Episode 006: Conversational Technology in Customer Support

In today’s episode of the Conversational Messaging Podcast, Gupshup’s Beerud and Srini discuss the inevitable demand for customer support with the rise of social media. They explore how shifting from traditional call centers to automated customer support is the smartest and most cost-efficient way to make consumers happy. 

The Common Mistake of Most Businesses

 Although customer support is one of the most critical aspects of a business, it often doesn’t get the attention it deserves. Most businesses compromise customer service because they think that it’s an expensive investment. Beerud believes that with conversational technology, quality customer support becomes cost-efficient as it can handle peak demand but requires little manpower. 

 “Using conversation technology is the best way to break that false choice. You can have great and high-quality customer support, and do it at an affordable cost.” - Beerud Sheth

Always Prioritize What Makes Consumers Happy

Now is the time for businesses to adapt to conversational technologies to interact with consumers. It’s easier for consumers to interact with brands using the very channels they use to interact with their families and friends. Shifting from traditional call centers to automated customer support can positively impact customer experience, leading to improved customer loyalty. 

“Ultimately, the only reality is the customer reality. The way that customers experience your company, your business, your interactions, is what matters. If they have a difficult time doing business with a company, if it’s challenging, they'll go find somebody else that they can do business with.” - Beerud Sheth

Always Go for the Smartest Solution

Conversational technology is easy to incorporate in customer support because of the simple type of questions. Through AI, Beerud proposes to create a chatbot that understands structured queries, unstructured queries, or a combination of both. For everyone’s satisfaction, human agents can handle the most complicated ones, those the bot can’t resolve, making the solutions inclusive to all types of queries. 

“It depends on the specific business, the brand, their consumers, the nature of the product, the complexity of the conversation, and so on. There's a lot of different ways to slice and dice this problem. There's no question that taking this approach is indeed the right way to automate customer support.” - Beerud Sheth

To know more about Conversational Technology in Customer Support, listen to this episode.

Bio:

Beerud is the co-founder and CEO of Gupshup, the world's leading platform for cloud messaging and conversational experiences. It is used by over 100K+ customers and developers and handles over 6 billion messages per month. He previously founded and led Elance (now Upwork, a publicly listed company), the pioneer of online freelancing and the gig economy.

Prior to founding Elance, he worked in the financial services industry – modeling, structuring, and trading fixed income securities and derivatives at Merrill Lynch and Citicorp Securities. His graduate research, at the MIT Media Lab, involved developing autonomous learning agents for personalized news filtering. Beerud earned an M.S. in Computer Science from MIT & a B.Tech. in Computer Science from IIT Bombay, where he was awarded the Institute Silver Medal.

Srinivas has over 2 decades of experience in running marketing for high-growth technology companies and managing corporate marketing, product marketing and demand generation. Prior to Gupshup, he held marketing leadership positions with Qubole, Pluralsight, Mobileum and Tejas Networks.

Resources:

Gupshup.io

Transcript

0:01

Listen to insights on how conversational messaging is changing the way businesses and their customers engage. Join Gupshup CEO Beerud and VP for Marketing Srini, and an array of guests for conversations about conversations. This is the Gupshup Conversational Messaging Podcast.

0:28

Hello, and welcome to another episode of the Conversation Messaging Podcast, as usual Srini and Beerud joining you, today we're gonna talk about a function or a capability that every business needs to have if it has to survive and thrive and build loyalty, and that is customer support. And customer support is as old as businesses. And this particular industry, or function has gone through several transformative journeys over the last 30 to 40 years. And initially, you had armies of customer support agents in a linear model. So more than number of queries and more than number of calls more than number of agents that were added. But with the introduction of technology, and with the introduction of automation, this particular function or industry is also getting transformed. So Beerud, how have you what have been your experience with customer support over the last 20-30 years? And how have you seen this function evolving?

1:28

You know, customer support is perhaps one of the most critical functions that any business does. I mean, if you think about it, businesses succeed only when their customers are happy. And the primary responsibility for that lies with customer support, or its related functions like customer success, you know, customer service, call it what you want, it's the same thing, okay. But the challenge is that oftentimes, this function ends up being seen as a cost center, and therefore it doesn't get anywhere near the same attention as many other parts of the business, right. And, either I think, if you think about traditional call centers, let's say, you know, in the US and so on, oftentimes, a running call center call centers with voice-based calls is very hard, because then you have peak demand, you know, you haven't, you don't have enough support agents, but at other times support agents are sitting idle, now they're trying to, you know, have some flexibility there, and so on. So that's sort of one part of the problem where customer support, you know, just hard to get it right. And it's, it really suffers. And then if you look at emerging markets, right, India, Brazil, and other places, many businesses say, 'You know, what customer support is too expensive, we're just not going to have customer support.' Right. So they say, you know, they make the choice of, you know, it's a hyper competitive environment, you're selling products at low margins, one of the ways to reduce cost is to reduce support. And I think it, I mean, you know, some sometimes it's hard for business, or if somebody, if some business invest a lot in customer support, they don't get the price premium that they need to continue to sustain it. So it's sort of, you know, you end up with this false choice, where, you know, do you have competitive pricing on the one hand and cut customer support? Or do you invest in customer support, and either increase your price and lose market share, or keep your price the same and low, lower your margins, right, so it really gives you a set of false choices. And, and these are hard things for business managers to do decide. And and that's the reason why write it often is a continuing source of frustration for consumers. And fortunately, you know, fortunately, with new technology, conversational technology, as we're going to talk about today, we can sort of really break through this false choice. And sort of say, look, you can have low prices, and good quality customer support. It's not either or, and so on. So we'll, we'll talk more about that. But I think this has the potential to fundamentally transform the whole sort of customer support industry, the customer service experience for consumers, and so on. Right. So I think it's super exciting.

4:30

Yeah, and I think the other trend that's happened is customer support is now sort of subsumed or integrated with the whole cx view of things that companies are taking. They're looking at the entire lifecycle of a customer from, from a lead to onboarding, to supporting them through their lifecycle. And I think with these, with technology coming into customer support that's becoming part of the whole cx tech stack. And there are seamless handovers that are taking place. which never existed earlier?

5:02

Yeah, exactly. I think, you know, this sort of the customer experience sort of mindset has sort of continually increased over the last few years. And I think, ultimately, the only reality is the customer reality, the way that customer experiences your company, your business, your interactions, is what what matters. And if they, they have a difficult time doing business with a company, it can be challenging, you know, they'll go find somebody else that they can do business with, right. So ultimately, and customer experiences, you look at it holistically, how do you, how do you market to them? How do you sell to them? And then how do you support them? Over time? How do you deal with exceptions? How do you deal with problems, you know, and so on? How can you how do you convert adversity into opportunity convert problem situation with the customer into, and as you know, as with the, with the rise of social media, and you know, the word of mouth, power has grown dramatically. So if a customer is not happy with the business, they can instantly tell hundreds of their friends are 1000s of followers. And vice versa, if they're delighted, even that gets magnified tremendously very quickly. Right. So So if anything, you know, so it's sort of funny, in the old, in the olden days, you know, maybe decades ago, businesses would spend far more in advertising. And usually, you know, good advertising could beat good product, or good support. Now, with social media, with word of mouth, and so on. It's, you know, it's much harder for businesses to really just create this marketing-based perception, because now consumer communities talk to each other, they influence each other, irrespective of whatever advertiser is doing. And therefore, the product experiences support experience, right, the overall customer experience matters a lot more. Today, you know, good product beats, and good support beats good advertising in any day. And I think it's super important to look at, look at support as a core part of the company offering.

7:22

That's right. So we've looked at what you might call the second coming of customer support, where it's now an investment function, and, as you said, a core part of building the CX and building loyalty. So walk us through, you know, some of the technological advances, you know, we always had a phone line that you can call into, or there's always been an email ID that you can write into. But what are some of the significant changes that have taken place in technology, around customer support that's leading us to where we are today, which is the conversational era and support?

7:54

Yeah, I think, well, even before we get to customer support technology, I mean, just look at consumer behavior, which has changed dramatically. Right. Thing, let's say in the 80s, and 90s, you know, if they had a problem or needed to reach a business, they find the phone number and make a phone call. And after that, when the web came about, they would look up the website and maybe submit a query there or send an email, right? And then, but more recently, you know, with the with the smartphone revolution, with the mobile revolution worldwide, you know, I think people are in there on the small screens. Consumers, by a very wide margin prefer to interact through messaging channels. I mean, this is how they interact with their friends. This is how they interact with family. The most heavily used app on mobile devices is the messaging app. Right? It's not so much the voice call anymore. It's not so much the mobile browser, and even sort of apps have a very skewed distribution of messaging-based tools and really adopting conversational messaging, right, developing maybe chatbots, and software agents that can have normal human conversation with, you know, with consumers, therefore, I think we're just just in direct response to consumer preference for messaging. I think businesses and customer support teams are starting to adopt conversational technologies to interact with those consumers.

9:24

Right. So Beerud, let's talk a little bit about customer support use cases that typically manifest in a lot of businesses, especially on messaging apps. We can start with something that's basic and maybe build it up to something really complex that's out there in the field today.

9:40

Yeah, I think, Well, firstly, for most companies, customer support queries that come in, usually have a very skewed distribution, right? There's a few queries, a small number of queries that are frequently asked. And then a lot of other queries, you know, a large number of queries that show up you know infrequently. Right. And clearly, with the use of technology, the first thing to address is this sort of low hanging fruit, the frequently asked queries, right? The FAQs are easy, or relatively easier to automate, it's very clear what they are. So oftentimes, it's like, you know, I've lost my password, can you reset my account? You know, just some basic... well, what are your policies? What are your hours of operation? You know, how do I do this? How do I do that? I mean, those are usually sort of the common frequently asked, you know, the views and, and that would be a good starting point, right? And then beyond that, course, there are more advanced queries, they want to do returns, cancellations, you know, it could be, or maybe even more complex or compound queries, right? I did this and that happened. And that happened, therefore, what do I do now. So I think if you just lay out all the queries on a spectrum from sort of relatively easy to understand and respond to relatively complex, I think on that continuum, right, using conversational technology, conversational AI, I mean, the good thing these days is AI, the state of the art in AI has got very good in terms of being able to do you know, language understanding, there are off the shelf language models that are developed by academia, by research or associations that are available, and that could be fine tuned for a particular domain. Right. And given that, you can, I mean, businesses can really, you know, do a good job understanding the user query, right? First part is identifying the intent, extracting the entities, right, basically understanding what the user is asking, right and, and then translate it into a set of actions or commands, you know, some of it may require an API integration with the, with the transaction system on the back end. Like if you want to, you know, look up the account balance, you want to, you know, cancel an item refund, and amount, things like that on trade, those would require some actions on the back end. And then, you know, the chatbot can provide a response, right, so those are the basic, the first step really is, you know, doing these, taking the FAQs, doing leveraging AI tools for language understanding, and translating it into actions. Now, you know, if some businesses are a little hesitant about using AI, or you know, they might be concerned about the accuracy of AI, well, then in those cases, they can just do completely structured interactions as well, you can provide, you know, a list of sort of quick replies, right? So you ask, you send a message user with option one, two, and three, and then if they select option one or two, then it goes to a different path. And then again, so it's sort of navigating through a series of menus. Right? And that that is also a conversational experience. It's not, you know, it's a very structured experience. But, but it's guaranteed to, you know, it's guaranteed to work. There's no failure modes. It's, it's very deterministic, it's very clear what to do at every step. Right. So there's no ambiguity about that. Of course, with a unstructured natural language conversation, you know, there can be a little bit of ambiguity if the AI doesn't understand the user query, or if it's not phrased, or if it's raised in a way that the AI doesn't understand, then that can be, you know, that can lead to some sort of unknown paths, right. And our one option is to use a combination of structured and unstructured queries, right? Or if the you, the bot doesn't understand the query to state clearly that, oh, I don't understand, or did you mean, A, B or C, right, and then allow the user to disambiguate and guide the bot in a certain direction, right. So, so that's where it is, you know, those are the approaches to sort of automate question answering, right?

14:12

And then it can get pretty advanced and complex as well, right? AI systems now can also handle changing context, right? Meaning that you may be talking about one topic and say I, you know, let's say if the board says, Are you sure you want to cancel this? And then the person says, Wait a minute, can you tell me about your policy? Or tell me about the pricing? Or how much will it cost if I repurchased it? So they completely change the context from one topic to another, and then maybe come back to it right. So AI has gotten very good, where even the dialogue management can maintain context and retain, you know, sort of remain intelligent about the conversation even as the user is jumping from one topic to another. Anyway, so I think all of these automations are possible, but the ultimate fallback is you know, of course, the the human customer agent team is still there and still available. So. So the right way to implement some of these is for the bot to say, look, if I do not understand the query with a high degree of confidence, then I'm going to escalate it to a human agent, right hand over the conversation to human agent and let them take it from there. Right? So and that's perfectly fine, right? So think of it, think of it as a dial where you see as the AI, is learning more and more things, you can turn the dial up or down, depending on what the optimal performance is, right. And in that respect, by the way, you know, you should think of conversational AI is just a new employee, right? Imagine, in the customer support team, you hire a new employee, right? You put them to training. And then in the beginning, you give them some, you know, simpler conversations and see what they do. But if they don't understand they will come to the supervisor, and, and get guidance and take it from there. And over time, as they get more confident as they get better, they can take on more and more responsibility, right? So in almost exactly the same way, think of the AI not not as an all or nothing proposition where either it does everything, right, which can fail, or it does nothing, in which case, you're not taking the advantage of it, I think the right way to say it is okay, but let's start, you know, with the simple frequently asked queries, and then kind of gradually take it from there. And you know, these days with technology, it's easily it's possible to do AV testing. So you can have the bot answer, you know, queries, some percentage of the new queries, and if it does a good job, then you sort of deployed to everybody and keep, you know, the bot capability can keep keep growing as well. So there's many different, you know, ways to sort of mix and match these different approaches. It depends on the specific business, the brand, their consumers, the nature of the product, the complexity of the conversation, and so on. But there's a lot of, you know, different ways to slice and dice this problem. But but there's no question that, you know, taking this approach is indeed, the right way to sort of, you know, automate customer support.

17:16

Well, thank you so much for your views, Beerud. And for our listeners, those of you who either work or manage customer support teams, or looking to make customer support more efficient in your companies, I think, do look at conversational technologies, and especially marry them with messaging channels, and look to improve the impact that your brand has on your customers.

17:40

Yeah. I mean, just being like I mentioned earlier, right, I think using conversation technology is the best way to sort of break that false choice, you can have, you know, great, high quality customer support, and sort of do it at an affordable cost. Because, you know, with technology, you know, you don't have it can technology can handle peak demand. It doesn't have to, you know, you can you don't need a large team. So so you can keep the team focused on, you know, simpler. I mean, the more complex queries that are fewer in number, while automating a lot of the other stuff, but I think this can have a huge transformational impact on sort of business consumer interactions,

18:23

Right. And yes, I think customer support then moves from being a traditional call center to something that's really valuable to a business and it's making a difference to customer loyalty. Alright, thank you so much as usual, listeners for tuning in and thank you for your time Beerud, and your views. We look forward to welcoming you and all our listeners on the next episode.