Part 2 of our Listen, Learn, Love trilogy. There are now over 4 billion people using social media. And those billions of people represent a goldmine of unbiased, unfiltered data you can learn from… if you have a powerful enough AI to help. Today we talk about the potential of AI-enabled Listening.
All right, it’s time to rock today’s podcast. Welcome to the CXM experience. I am Grad Conn CXO at Sprinklr, and I’m here to talk about customer experience management. CXM.
So in yesterday’s podcast, we talked about this idea that there’s a system for being highly connected to customers… a customer centric approach to digital transformation. And we call that Listen, Learn, and Love.
And yesterday, we talked about “Listen.” Listen being, how do you find out what people are saying across all the modern channels. And modern channels are everything that’s been invented in the 21st century. So those are social platforms like Twitter, and Facebook. Those are going to be messaging platforms like WhatsApp and WeChat. It’s going to be things like blogs, of which there are millions, and forums, of which there are hundreds of thousands, and review sites, of which there are thousands. And you need to pull in all that information from all those places to really understand what’s going on out there. What’s amazing, is there are more than 4 billion people now actively using social media. And the usage of social media is increased rapidly over the last few months in 2020. I don’t know why. I can’t imagine what’s happening.
But anyway, something’s happened to cause people to increase their usage by double digits. And what that means is most people who are online are also reachable across these platforms, and it’s their primary way of communicating and interacting. If you’re not on those platforms, if you’re not in those places, then you’re not going to be where your customers are. And the result will be: you won’t know what your customers are thinking, and they won’t be able to see what you’re selling. So that’s “Listen” in a nutshell.
So let’s talk about “Learn.” All this content coming across these platforms is unstructured, and unsolicited. And what that means is that no one’s asked for it, which is great, because that means it’s unbiased. If I asked you to review me, if I asked you to give me feedback, I immediately bias the system. It’s something known as the Heisenberg uncertainty principle. It comes from quantum mechanics, that the act of measuring a system is also an act of altering the system. If I watch someone write something down, that they wrote down on their own, because they wanted to, because they were talking to friends, or because they had some passion, that’s something I can really trust.
There are millions of these conversations going on. In fact, there are billions of these conversations going on — the data store in Sprinklr is 16 petabytes. You can’t even imagine how much data that is. The average human can’t really process that.
I was on a great group call today. We had our advertising agency council meeting today at Sprinklr with representatives from many of the world’s biggest and greatest agencies all talking about how they’re using Sprinklr: to collaborate with clients, and run advertising, and do other interesting things in the marketplace. And one of the people said they’re seeing a lot of comments on ads. The ads themselves get a lot of comment. And they need to track those because those comments are not always complimentary to the ad. Sometimes the comments will un-sell, what the ad is selling, and sometimes they say things that are great.
But to track it all, and to stay on top of it all, is almost overwhelming. There’re thousands and thousands of comments across thousands and thousands of ads. It’s really hard. So, what do you do? And this is where Sprinklr is incredibly powerful. Because Sprinklr realized about six years ago that the volume of information was quickly overwhelming community managers and human beings. And the only way to solve it would be with artificial intelligence. And the AI work that started many years ago is now a dominant part of what we do in R&D. We have thousands of people who are using the product every day and training it. We’ve got petabytes of information that are part of the data set that make them more intelligent. And we’ve got a sophisticated group of PhDs and academic institutions we work with on algorithms. And there are algorithms that sit across 90 languages now, and 40 different categories.
It’s important to have it by category because things don’t mean the same in different categories. One of our clients, for example, is the Mayo Clinic. And for Mayo Clinic the word “sick” has a very specific meaning, an important one. Another one of our clients is Red Bull, and for Red Bull “sick” is a very different context. And so you need to be able to know that “sick” is cool for Red Bull, and bad for Mayo Clinic.
Another one of our large clients is Microsoft. And Microsoft has some really challenging brand names: Surface Word, Windows These are really tricky things to separate out. And so being able to use artificial intelligence to understand the words that are adjacent to that, so that you can understand when someone says Word in a Microsoft context, that they’re actually talking about Word the program, and not just “words.”
AI has become the way that you learn. And we actually have seven different AI layers and filters that we process everything through. We’ve got a great AI webinar series hosted by yours truly, and it’s got engineers from across Sprinklr talking about how they’ve implemented different features with AI. It’s a six-part series, and the links are in the show notes. So take a look and I encourage you to watch it, you’ll learn a lot about AI, and machine learning. And you’ll also learn a lot about Sprinklr and how we’re doing it.
AI does all sorts of things in addition to just listening. One of my favorite ones is smart budgeting. So what it’ll do is actually look at the way money is being spent, and then optimize it on the fly and allocate it correctly to the right advertising channel based on what’s happening from a reaction standpoint. All that is really hard for humans to do in real time. And it allows people to optimize campaigns in a compelling way.
What else can you do and learn? Well another thing that is fascinating is you can get a pretty good idea of what your brand is all about. I mean, there’s two ways of thinking about brand. Brand is what I want it to be — our brand is what we say our values are, our brand is what we have written down or etched in stone or put into decks. The other way of thinking about brand, maybe the way I think about brand, is brand is what other people say about you. Because no matter what you’ve put down, if other people don’t say it, it’s not your brand. Your brand is what others say about you.
And great thing about “Listen” is we can now find out what everybody is saying about you, or about your CEO, or about your brand, or about your people, or about your product, etc. One thing I rely on heavily in Sprinklr is the AI-based brand attributes. I can see what sentiment looks like, I can see what core attributes of my brand are, I can benchmark against other products in my categories or other areas, I can see word clouds of what people say about it, I can look at all sorts of different ways of understanding the brand and understanding what’s going on. And then take action. For example, one of our core brand sentiments is innovation. So not that surprising. But I can click on “innovation” and drill in and understand who said that and how they said it and what context it appears. And if I want to, although is there’s a lot of them, I can get down to individual messages. So all the individual messages are still there. But then they’re rolled up with AI, and AI dynamically organizes the comments into categories and attributes.
The other thing that I rely on heavily is that I can see which posts and which ads people are responding to — clicking on, sharing, etc. Typically the content that’s getting the greatest reactions is not the majority of content that we’re sending out. So, for example, we’ll send out content that has “hopefulness” as being a core attribute. But people are clicking on content about “feminism,” which was a really big topic a couple weeks ago. So “feminism” was an important part of what people were clicking on. And so what we can do is say, hey, people are clicking on this type content, we should route it more, and people aren’t clicking on these other ones that we’re doing a lot of, so let’s do less of them.
So that’s “Learn.” We’re going to come back to “Love” tomorrow — and love is a many-splendored thing — so we’ll be able to talk about that at length. I’ll probably do a few sessions on it but, for now, this is Grad Conn and the CXM Experience, and I’ll see you tomorrow.