Eighteen months ago, there was a lot of talk around voice and the possibilities of using human speech as a new automated channel in financial services. There’s less talk now but the technology continues to improve.
Today’s guest on the podcast is Clinc’s Jon Newhard. He’s been around technology a lot for the past 25 years and as Clinc’s newly appointed CEO, he’s juiced at the opportunity to be able to understand unstructured speech and interpret the underlying meaning of customer support inquiries. Clinc provides its customers — primarily banks and financial institutions — with the tools needed to improve their customer support services.
I’m not a founder. Investors brought me in to be CEO. That’s sort of typical as you go from founder stage to more scalability. I’ve got a couple of highway miles on me. I spent about the last quarter century in technology businesses — from stuff in Silicon Valley to some bigger companies like Siebel Systems and SAP running individual businesses. I was a CEO of a company in Houston from 2013 to 2018. Sold that and stayed on with the acquisition, and then moved on to Clinc a few months ago.
I don’t think I’ve ever seen a bigger idea. I’ve been telling friends that Clinc will only sell to companies that want to interact with their customers by voice, which, of course, is like, a TAM of the entire planet. If you talk to a lot of technologists, they believe that in five to 10 years time, voice will be the primary way that we engage with technology, which is not hard to wrap your head around. Most of us speak a lot faster than we can type, let alone thumb on our iPhones. It’s just a much easier way to get information, to get things done.
If you think about voice from a design perspective, it obviates the need to think about how many links you need to get to in order to do the thing you want to go do. So if I want to transfer money on my Bank of America app, I’ve got to click through about five different hyperlinks. If I want to go do it on US Bank, which uses Clinc, I just say, ‘Transfer 100 bucks to Zack,” and it goes through. The second transaction is a lot smoother, easier. And sort of more to the point that the first one. Less time for the consumer, cheaper for the bank. Everybody wins.
Conversational AI making a comeback
I think there’s two things. I think one is that trend of hype and then quieting around a technology is not atypical for the way technology is adopted, generally. There’s a second piece which is, in the last six months, COVID has had an interesting effect on the market. What we found is individual banks moved more slowly, but the market accelerated, which is a very weird dichotomy. And so what we’re finding is that the level of interest from major banks has actually spiked up considerably.
We’ve deepened our pipeline with a number of really large retail banks, primarily here in the States. They’ve kind of looked at the first wave of digital assistants, kind of learned the lessons, and are trying to do that second mover advantage piece very effectively.
I’ll give you an example. You mentioned Bank of America’s digital assistant, Erica. So the data that I’ve seen publicly on Erica is that it took about four years to build it from scratch and $300 million or so to go do it. The data that we have is a little bit old.
I want to make sure that that I don’t overstate the case: the negative containment rate on Erica was about 35% or so. Containment rates basically mean I engaged with Erica, I could do the things that I want without having to either go into the app, a call center, or whatever else. So 65% of all activities could be resolved within Erica.
US Bank launched a voice forward digital application with Clinc over the summer. A few weeks into it, they’re getting 90% to 92%. They did it in about a fifth of the time — 10 months instead of 48. And they did it for 1/20th of the cost.
Technology marches ahead
The technologies have improved. I think the bigger piece is the banks wrapping their head around how best to go do it. The banks are realizing that there’s stuff off the shelf that they can plug and play into the rest of their enterprise architecture in a way that’s not onerous to the other things that they’re doing on the back end, and is complementary to some of the other digital transformations they’re making inside of the bank. It just makes a whole lot more sense.
Building a conversational platform
What we found in talking to customers is there’s a lot of things that most banks want to do. Building those competencies out of the box just makes the ability to implement conversational AI that much faster. We’ve already built a small library of competencies, so that if you’re a major bank, or even a digital upstart, you can go ahead and use those competencies, largely off the shelf, and just be able to move faster.
If you’re a large global bank, you have a lot of IT staff around. If you’re a credit union in Piscataway, New Jersey, you probably don’t have a large IT staff, and the IT staff that you have is probably a little bit more quotidian. So what we’re trying to do for those folks is work with partners like Fiserv and pre build the integration to the Fiserv stack, so that we are literally a plug and play.
A bank like US Bank uses about 40 competencies. Some of those were things that Clinc built off the shelf, and some of them were things that they built themselves. Some of those things would I would categorize as sort of informational, like, tell me my balance. Some of them are transactional, like, pay a bill, or transfer money using Zelle or the equivalent, or move money from my checking to my savings.
And the interesting thing, too, is they built a bunch of competencies, and when they tested internally, they found a lot of people were asking the digital assistant, what can you do? Which they didn’t have an answer for. So they built a competency around it. There’s nothing like putting something out there and using it to understand what consumers really want, and being able to be super nimble about those sorts of things.
Who uses digital assistants
The demographic is actually fairly broad. The interesting piece is how it’s used. For the digital natives, there’s a different way of using it, and a different set of expectations for the technology. My parents know Siri. But for them, Siri is like, go play Bob Dylan. Right? That’s their engagement with Siri, while a younger demographic is having a conversation with the virtual assistant. Language can be super sloppy and it’s sometimes a heavy use of slang.
Voice’s role in the future
Voice has a number of advantages. There are also some distinct disadvantages. A lot of the banks that we’re working with are being pretty thoughtful about that. For instance, if you’ve lost your credit card, voice may be a way to kick off something but not a way to continue to interact. You don’t want sensitive information like your last seven credit card transactions to be spoken aloud in an airport. SMS and apps are better for that.
There are just some things that actually don’t make sense for voice and things where voice is generally better, but not always. And so we’re seeing more thoughtful banks do things that is called voice forward, meaning that they want to lead with voice because it’s just so much more efficient and a better experience. But where it doesn’t make sense, they give the option of doing things in a different way, like having the list of transactions pop up on their phone.