Podcast | Why now is the time to build chatbot optimization tools – not just chatbots?

Jean
Shin

Director, Strategy & Content | Podcast Host of Mobile Interactions Now

21 min podcast
Why now is the time to build chatbot optimization tools – not just chatbots?

In this Part 2 of our discussion with Dan Leshem, we venture further into what the tech community needs to deliver in order for chatbot interactions to meet business requirements. Yes, we’re talking about the next-gen tools, not just for building bots but – perhaps more importantly -- for optimizing to improve performance. Find out why Dan is calling for chatbot optimization tools.

Podcast transcript

Jean:

Welcome back to the show, Dan. In our previous episode, we touched on some overall trends that is happening, particularly with chatbot, and focused more on customer facing interactions. And now I would like to delve into some specific use cases, and maybe even pick up from some of the delivery use cases we started out with.

Now, the way I look at it, is that some of the industries have kind of a longer history dealing with automation in general. So I'm curious whether some of these industry sectors are going through a different adoption curve when it comes to what is available today?


Dan:

So I think each company has its own timeline and customer demands when it comes to adopting AI. And not necessarily in customer communication channels, but also one use case that I really like is companies adopting AI workforce and their employees to communicate better and to collaborate better.

So, that's a really great use case that I see. We're a little bit far on that end. I think we see companies using AI for external communication rather than internal communication. And of course the companies that will adopt and embrace AI initially are going to be very heavily consumer-based companies. If they can ask their customers the conversational interface that really works and offer a real value to the customers, this is what they looking. So we can see this in e-commerce and digital-first brands first. And then later on, we're going to see companies using AI to help for internal communications as well. And maybe later, large enterprises in industries like utilities, or even SaaS companies that today they don't really need the competitive edge of customer experience.

It might not be needing it compared to e-commerce companies. So I think it really depends on your industry and how customer experience is essential to the success of the business. We see companies in the SaaS industries and they embrace AI. They adopting AI even more than e-commerce companies for some reason, although it might not be the great use case, in my opinion, because if we come to SaaS, it involves very technical support and troubleshooting. And I see the value from the company standpoint, but I don't see the much of the value from the consumer standpoint.


Jean:

You gave me another opportunity to chuckle because I tried to get human service from Intercom. It's not a small thing. And whether it's there for operational efficiency and cost saving reasons, or it’s helping users? Because by the time you’re actually consuming SaaS product and you have a reason to contact, I guess at that point a human service is what they are looking for.

So I appreciate the thrust of that question, but that actually is a very interesting point. So when is just a simple RPA, a simple back and forth automation, is better, and when can real AI driven machine learning do something? At what point can you comfortably say, this is where AI can be really meaningful?


Dan:

I think everything that involves pattern recognition, when we need to categorize something by its pattern, this is where AI can help. A good example might be routing the conversation, not automatically answering the question, but rather just routing the conversation to the right agent that has the skills to answer the question. But really AI can help us complete this task way better. Why? Because we can use AI for example, to look for, to understand the question, and then to direct the question to the right agent, that has the skill, the most experience in answering that kind of questions. So, that involves patterns recognition. We need to understand the pattern of the question that we are being asked. And we also need to understand the pattern of the answers answered by the agent. And this we can match those. We can provide a better experience for the users because we can direct them, direct the support or the questions directly to the right agent. And this is where we can solve the issue in the most efficient way.


Jean:

So let me flip the coin here a little bit, because we talk a lot about what machines can do to recognize patterns and understand what a human is asking for, but what can humans and users do for a machine to quickly learn to service? Is there something that end users can do, is there a pattern in how chatbots are functioning. Is there a way for human beings can speed up their learning by asking differently and so on? Do we just get rid of using articles or do we just use nouns only? Is there something I can do to get a quicker result out of a machine?


Dan:

Absolutely. I think that there's a lot that we can do to increase adoption and to help users better interact with chatbots. And this is where conversation design comes in. It all comes down to what I said earlier about this upfront contract with the users. You need to manage expectations. You need to be specific with them and to ask them questions that can help the machine better understand their users. In some cases we want to talk about the AI and then we want to think about chatbots as something that is fully capable to understand the user and to speak freely in free language with our users. Where in reality, at some point you just have to present the right question to real users. I think we have seen this as well with mobile apps because 10 years ago, or 12 years ago, we tried to force websites into mobile apps.

We try to... When we build mobile apps, we just copy paste the website into the mobile app. And then we understood that it's not the case. It's a different medium, even though we're talking mobile first, because it's a different medium. And therefore it has different use cases. And I think the same implies to conversational interfaces and chatbot. Not as exhausting as should be done being in a conversational way. Even if the technology is there, even if the language understanding technologies are at its best, there are some use cases where it's just easier to explore in the website.

So for example, I personally don't think chatbots are a good use case for exploration tasks. So if I want to shop around for fun, sometimes it's way, easier to go through the website and through incident feed of a product and play around with the filters to find what I'm looking for. If I know what I want upfront, this is where it conversational interface can be great use case. If I want what I'm looking for, I don't think chatbot is the answer for everything. It should be part of the customer experience.


Jean:

What else am I not thinking of how to use this? Open my horizon. I'm actually curious about some of the AI use cases, the employee use case that you talked about in the beginning. If you can talk about that a little bit, because I think there is some industrial examples as well.


Dan:

Yeah. So I think that the AI can really do amazing job behind the scenes. Even if it's not customer facing automation, even if it's just, helping agents to work more efficient and to provide more personalized and faster answer. So I think that when we think of AI, it's not just about 100% automation. We need to think of it as an ecosystem of tools that can help us improve the customer experience like chatbots at the first 100% automatic conversations. And then you have the conversation handed over to an agent and there, the AI continues to help the agent provide more personalized experience.

And lastly, the agent can bring back the conversation to chatbot. He doesn't have to end with the agent. For example, the user can ask for a recommendation in your example. And then they talk with the human and you may ask them few questions and then the human can hand over the conversation back to the bot. After the user, knows what they want. And this is where the user can continue the conversation with the bot, to get a faster response. And this is where the experience can really... Companies can really improve the experience.


Jean:

Right! And it's such an obvious thing, but when you put it that way, that is one of those aha, moments. Often times you call in for some kind of support, after you get that, sometimes they do ask you, would you mind filling out a quick survey... Answering few questions for satisfaction survey or something like that. And you get into that automated mode and you don't mind that. Pushing a few buttons to fill that out. So, seeing it more fluidly back and forth, from human to bot , back to human and so on, I totally get that.

And I think this probably will be my last question. So, now this is very interesting. We talked about some of the technology that goes into it. And now let's talk about some of the tools available to do this. I mean, as a business leader, as well as a developer yourself, are you satisfied with some of the tools that are given to you, to kind of manifest all this ideas. What are some of the tools that you like? You know, help some other people play with it.


Dan:

So I think in terms of the ecosystem, the industry has a lot to do, and it's not very mature yet. We have some NLU and NLP technologies, which are very advanced these days. IBM Watson, Dialogflow. Both of those tools which are great, but I think what we are missing, are tools that help you optimize the experience and, you really understand what our customers want from us and how they engage with chatbot. So I think it's still... I haven't seen satisfying tools yet, to help us solve those questions. Most tools today are focused on building the chatbot and not optimizing the experience of the chatbot. And these are great tools, but as we move forward, we going to have more tools that help us design better conversational experiences.


Jean:

Is it more of a data issue or what are some of the components that are missing?


Dan:

I think it's the data issue, but it's more of how we design the experience. It's how we get this feedback because building a chatbot or building conversational interface, it's no different than building a mobile app. And when you develop a mobile app, you have the sole ecosystems of tools that can help you optimize the experience of the mobile app. You have analytics, you have developer tools, you have many different tools out there that can help you optimize the user experience. But when it comes to chatbots, all that you have left is that NLP chatbot builders. And I think there's still a lot of room for a chatbot supporting tools that can help you better manage the feedback from the users and optimize the conversational experience. Because again, as in mobile apps, it's not a one thing, it's not that you feel, you have that idea and you develop the mobile app, you release it and that's it.

It's a process, you have to evolve, you have to see how users interact with your mobile app and then you improve the experience over time. But you need to have those tools that can help you ship better products. So I think the same applies to chatbot. Chatbot is just an interface like mobile app, like website. You need to reiterate through the product and evolve over time, improve the experience and go with your users. As you grow, you add more features, you improve the current experience. I think there's still need in the market for tools that can help us really refine the experience and help us design better experience. Especially, going through what we discussed before, throughout the conversation and not just understanding the initial intent of the user.


Jean:

That is actually very profound. And I think you should give it to the whole community in terms of building that, nurturing that ecosystem. Because, I remember…you triggered a memory here. Was it Adobe or something like this when people were so excited that they have a program that they can design a logo or something graphic with. And then, all of a sudden people are like, but it's so flat. And then you have these little features that you start building on it. We can make it flame. We can make it sing. We can make it like, whatever kind of experience you want to have. And you start having those tools pre-built in a way, so you can quickly make these. And I think that's a sign of industry evolving and maturing. Isn't it? If you are thinking of like, what's the next generation of tools that can really optimize it, right?


Dan:

Yeah, absolutely. Because as I said at the beginning, chatbots, make great promise. They must win. If you think of it, it's a great win for the company. And it's absolutely a great win for us as customers. So, I think with the technology, the point where it is now, it's only a question of when is it going to be an integral part of our lives? And I think that the industry is still has a lot to do, and there's a lot of room for innovation, but we are starting to understand that it's not that hard. We're moving in the right direction. It's going to take probably few years. I think now with the pandemic, absolutely accelerating things, but yeah, we are in the right direction.


Jean:

Awesome. Now that sounds like a wrap. However, I have a one little question I ask my guests, before I let them go, just so that we get to know them better. So I'm going to ask you a little quasi personal question. So could you please name the three things you do the most on your mobile phone?


Dan:

Yeah, so let's see. I'm going to open my phone here. So the stairs come to mind is obviously WhatsApp. And I'm tempted to say the second is also WhatsApp, but I'm going to go with probably Gmail and my Google Chrome. So I think these are the three, and maybe I have the GPS here, but I think WhatsApp is the number one app that I use throughout the day. I even manage my personal tasks in WhatsApp. I have this group with myself in WhatsApp where I use it to leave notes. So yeah, these are the students anyway.


Jean:

Yeah. God forbid you use it the standard way, the way I would use WhatsApp. So awesome. I didn't expect anything less from you. So Dan, that was absolute pleasure. Hope we get to speak with you again soon. So stay safe and have wonderful weekend. I know we are taping this on a Friday.


Dan:

Thank you very much. Jean it was a really pleasure to sit down for a conversation. Thank you very much.