back to main menu

S3E06 – Why Hyper-Personalisation and Omnichannel Support is the Future of CX – Nitin Somalaraju, Tech Mahindra

Duration 25:46

Guest Speaker

Nitin Somalaraju

Nitin Somalaraju

Conversational AI - Product Manager, Tech Mahindra

Industry

Other

Nitin Somalaraju is a Conversational AI, Product Manager at Tech Mahindra and comes with over ​​9 years of IT and Management experience in Software Development Life Cycle (SDLC), business product management, and data-driven decisions. At Tech Mahindra Nitin works as a Product Manager for the Conversational Intelligence & Automation platform (Sayint), and his role requires him to work closely with the Product Design, Development, Marketing & Sales team to create a Product strategy & Implementation plan.

We cover:

  • How conversational AI has helped Customer Support?
  • What metrics should you keep in mind to effectively track customer sentiment?
  • Where is the Conversational AI space heading in the next 2 years?
  • What to keep in mind while investing in an AI chatbot?

If you enjoyed this conversation, please subscribe on Apple PodcastGoogle PodcastSpotify or wherever else you listen from. It means a lot, and it helps support the show. 

Transcript

Jude Gerald Lopez

Welcome everyone to another episode of the Twenty Minute Moat podcast.

We’ve, I’ve got a very special guest this time on the show.

I’ve got Nitin Somalaraju with me today.

He’s a conversational AI expert and a product manager at Tech Mahendra and he’s got a lot of thoughts on conversational AI and where the space is heading. 

So, I won’t take up too much time, I would hand over this to him to give us a brief intro about his role and his work in the conversational AI space.

So welcome to the show Nitin.

Nitin Somalaraju

Sure, thank you so much, Jude. Really excited and I was looking forward from the last week to this particular session to be able to discuss and share some views that I have in this particular space, right?

So, I work as a product manager in Tech Mahindra, specifically in the conversational AI domain. And, we have been working on both, intelligent automation solutions as well as intelligent analytics, from across all different channels, right, voice channels, text channels, that we are looking at.

And we have a, in the process of the last 2-3 years, right, we have built some very interesting use cases across verticals, BFSI, telecom, aviation, automotive, any of the verticals that you look at, I think that conversational AI has a very important part to play.

And, I’m feeling glad that I am operating in this particular domain and really excited to share my views on this particular show, right.

Thanks, thanks for having me.

Jude Gerald Lopez

Oh my pleasure, you know, I think the last few years it’s showing that digital-first is the new norm.

Nitin Somalaraju

Yes. 

[01:44] Jude Gerald Lopez

And, you know, this conversation that I wanted to have with you is going to be along the lines of that, and, where conversational AI plays into customer support.

So you know, we’ll just get right into the meat of the discussion. So, conversational AI and customer support, right, so, how do you think this, this tech part of it, of conversational AI, how has it helped customer support in the last few years?

Nitin Somalaraju

Sure, the last few years have been really demanding, you know, with the entire pandemic going on, it was a certain shocker for a lot of organisations with different business process transitions, right?

Even we, as employees, found it very difficult to be able to just transition to work from home scenario, although many of us were like really excited about it.

And I’m sure, you know, you know, bigger organisations, they would have seen this at, on a much deeper level, and since any of these businesses, the crux of it, you know, since they depend on the customers and how, how much of a good experience that you are giving to customers, right? 

It has become very demanding to be able to provide that customer experience at the cost of losing the personal touch. 

So the pandemic has given us a, you know, an entire new opening that, you know, the positive customer experience doesn’t necessarily need to depend on a personal touch. Rather, personal engagement is more crucial, right?

And we have seen a lot of businesses and organisations reinvent themselves, you know, keeping this in mind, to be able to get that optimum customer experience, even at the cost of not having that personal touch, right.

And this is where we also see conversational AI playing a huge part in meeting some of these customers demands that you know, we have been looking at, right?

And, when we also look at, you know, customer support, you know, many of us, we think that customer support is just restricted to the post-sales of any particular product, right?

But I would say customer support extends far beyond that, you know, being able to help the customer order a particular product or search a particular product is also called, you know, some sort of for customer support that you can provide to enhance that customer experience.

And, I’m glad that businesses have started seeing this new, angle, right, into providing customer support.

And we have been able to engage in some very interesting opportunities across all the different verticals in trying to help businesses provide this enhanced customer experience, right?

So, we see, you know, it’s not just being restricted to conversational AI solutions that you usually see in the form of voice bots or chatbots, it extends beyond that.

It’s about, you know, what kind of analytics can you, you know, gain from these conversations.

What kind of metrics can you gain from these conversations?

Are you be able to under, you know, what your customer sentiment is like, right? Or, are you able to understand what is the customer’s intent like? Even before he asked the entire question or, you know, asks the, or mentions the entire problem, right?

So those are some of the interesting areas where we have been seeing a lot of opportunities developed. And, you know, obviously, the benefits of having conversational AI is having round the clock, 24/7 availability of support, right? 

And, at the same time, being your personal concierge to help you with different processes of, say, ordering a product or taking a particular service, right? Whether it is in the searching process or whether it is trying to order a particular product or service or, you know, post ordering support.

All of these have been certain elements where we could find a conversational AI play a very important role in trying to get that customer experience up, right?

[05:26] Jude Gerald Lopez

Interesting. I mean, I was just thinking about one thing that you mentioned that, you know, the kind of opportunities that conversational AI has brought in. And when you’re talking about that, you can’t, I mean, you have to talk about how this has also brought in personalisation at scale.

With a lot of opportunities, what do you think is at one point there is still a kind of a challenge in support, despite having, you know, all the help from a tech like conversational AI.

Nitin Somalaraju

Absolutely. So you know, over the last couple of years, we have noticed a lot of different challenges that we have been working on the solutions for, right. But, I cannot stress enough on one particular challenge that still exists.

So right, so especially in the customer support industry, when you look at the plethora of conversations that happen, right.

You know, these could be, you know, pre, you know, pre-sales support or post-sale support or during, you know, the product purchases.

You know, it could be, you know, trying to be able to get the right kind of installation for a particular product or, etc etc. right? In any of these, scenarios, we have seen one big challenge recite with the kind of, you know, audits that these conversations go through, right?

Jude Gerald Lopez

Yeah.

Nitin Somalaraju

You know, every particular contact center, every particular organization, they have to audit these conversations to essentially try and understand how good they are doing, right?

What’s going good. what’s going bad, you know. Are the agents be, able to provide the right support? Is the customer feeling satisfied with the kind of experience that he’s getting in? 

And the problem is that you know, there, the number of conversations, they are so huge, that it becomes difficult for any particular organisation to go over them manually, to check what each and every agent is doing or each and every business process is doing. 

And, that has limited all of these organisations to, or restrict their auditing process to less than 5% of their conversations.

Jude Gerald Lopez

Okay.

Nitin Somalaraju

So among a billion calls that happen, only 5% of the calls are ever audited for any of the KPIs that we look at, right, the performance indicators for the frontline guys like the customer service agents, right? 

You know, less than, less than 5% of the conversations are being audited. That means 95%’s views are not being heard of.

Now that’s a huge gap in terms of what the customer is expecting versus what is actually being delivered to them.

And, this still exists till today. And, we have been developing some interesting solutions. In fact, the whole industry as such has been looking forward to be able to solve this problem, be able to get that 100% automated audit processes, where we are able to try and understand what are the different intents of your customers, or what’s the varying sentiment across different type of product lineups that you have.

Where is the exact problem that the company needs to work on, right?

So, you know, these are some of the interesting areas that we have been working on, especially utilising the 100% audit automation features that, that, that, that the industry is moving forward to, right, you know, including that of Tech Mahindra. 

So that would be the biggest challenge that we are still battling it out of there, Yeah. 

[08:33] Jude Gerald Lopez

Makes sense. I was also thinking about, on the lines of, you know, like what data points people have to work with when you’re working with the conversational AI platform.

So when you’re talking about support, and when you are analysing support, do the kind of metrics people track or the kind of metrics that are most sought after, do they accurately or judiciously represent customer sentiment?

Is that, or is there a challenge there as well?

Nitin Somalaraju

Sure, so there are two parts to this particular scenario, right?

So you need to understand that each and every different vertical that a customer support team has,  right, you know, it could be, you know, the BFSI or automotive.

Each of these different verticals, they have different sort of KPIs that they need to be looking at. And sentiment analytics, as such, might not be a really applicable and fruitful across all the different customer support.

Say, suppose, you know, I’m looking at a customer grievance support, you know, team, right?

You know, I have an escalation that I need to make.

Obviously, my sentiment is going to be negative as a customer, right. Having sentiment analytics over there might not be able to give you the right insights to take a business decision, right? 

So, we need to go a little bit beyond that, you know, we need to start looking at, you know, what the customer’s intents are, right?

Are you able to identify what those specific customer’s problems are, problem is, in the very first go, right?

Being able to improve the first call resolution is going to be an obvious driver for sentiment, and not just trying to understand the sentiment patterns, right?

So, you know, I would say trying to be able to identify the root cause analysis for either the positive or the negative sentiment, which can be essentially understood by, you know, analysing the intent of a particular customer, right, is the right way to go about it to solve that particular challenge. 

And, you know, like I said, right, you know, there are different verticals.

They demand a different type of KPIs that we are looking at. And, I think, when businesses are trying to look at automation solutions or analytic solutions like these, they need to be able to identify which specific business processes demand which particular KPIs. 

And that, again, you know, even for companies operating in the same vertical, right, they might be completely different for each and every organisation. 

And that also, you know, exposes us to one of the biggest problems that is there in the industry, which is, you know, the kind of user configurability that is required.

What the industry, you know, players are doing at this point of time is they are, you know, trying to roll out, you know, very generic product solutions which include some of the KPIs that you were talking about, right, sentiment analytics, or being able to have call metadata of how much silence time is happening or what are the crutch words that are there. 

But that ultimately might not be able to drive a business decision. So, organisations need to be looking at what are the specific KPIs that would really affect and transform their business processes. 

And also be look at, also be able to look at what are the specific objectives that they want to achieve at the end of the day.

It’s not a fancy product that, you know, you need to have a chatbot, you know, deployed because everybody is doing it.

Jude Gerald Lopez

Absolutely

Nitin Somalaraju

You need to be able to try and understand if that is going to give you a positive outcome in some of the specific areas that you are looking at, right. It could be trying to reduce your Opex, right, operational expenses, or it could be trying to be able to increase your NPS. 

So we need to be able to try and understand what are the specific KPIs with the end objective in mind.

And that’s how we will be able to customise each of these, right. So, customisation is the key over here now.

Jude Gerald Lopez

Yeah, I mean I, I think, that’s support 101, right? 

Nitin Somalaraju

Yeah. 

Jude Gerald Lopez

Where there’s no, there’s, there, there are only tailor-made solutions. 

Nitin Somalaraju

Yes

Jude Gerald Lopez

You can’t have, you know, cookie-cutter kind of a setup.

Nitin Somalaraju

Especially in the AI domain, right?

Jude Gerald Lopez

Yeah.

Nitin Somalaraju

You know, because, the thing is, is that Jude, it’s still, it’s still a very learning experience for all, you know, the industry players over here, especially in conversational AI, right.

We haven’t yet reached that peak phase where we can say, oh, we have solved this particular problem. This is 100% done, right?

Like we see in other technology-driven solutions.

So it’s still a very learning phase that all the companies are in. 

And, you know, I think the end result will come towards, you know, who can bring that customisation as a pure product play, right. You know, customisation should no longer be solution-driven, shouldn’t be that huge teams are working on a very specific solution for a particular company, right?

You need to be able to productise those customisation elements too.

I think that is where we will be heading towards in the next couple of years too, yeah.

Jude Gerald Lopez

Interesting, interesting. 

Nitin Somalaraju

Yeah.

Jude Gerald Lopez

You know, I was also thinking from the other point of view, from the customer’s point of view. So, support has changed a lot from the time when people had to dial an 1800 number to something that you can resolve in under 30 seconds. 

Nitin Somalaraju

Yes.

Jude Gerald Lopez

So expectations have been growing, and it’s, it’s always good. It’s a driver for change. And how a, how, how should a company react to this?

And how are brands, you know, reacting with these high expectations, higher volumes and again a very digital first approach, so everything is done on chat or now, voice chat, voice AI is also increasing, right?

Nitin Somalaraju

Yes.

Jude Gerald Lopez

So how should a, I mean, what should a brand be wary of when they are dealing with a scenario like this.

Nitin Somalaraju

Absolutely. And, very well put, you know, when you said that customer’s expectations have been increasing, right?

Fortunately, we live in an era where, you know, customers expect what, except that the organisations need to know what they want, even if, even before they even address or ask that particular scenario to companies, right?

They expect us to know it. And that actually, you know, drives us towards one of the key elements in this particular industry, which is hyper-personalisation. And that’s what, you know, even organisations are looking at.

Gone are those days when, you know, you can send in the random email promotions of products, which I’m really not interested in.

Jude Gerald Lopez

Yeah.

Nitin Somalaraju

I have myself received a lot of retail promotions, you know, which, which showcase women’s clothes, woman clothes, right, in the mail itself. And that’s not applicable to me. I’m looking at personalised content, right?

Whenever I call a particular organisation, I mean, you know, it could be, you know, one of the recent experiences that I have had, was, you know, I had to call, I had to, you know, probably wait for 10-12 minutes when I called in my personal banking organisation, right, just to be able to cancel a particular credit card, right?

You know that gets, that’s, that’s valuable, that’s a waste of, any particular customer’s time, you know, it’s a simple problem that I have.

I want to be able to cancel my credit card, and I want to be able to do it in a couple of quick steps, like you said, within 30 seconds. And that kind of personalisation, to every specific problem that a customer might have, is one of the key things that companies should be looking at. 

Hyper personalisation, you know, having very personalised content delivered to them, having very personalised discount offers or promotions given to them, and also being able to have and expand into more of a personal concierge, concierge service, right?

When you look at conversational AI too, you know, instead of going onto a particular e-commerce website, and, you know, probably if I’m looking at, you know, say one of the basic products like a shampoo, right. 

Instead of trying to, you know, search for a shampoo and then use filters, saying that I’m looking for, probably dry hair, et cetera, et cetera, et cetera, 4-5 options, then probably go through the entire reviews and check for which of the products have the highest reviews, and then try to probably filter it according to price, all of this is going to take me 10 to 12 minutes of valuable time, right. 

You know, instead, imagine if I have a voice assistant deployed on a particular e-Commerce website.

And I have an idea about what I’m looking at, you know, maybe not the exact product, but I know what kind of filters I’m looking at.

I can just speak out to the particular voice assistant saying that I’m looking for a shampoo, that’s dry hair, above 4.5 rating, and, you know, I, I want it in the price range of ₹150 to ₹200.

That entire query, you know, took me less than 20 seconds. 

Jude Gerald Lopez

Yeah.

Nitin Somalaraju

And if this also, if the intelligence system is being able to suggest based on my previous, you know, purchase history or similar purchase history from other individuals, that’s going to be a great experience for me.

Jude Gerald Lopez

Yeah.

Nitin Somalaraju

So hyper-personalisation becomes the key, and, hyper-personalisation not just in the common language that we speak, but also being able to expand it to the regional languages, right, that could be, that could be a great way to move forward.

Multilingual support is always going to be one of the key, you know, areas where, you know, the conversational AI solutions can work on.

And one last thing that, you know, we should always keep in mind as businesses is, you know, you know, the omni-channel customer experience that you can get in there.

Am I able to do, you know, all of my purchases you are tracking, say from one of the channels, right, at the right time.

Probably, it could be probably WhatsApp, right? Or, am I able to integrate that seamless experience, you know, both, or, you know, in-store, both in-store as well as online.

So one of the beautiful examples that I always talk to about, with my colleagues is what Lenskart and Decathlon have done, right?

You know, they have been able to provide that service where I can shop around, look around for what are the different products online. And probably place an order, so that I can pick it up from the store itself, which is going to cut down my delivery charges as well as improve their logistics planning also in a much more efficient way.

Jude Gerald Lopez

Yeah.

Nitin Somalaraju

They don’t need to incur the additional overhead operational costs, or efforts, right.

Jude Gerald Lopez

Yeah.

Nitin Somalaraju

And be able to go and search and try out different products.

Being able to create that integrated seamless experience across all the different channels is something that organisations need to be looking at as well, right.

Jude Gerald Lopez

I think you highlighted a lot of like important aspects that customer support as a space has been trying to solve. Few people have solved it well. 

Nitin Somalaraju

Yeah.

Jude Gerald Lopez

And a lot of others are still in the process.

Nitin Somalaraju

Absolutely.

Jude Gerald Lopez

You know, I was also thinking of, when you spoke about the whole banking situation, it also brings us to the necessity of self-service. 

Instead of getting to a stage where, you know, you have to raise a support ticket or stay on call, get to the point where you can speak to a customer service executive and then tell him the whole thing.

This thing can just be a product change where you can just add this button saying, you know, I want to know this or I want to do this. So empowering the customers, makes half the work done already.

Coming to something we discussed a little earlier, you spoke about how there is no one, what do you call, one generic solution, right?

So when you’re talking about brands now investing in conversational AI platforms and chatbots and all these things, are there a few pointers that they should keep in mind?

Nitin Somalaraju

Absolutely. So, like I was mentioning it before, right Jude? You know, so, you know, it’s not going to be like, a success or a failure scenario when you’re deploying any of these AI solutions, right? It’s going to be more of a learning scenario, right?

Every, every particular deployment that you make, there’s some sort of learning that you can make out of it. And what organisations need to focus on is to try and understand what are the different demands for each of these business processes, right?

So, one of the examples, like you have mentioned a couple of seconds back, right, self-service bots.

Jude Gerald Lopez

Yeah.

Nitin Somalaraju

So, you know, is my FAQ going to be listed on a particular page, you know, where they have to navigate to, by probably doing 6 or 7 clicks and then search 1 new corner or in the website where they can get their questions answered?

Or is it more, like, going to be a scenario where I click the button and speak out what my question is and it understands my intent and gives me the right direction, right, or right answer in the right direction.

Now organisations need to understand which of the business processes can actually increase the customer experience by, you know, doing an end-to-end, you know, implementation, right? 

Like the FAQ bot that I was talking about or the self-service bot which can address, you know, say returns or you know trying to be able to order a specific product or, you know, being able to track a particular product.

 Any of these self-service bots or FAQ bots, right, can maybe be some of the simplest forms of being able to reduce the operational expenditure that you have as a company to be able to support addressing these queries. 

At the same time, it’s going to be a quicker turnaround time for customers to be able to make any of these purchases or get their questions answered, right?

Because we need to understand that customers have a very short span of time, very short span of attention, right. 

And especially when we live in an era where almost every product that is there in a similar space is equally technically driven, right, equally driven from the technology perspective, equally driven from the product perspective and equally driven from the pricing perspective. 

The biggest differentiator that you can create is being able to drive up that customer experience by personal engagement, right.

Now, businesses need to try and understand which of these business processes are going to help them drive that particular positive customer experience through an end to end solution.

At the same time, they should also be considering hybrid scenarios, right?

Because with certain business processes and clients, what we arrived at was rather than having an end to end automation solution which can, or a bot which can perform end to end tasks, right, there are bots which can help boost up the productivity of a particular employee. Say, suppose an employee is working as a customer care agent.

You know, you know, if a bot can help him try and identify what is the specific intent group that he’s looking at, is it a balance inquiry that he’s looking at, or is it looking at some sort of a refund or a  reimbursement? Or is he looking at trying to track a particular ticket, right?

If this can be addressed, or this can be guided to the employee even before the customer, you know, gets into the niche problem, right, that could help boost up the productivity. 

If an employee gets to know what was the previous communications that was happened with a particular customer, right, you know, it doesn’t need to be in the same conversation, right? It could be 3 months, you know, before the particular conversation.

If a guided customer journey can be provided to the employee, right, you know, that can always help him take the next best action too. 

Might not be able to drive the entire conversational AI experience where end to end is managed by, you know, managed by an AI bot, but it could also be taking a part of the entire task to be able to help and boost up the productivity of an employee.

So organisations need to consider which would be the right way to go about it for the specific business process that they have, rather than just trying to, you know, drive an end to end solution.

Which might not be really efficient and which might not really help, you know, what their end objectives are, right?

So, I think, careful consideration and proper due diligence are required. And, always we need to remember that, you know, especially when you go to the more complicated use cases, it’s always a learning experience where you develop it over a period of time to arrive at that most efficient, you know, phase where everything is taken care by the conversational AI, right?

I think so patience and perseverance, as well as proper due diligence, is what organisations are expected to do to try and make the most out of the technology that is available out there, right?

Jude Gerald Lopez

Brilliant, I think that was a very comprehensive answer.

It makes a lot of sense when, you know. 

Nitin Somalaraju

Yeah

Jude Gerald Lopez

You look at, you know, how sometimes chat bots are implemented in a hurry.

Nitin Somalaraju

Yes.

Jude Gerald Lopez

It doesn’t really make the experience better, it makes it worse.

Nitin Somalaraju

Absolutely. Couldn’t agree with you more.

Jude Gerald Lopez

And I think it makes the lives of the human agent at the end of it also a lot difficult. So,

Nitin Somalaraju

Yes, yes. Absolutely.

Jude Gerald Lopez

But I think it’s been a very brilliant conversation, Nitin. 

Before I let you go, final question, since you’re in the conversational AI space, anything that you’re most excited about in the next two years in conversational AI, in terms of the strides they are making?

Nitin Somalaraju

Absolutely, absolutely. So, in fact, I would say the whole domain as such, is in a pretty nascent phase. And every particular opportunity that we are looking at is, you know, very exciting for us.

But, you know, looking at the next two years, right, you know what we are focused on as a team, is the amount of, you know, hyper-personalisation that we can bring in this particular space, right?

I think that is going to be one of the key differentiators that a lot of these organisations will be focusing on.

And, also, you know, before I talk a little bit more on the hyper-personalisation part, right, now I think, you know, organisations, especially the industry players in the conversational AI space have realised the importance of trying to be able to prioritise these technology solutions, right?

You know earlier, a couple of years back, we were more looking at, you know, the technical modules like speech to text modules, or NLP, or, you know, text to speech.

You know, we were talking mostly in terms of, you know, the technology which was driving the solution rather than the specific productisation or the use case. Now, you know, companies are coming up with very beautiful brilliant use cases, driven towards a very specific problem statement in a very specific vertical.

It’s not, or no longer an HR-AI solution or conversational AI solution. You have a separate conversation AI solution for the onboarding process versus anything etc. right?

So they are very productive, they’re very domain and very problem statements specific, and I think that’s the right way that we are headed to, you know, all of the organisations, right?

And that is something that I am going to see, where we are going to see a lot of new, you know, solutions coming up which are very specific focus area, right? 

And, talking about personal, hyper-personalization, right, this is where we are going to see most of the upcoming trends going about Jude. 

And, especially if you look at, you know, the kind of models that are being built in the multilingual, you know, capacity, they have been doing bonkers, you know. 

We have been working like, we have been working on a lot of Indic languages. And that kind of exposes us to one of the biggest challenging scenarios, right? 

You know, we see that there is a lot of language mix, you know, when we speak, right? You know, especially with Hindi and English.

You know we always have a habit of making it Hinglish 

Rather than, you know, Hindi or English, right? So, I think, you know, we are going to see a lot of changes coming about in that particular space.

And also, you know, trying to be able to, cater to the emotional intelligence that is being derived out of these conversations, right, becomes a key. So, I have been seeing some very interesting solutions being offered.

You know, especially for the sales and marketing guys during the pandemic, right, where we had a certain use case being delivered, where based on either the facial expressions of a particular customer right, you are able to figure out, you know, what else can be the next best action that a sales agent has to do in a virtual call, right?

So improvements like that, especially in the emotion AI space, is going to be a really good moving forward. 

And also, you know, being able to draw transition from, you know, the regular chatbots that we are using, towards the voice technology, right, is one of the biggest advancements that’s going to happen in conversational AI.

We see chatbots every now and then, almost in more than half of the websites that we go about.

Jude Gerald Lopez

Absolutely. It’s been, it’s been an absolute delight talking to you Nitin, on The Twenty Minute Moat.

So very, very happy that you could make it.

Nitin Somalaraju

Absolutely, absolutely my pleasure, Jude.