How should startups think about AI and customer service?

“Definitely. Yeah, I think the, you know, I'd say like two things. Like, one is I think we're seeing a very new role come up within a customer service organization. It's a, you know, customer service for the longest time was a people business. It was like, how do I find great people who represent my brand? How do I hire them? How do I train them? How do I retain them? And how do I, as that team grows and we scale as a business, how do I keep bringing those people in and sort of like scale through people? I think we're hitting a point where a large portion of those interactions and that design experience is happening through technology. And that's taking a very different role that sits within operations. You know, it's, it's, we have all these like, you know, for years I've worked with these like highly creative operators and admins that are technical enough to be dangerous and they know what they need to do, but they need the right tools to be successful. And I think we're really seeing that progression happen right now. You know, these sort of like, everyone calls it something different. You know, I was just talking to a customer yesterday. They call it like a service designer. I've heard it called like an automation engineer, an AI architect, but it's this sort of like you're part data analyst to sort of like understand where the problems are. You're part, um, sort of experience designers, sort of thinking holistically around, you know, the journeys that you're trying to design for a customer, almost like a product manager type role. And then you're part like system integrator, like how do I connect into the right systems to get things done for my customer?”

“And I'm not going to, I'm not going to start chat. I just don't have time right now. And I think service is very much like that in that, you know, that one customer is doing you an amazing service by actually reaching out and going, this is the problem I'm having. Because behind them, there's a whole bunch of other customers that had that same problem that are disenfranchised that you can go then solve for. And the thing that's really exciting to me about the personal automation side is like, as friction to sort of like interact with a brand decreases, I think we're going to see actually a lot more interaction and a lot more sort of feedback to brands. So you could imagine as personal automation comes up, like actually the number, number of service interactions being handled by a brand, like doubling, tripling, quadrupling over the next few years because those friction points are really being reduced and it doesn't have that same time commitment to actually sort of like give feedback or deal with that sort of return that, you know, just has been sitting in your closet forever or whatever it is. Right.”

“Yeah, I mean, there's a lot of sort of counterintuitive data in service, so, and it's very unique to each business. So like a good example was one of the early data products we launched was something called Satisfaction Prediction. We had all these customer satisfaction surveys, but we also had all this operational data around, you know, what led to that outcome. And, you know, back then we'd build like one logistic regression model per customer. But you could see actually what weighted higher for different customers. So the red thread was time between responses and first reply time highly correlated with positive outcomes. But then there were sort of customers that had counterintuitive data. So we worked a long time with a customer, Upwork, and they work with these freelancers, people that don't necessarily have a team around them. And most of the times, the number of touches on a conversation would be highly correlated with negative satisfaction. But for them, like the more they engaged in conversations with their customers, the higher the likelihood was it would actually be a positive outcome. So, you know, every business was a little bit different. There were sort of like red threads that sort of like patterns that were highly correlated with, you know, positive outcomes, but definitely businesses that were sort of unique in the way that, you know, customers wanted them to engage as sort of like with service as the face of their brand.”