chat? If I come in with a problem, the AI responds, and we resolve, we probably had a good experience. But if that chat goes back and forth a couple times, we should be worried. We should be thinking, you know what, maybe we should open a support ticket. So it's kind of like setting some backstops, some pressure valves. And always give your customer the opportunity to say, I want to talk. I want a person. Whether talking is in a live chat or talking is on the phone, you have to give a pathway from the AI to the person. I'll tell you one quick anecdote. I have a banking relationship. I pay all my bills online. I bank online, but I have a safe deposit box. And all I needed to do was get my safe deposit box rekeyed. I called my branch. It took me 14 times through their auto attendant. 14 times to get to a human because software couldn't rekey my box. That is the failure, not of AI, but of technology to serve customers.

Yeah, that is— so it's human error will drive artificial intelligence

error.

Absolutely. Because somebody didn't think through all the questions we need to ask. Just so I understand this right, is that what Modria does? Do you think through every scenario that is possible and then help to—

Yeah, we design resolution journeys. We've designed a set of pre-built resolution journeys for e-commerce, but as you said, we've also powered online divorce, consumer protection, landlord-tenant issues, and we come in with, hey guys, we've done this, and then we tailor around the edges to our customers' specific business environment.

It's, um, it's a lot of resolution.

It is that.

Saving one marriage at a time. That should be the tagline.

Um, actually, the tagline is separate together, believe

it or not. There you go. Um, okay, so Cindy, uh, Millett or Malay, I'm— again, I apologize. What do you think will be the percentage of AI used for the future of customer service in big corporations?

So I think we're going to— I'm going to answer the question a little bit differently. I think we're going to trend towards seeing 80% of all customer service issues resolved in software, and that software is going to be a combination of rules, machine learning, and AI. And I think that's what we're headed toward, because when we look at the nature of problems that businesses face, there's a large percentage that are patterned, but there's always going to be 10 to 20% that you're going to have to go to a person with.

Yeah, that's— it'll never get to the Terminator.

No.

Good. Okay, so let's see. I apologize. For some reason, the name didn't come through with the question, and I'll have to search it out. But the question— so I apologize for that. But the question is, what is the biggest challenge you see in the space overall? What is going to be the biggest challenge?

Yeah, so to me the biggest challenge is that customer service gets underserved inside the enterprise. So when I go in and talk to a customer success team, what I see is they struggle to get dev days, they struggle because they're not a profit center, and so sometimes they get starved off from a resource perspective, which limits their ability to innovate and limits their ability to transform the post-sale journey. And I think that's the biggest challenge. It's not that we don't have technologies to bring to bear to these problems. It's that companies haven't all transitioned to the idea that creating an effortless experience from front to back is how you're going to compete in this century, because you can try and compete on category and price alone, or just product, but honestly, that service experience is what's going to make the difference between the explosive growth and the moderate growth.

So that was, that was Alan Berkson, by the way, from Freshdesk. So just wanted to give a kudos to him for, for the question. And now that makes sense as to where the question is actually coming from, too. Okay, so moving on. Monica Smith-Mullen said, when building a journey, what data elements are most critical for an experience that delights with fast and fair resolution?

Great question. So we need to understand the nature of the problems that our customer's customer is going to face. That's the first and foremost issue. What is likely going to happen? And how does our customer view what they want that outcome to be? So if you tell me you didn't get your item, if you tell me it was not what you expected or the service wasn't delivered, in designing a journey, we have to essentially know what's that relationship you want to have with your customer when things go bad. It's kind of like doing a little bit of marriage therapy, right? What do you guys do when you get into a fight? How are you going to solve these problems? And that's what the journey design is. So some of it's raw data. Give us numbers, tell us call statistics, tell me what's showing up most often. But the other side of it is, how do you want to be different than your competition? How is it that you're going to create that experience differently than the guy down the street is? And then we try and embody that in the journey to reinforce your brand.

Now, could AI tell you, I wouldn't say that?

So, we have an advantage, which is we get data across all of our customers, and we get essentially a license to aggregate and anonymize data. And it lets us come to our customers without— there's no competitive data here, but to say, look, you know, in the top quartile, here's how people are dealing with this issue. And you might be able to learn from that. And the good news is all Modria customers get lifted by our ability to improve those experiences.

Phenomenal stuff. Again, mind is being blown over here. If you guys have any more questions, keep them coming. I'm going to keep zipping through them as we go here. Next question is from Mitch Jackson, who is our resident lawyer, and such a wonderful guy, one great friend to me and to the HHA Chat community. He said, how important is the The mobile tech aspect of the AI/chatbot/voice tech you're discussing, what devices are consumers using?

Yeah, I mean, I think we all sort of know this intuitively, and it's a great question. We've crossed the tipping point where the majority of interactions today are mobile. And when I walk into e-retailers and marketplaces, and even payment networks, whether the number is 55% or 70%, Their customers are mobile. And now the geographies that we're in, mobile means something different. So here in the US, that's an Android or an iPhone. When I go to India, it may or may not be a smartphone, it may be a flip phone. And so mobile oftentimes can still mean SMS. So when we think globally, there's this shift to mobile, that's my device, but still not all those devices are smart. So we design for essentially both a desktop and a mobile experience because we're still in transition, but I don't think it's long before we're going to see that whether it's a phablet or a flip phone, that that really is how we interact with the internet.

Great, good stuff. Ecovia Solutions said, how do you make that human connection with your customers through AI? How do you build rapport even when it's not a dispute? That's a good one.

So it's interesting to me, having been a technologist for my whole life, that when we go deploy Modria with a customer, the majority of our time is not spent on technology, it's spent on language. It's spent on the tone that's created by the way we talk during that journey. What do you call your customer? Are they a client? Are they a customer? Are they a partner? And what do you call a dispute? Is it really a dispute or is it an issue? or a problem or challenge, right? So the way we construct that language sets the emotional tone of the software-human interaction. And we can't forget that because, you know, we're going to respond as people to the language that we see. And that also goes to the colors and fonts and styles in that journey. It needs to feel soft, it needs to feel inviting. And so these pieces of human experience and user experience, I think, are in part how we create

Absolutely, absolutely. Good stuff here. Stephen Rothman said, what impact does this have to the customer service centers? He said, such as layoffs or redirects.

Yeah, it's a great question, and it's a human question, because we're going through a transformation in customer service, and what does that mean to the people? Now, when I talk to Our customers who run big customer service centers, what they say is they want that talent to solve the messy 20%, to broaden their relationship with their own customers, and do upsells. So to me, it's really a redeployment of human capital. It's not an outright displacement. But we all know that the more we can build relationships with our customers, the longer they'll stay with us, and the more we'll have this growing business. And so, what I see is that the human capital shifts from the things that we can do easily in software to the hard stuff that we can't.

Okay. Fair enough. Okay. Where did we leave off here? Okay. Gary McIntyre said, will AI be available to segment those who prefer human intervention sooner versus later?

That's a good question. We have lots of signals in our interaction with software that I think AI should be able to detect. When I say agent to United Airlines, there is no question I want a human. When I go online and I'm clicking through software, I should always have a pathway to show my preference, and this is about creating personalized journeys. Now, you can imagine without too much science fiction thought. thought that these AIs can get to know us as people and detect when we would prefer to have a human interaction versus when we can't. And that's this emerging sort of digital assistant that we might see from a Google, from an Apple, from a Facebook, where you pattern my interactions over time and you say, you know what, I think you probably would like to go straight to a live chat.

Yeah, absolutely. So you think that there's a digital body language that may eventually be read by a somewhat intuitive platform that knows and learns your preferences over time?

Yeah, I think this is that crazy worry about the intersection between what does the net know about me and how much do I let leak into software. You know, if you were Google and you watched that I just surfed, I need the executive of customer service, I want to know how to file complaints, I went to the Better Business Bureau, and then I went to your customer support site, you might be able to begin to infer I might be upset. Right. And so the question is, how much of that data gets integrated into these AI experiences, and who in the ecosystem actually has that purview? Because if I'm buying from Newegg, Newegg might not know that, but if I'm buying from a Google Store, well, Google probably knows

that.

So where do we create those boundaries and how, or should we, take advantage of that data?

Oh man, do you just like go home at night and just like, I gotta rest?

This is the life of an entrepreneur, Bryan.

Every, every, uh, every time you get in an argument with your wife, do you just pull out AI and say, here, here you go?

I'm way smarter than

that. Not good. Way

smarter than that. She's got much more emotional intelligence than I do, trust

me. Very, very cool. Um, okay, more, more stuff, great stuff continues to come in. I— we're having such a great time here. Thank you so much, Scott, and everyone out there. We've got just 10 more minutes here. We're going to— great questions here that I'm going to cover. First one is from— another one from Mitch Jackson. This is a good one. What if human mediator, the human mediator feels that AI proposed a resolution that's contrary to the law? Can he or she override the automated decision?

Absolutely. So let me give you a real-life journey that goes on today in the Netherlands. We power As we said, online divorce. And the way this works is one spouse goes in and answers a long set of questions about everything from assets to children to visitation. And then the other spouse comes in and answers the same questions, and the software tries to compare them. And wherever the spouses picked the same answer, then we check that box and say, sweet, that's, that's good. We've agreed on that topic. Where they disagree, we bring a mediator in, and the mediator can mediate that one topic. Now, if they still can't agree, we can bring an arbitrator in and make a decision. But here's the catch. At the end of the whole process, we have a marriage settlement agreement which goes to a reviewer, and that reviewer's job is to make sure that the settlement agreement conforms to the law. And that's important because the parties, not being lawyers, could negotiate something that either radically disfavors one party or is just simply not legal. And so you have to build in these checks and balances. You can't just assume that the software is always going to get it right.

Yeah, that's— well, that's good. I'm glad to hear that. You just gave us all hope that the humans are really going to run the—

We still have a job.

Run the world. We're okay. Okay, good. So Kara Brookins said artificial intelligence doesn't have to feel emotion. Only to simulate it. Can wit or clever responses take this place?

Yeah, I think this is one of those appropriateness questions, right? If we create a witty chatbot, sometimes it can diffuse the emotion, make it funny, and help me just get to the issue and solve the problem. But other times, the wit could seem incredibly insensitive, and the result could be to inflame the situation and make everything worse. And this is where I don't think AI has matured to the point that those nuanced differences are appropriate or that we can count on them, because when we get it wrong, it can go wrong really badly. So today we see that kind of rules and machine learning, I think they're doing a good job for the types of issues that we can resolve in software, but AI itself is still kind of behind the people.

Good stuff, good stuff, okay. So that leads us to another question from Gary McIntyre who said, Scott, can you formulate data arguments for using AI versus human touch in specific points throughout the journey to a CMO?

To a CMO, yeah, so I think the formulated response is we know when you can deliver a fast and fair resolution at the pace of commerce, The CMO, the marketing guy, wins the trust of his customers, and we can show that straight up that it's double-digit growth.

Wow.

And you— what you know is it's about time. Now, whether that's AI or rules really doesn't matter. What matters is that when I have a problem, if I have a buy button, why don't I have a resolve button?

Right.

If I can click resolve and I get an answer to my issue and I get it timely and it's fair, You earn my trust, and that's the pitch to the CMO. Now, whether it's AI or not really just depends on how much of the problem set can you cut into, and how adaptive is it. So the advantage of machine learning and artificial intelligence is to adapt those policies as the market changes, as your customer base changes, and that to me is where the science has real benefit.

So analytics are huge here. And we haven't really talked too much about the analytics side. This is a great question because it really does drive towards the time or speed to closing or resolution conflict. If you solve something between AI versus a human, you're not taking up someone's time, which means that you're not paying for that time or eventually going to pay for that time. How many of these KPIs exist in what you're talking about?

So we instrument the whole platform and we do it for just that reason. This is a business process. It's a business process that needs metrics, and we, you know, we're marketing guys too, we call it return on resolutions. You need to tune the amount of protection payments and returns, the loyalty earned, and the labor cost, and find that perfect balance for your business, and your equation is different than somebody else's. And so that is all about metrics and measuring the journey, and what we watch are all types of things. How many times did you go back and forth to get to an answer? What were the economics of the outcome? You know, how do we make sure that we are delivering experiences that delight shoppers? And I think that's critical in trying to design these journeys.

You know, I have so many questions for you, I think we're going to have to do another one of these. Okay, so let's keep Let's keep going with some questions here from some other people here. We've got Jason Martin said, I am super jealous of the job that Scott has. Thanks, Jason. How exciting is it to design AI to resolve human-to-customer issues?

I think what's really powerful about what we do is that it's really putting online technologies to work to solve these human-system interactions. And it's technologies of all kinds. Oftentimes we can go in and solve problems with rule-based systems that just cascade through a series of rules that are intuitive and well-designed, and there doesn't have to be artificial intelligence behind those scenes. One more comment that I'll make about an earlier question you had asked: as people, we hate to get on the phone and argue with someone. Most of us don't like conflict. And so when I have an issue and I want to do a return, I have to get myself all worked up to explain to you why I want to return it, and I'm afraid you're going to tell me I can't.

Right.

But if I can go into just software and I can say, I want to return the item, well, hey, now I'm having an interaction that doesn't cause me to enter into a fight, and that's a benefit.

Oh man, that is a huge benefit. Okay, let me keep going. Monica Mullins said, what will help to elevate customer service to an equal political level as sales and marketing within a company? With equal resources. Does that make sense?

It does, totally. So I think there has to be a shift in awareness to understand, you know, it's funny because the economics in the business are there. We know that there's a certain cost to acquire a customer, and we know that if we lose the customer, that it's going to cost us way more to get them back, and we may never get them back. So customer service is at that intersection. They own those loyalty moments, and if you want to drive profit in an organization, that's where you need to invest, because you want those repeat buyers. That's where all your margin comes from. Your first-time buyer was expensive. The guy who buys the 10th and 20th time, that's where your profit is.

Mm-hmm.

I think we need to elevate that into the financial organizations and the marketing organizations so that they see that they hit their objectives by empowering the customer service guys.

God, this is a— this is— you're going to be in business for a while. I hope you know that.

We hope so.

Well, conflict is not going away. Okay, so Vamsi, I believe is how you pronounce it, said, will these bots work autonomously? For example, if I ask Alo for an answer, can it in turn talk to another bot And then come back with the answer?

So I think this is a time horizon question, and I think that that's maybe— is that, you know, wave 3 of artificial intelligence where agents interact with one another? You know, a lot of us have been envisioning these possibilities since the mid-'90s, and we're only now to the point that software can interact in natural language. So, you know, how and when do people create open interfaces where bots can talk to each other? To me, one of the things that dropped that barrier is the fact that we're beginning to have natural language interfaces. Because previously the issue was nobody had built the APIs for 2 bots to talk programmatically to one another. But you can imagine there's no reason why a bot who can speak English couldn't ask Siri a question. So again, that still feels to me like it's out there in that kind of 10-year-ish kind of realm, but who knows, there could be an IFTTT

Oh man, and I just got the hang of IFTTT. Jason Martin said, maybe we shouldn't make chatbots look like live chat to make sure users don't think that there's a human behind the bot.

Yeah, so at Modria, we're big believers in transparency, and, you know, just like when we're gonna have a conversation and we have a discussion, We make sure you can see the whole discussion thread over the life of the dispute because you don't want to— one, you don't want to have to restart, but two, you want to remember everything that was said. So we're big believers that if it really is just software responding, say that it's the software that's responding, because otherwise I could think, really, a person could be that rude? And that misinterpretation can create huge issues.

Right. So many different possibilities, so many ways to go. You guys are doing some incredible work to solve solve some of the world's problems. I can see that, you know, and there are world's problems. I mean, we just had a debate on last night that could have used your technology. So, so I can't tell you how much I appreciate, Scott, you being on here today for answering everyone's questions, for just being engaging. Congrats, everybody out there for, again, trending on Twitter. This is fantastic topic. And maybe, maybe we can get Scott on for another interview sometime. and talk about some other things because this is a wide-open discussion. But Scott, thank you so much. Where can everybody find you?

You can find me at @J_Scott_Carr on Twitter. You can also find me at scott@modria.com. Bryan, thanks so much for having me today and for the opportunity, and thanks to everybody in your online sphere for their questions. It's been a fantastic interaction.

It really has, really has. And modria.com is your website, correct? M-O-D-R-I-A.com. Absolutely. Great place to go check out some good resources on all of this. Thank you, everybody. We will see you again next Monday. If I were ready, I would have already known who that is. But I am— oh, actually, next Monday— sorry, sorry, sorry. Here we go. Here we go. I'm going to have it in just a second. I'm going to have to tweet it out. I can't find it. So next Monday, I will let you know over Twitter where— who's gonna be. I'm sure it's gonna be fantastic. But next Monday, same time, same place. Scott, thank you so much. And we'll see you guys all again next week.

Thank you.