Mobile Interactions Now Podcast: Testing AI-Powered Point Solutions and Measuring the Results

Mobile Interactions Now Podcast: Testing AI-Powered Point Solutions and Measuring the Results

In Part 2 of Mobile Interaction Now’s Conversational AI podcast episode, ContactEngine CEO Dr Mark K. Smith talks to tyntec‘s Jean Shin about the signs of when a company is ready to test a point solutions powered by AI – and how to choose the right communication channel to use.

Listen to the full podcast here

Jean:
In our previous episode we started discussing how AI can remove a lot of friction and stress. You know, more often than not, I count myself as one of those people who get frustrated and agitated by poor service. A recent example, I’m getting annoyed by my bank back in New York that I opened an account with before I moved to Munich. And the customer services is insisting on calling me and, or better yet asking me to call them back. And despite the fact that you know, time zone has changed and you know, I’ve already asked them to write me email or message me and it’s not happening. So how far along are we really to get that kind of benefit into mainstream economy like banking, in my case?

Mark:
Very interesting question. And banking exposes some very interesting challenges. I was with an insurer actually a few weeks ago. A man who was, not dissimilar age to me, sort of late forties and he said, and I quote, “There is software in this company which is as old as me.” That was very difficult to understand how that can be true and he was probably exaggerating a bit, but a lot of financial institutions, the sort of core brain of what they do, can be somewhat archaic. And if it is that desire and the ability of smaller companies such as mine to actually solve their problems, there is a disconnect between those two things. If there isn’t data availability and in a sensible way that’s presented in a, you know, some beautiful API to allow, you know, my client service to pick it up and fire off secure messaging to people about this, that and the other, then I can’t do anything.

However, it is as clear as, as you know, as night and day that you have to do this. Your example with the, with the bank in New York and you’re in a different time zone. Well I know a bank in, in mainland Europe who did something which really was quite a surprise to me that they thought like this and it was a huge pleasure and they said this, if you can test the best time of day to reach people, and that’s a relatively simple thing to do, take it 10,000 customers, communicate with a thousand of them at 30 minute increments during, you know, acceptable hours or maybe seven in the morning till eight at night and then draw a graph and see where most people are receptive. Then if you prove that I will change the shift patterns of my people to match my customers desire and availability to respond. Now that is customer service.

Calling you because it happens to be two o’clock in New York when it is a different, I’m not sure my time zones are with Munich, but because it’s a, you know you’re in the pub then maybe? I don’t know. But I probably got, I’m making terrible assumptions there. But let’s say you were, it’s not a good customer experience. Being empathetic to the fact that a group of customers are in different time zones and then either using automated communication because that works for you or changing the shift patterns of people, it is just got to be like that because the bank you’re with has very little difference in the service from a bank in Germany or a bank in San Francisco or a bank in London. So customer service will be the differentiator. They either get on the train or that train stops.

Jean:
Little side question, cause you just mentioned automation. What would you say for people who are thinking AI is just a way to get to automation?

Mark:
Yeah. I don’t have an apocalyptic view of a sort of dystopian future where I’m just, you know, fed by tubes and machines and doing everything for me or maybe I’ve taken over the world and firing large guns. I mean I don’t, I don’t think that’s the way technology works. My answer to the question of automation and the relationship between the carbon and the silicon, the human and the machine, is captured rather nicely, no one’s ever really adequately explained this to me, than by this thought. So in my country we have the most people who have ever been employed ever. Our population’s increased, which partly explains it, but we have something in the region of three and a half, 4% unemployment. Now there are some economists can be brutal people. There are some economists that argue that that is in fact zero unemployment because there will always be a small group of people between jobs or what have you.

So how do you explain this then? Today is the most technically advanced we have ever been by definition. And yet my country, there are more people employed than there ever have been. And that’s with the progressive tax system and the minimum wage. How’s that happening then? Now I’ve lived quite a long time and I’ve seen these sort of apocalyptic views of what technology is going to do. Robots in the car industry, you know, AI in medicine, all this kind of stuff. Well it’s not happened, right? It didn’t happen. It changed. People started doing different things, but people should be used where people are best and machines used where machines are best. You know, to be honest, I’d rather my car was made by a machine, not some guy on a Friday afternoon who’s not feeling so well. Right? So there are moments where automation is absolutely fine and there will always, it will always be a synergistic relationship between the machine and the human.

Jean:
Okay. With that bright picture, let’s try to help businesses to get started a little bit. You mentioned looking at a transactional level versus differently. And so is there something of a good indicator how ready they are to embark on integrating AI into their customer experience?

Mark:
Well, companies certainly seem to be open to doing this. I think I, I’m fearful sometimes that there are a lot of sort of mid sized vendors that will, I’ve come across this quite often where a company will say, well I’ve got this magic widget that can do these things. And, and I’ll say, well, can I, can I see your magic widget? Can I just test the magic widget out please? And they said, yeah, of course you can. All you need to do is sign this contract for my most recent one for fairly ordinary piece of software, it was about a quarter million quid year. When I’d stop laughing, I then said, you know, the [inaudible 00:07:46] are not testing this out and seeing if it works or not. That’s what I actually want to do. So I think for, for corporates, they need to, they need to make several decisions.

They need to, they need to sort that data out. There have been, and I can remember one occasion with a large travel firm that’s in the UK who is in something of a mess who after two hours I stood up and said, I can’t help you because I didn’t have a CRM system, so how can I communicate with people when you can’t communicate with people? So I gave up there. So they’ve got, they’re going to get that data in some kind of order. But even that can be dealt with because companies can take data from different silos and bring it together. And then it’s just test, test to see if it works. Take a control group, take another group and see if the use of that piece of software that’s mine or somebody else’s actually makes a difference. Prove that difference. That then tends to lead to the justification for doing something.

We’ve just picked up a contract yesterday for several million dollars a year where we spent the last six months just proving again and again and again against control groups that we are delivering at 10, 15%, 20% improvements in a process which is delivering 50, 60, 70, 80 million dollars of savings. And that’s the way to do it. So they started cynical, they then became disbelieving, we then proved it. They then said okay, you’ve won. That’s what they need to do. And people like us as are speed boats as well, which is helpful because we can do whatever we want immediately.

Jean:
So it sounds like they don’t necessarily have to have a single view of the customer just yet.

Mark:
No, you can have a sort of a point solution. So one of the things we did with a bank recently was around an up sale conversation after a successful mortgage application. So basically asking people sympathetically, because it’s not a monopoly position, asking people after they’ve got their mortgage with bank X, you know, would you like to take buildings and contents insurance with us? So that’s a very discrete subset of the bank, which is eminently available to us because by completing a loan application for, you’ve gotten the money, you’ve got the contact details, and you’ve got the permissions to do it. So you don’t need to boil the ocean, eat the elephant in one go, all those hideous phrases. You just, why would anyone eat an elephant anyway? Oh you just need to demonstrate a small base of your customer base to deliver something exciting and then say, look at the implications of this across the piece and then you justify your existence.

Jean:
Let me just try to capture that because what you just talked about is sort of knowing your short term goals that you can realistically achieve versus maybe a more long term destination you can get to. So can you, can you just help us frame those two things so that we can try to understand and help the motivation internally.

Mark:
I’m trying to think of a decent analogy for this. I suppose drug discovery is not a bad one. So most pharmaceutical companies will spend an enormous amount of time testing and it takes a long time to do that and can be very expensive. But at the end of the day they end up with a product they can sell. So it starts with some simple experimentation. The experimentation goes to bigger sample sizes, bigger groups. It goes through regulatory hurdles. It goes through further R and D, further refinements, different models of efficacy, different animal testing. Perhaps if that’s still done, which I fear it might be, until such time as a drug is then available to make people better. Now the same sort of approach applies with the, with so called new technology. And one would argue that AI has existed in principle for about 75 years actually. So start small, test principle, see if it works, roll out more safer works, test principle, move towards the product at the end that delivers the benefits.

Jean:
And this is going to be my last question and I want to shift the focus from the consumers point of view. Do we know anything about how they are deciding what they prefer to use, you know, is it privacy, security, whatever will, what do they really care about?

Mark:
That’s an interesting question. And one of the things that I find most fascinating about my career is I’m old enough to have began working in an organization where I had a typing pool, and I’m not joking, I used to have a Dictaphone, which is a small plastic rectangle with some tape in it. I used to speak into this thing and my words had been magically recorded and then placed in an envelope and man called Frank would take that and he would move it to a typing pool and maybe 48 hours later I’d get back a piece of paper, which I’d correct. And another day later I got that back. And then I have to find an envelope and a stamp and I put it in the post, Christ. This is not that long ago, right? This is 25, 27 years ago. Not even that long ago. So 25, 27 years ago, your means of communication were pretty much a letter and phone call. Email [later]. But I’m also old enough to remember my late twenties when I asked a group of well-paid chemical engineers in a room, how many of them had email addresses and two hands went up. So, it was slow. It’s slow progress. You know, the postal system, 200 years, you know, telephony, a hundred years. Email, maybe 30, something like that. Text messaging, sort of 25 ish. And then now it’s gone boom. So you have hundreds of instant messenger platforms. You have people using communication methodologies that, you know, are almost unbelievable. You have thousands of different channels. If you choose to dictate terms to your customer, then that is wrong.

It’s not just the app, it’s not just email, it’s not just a phone call, it’s not just a text message, it’s not just WhatsApp or Facebook Messenger, nor should it be. I’ll come back to that one day, but nor should it be. It’s not just Alexa or Google Home, et cetera, et cetera, et cetera. It’s arguably all of those. And then you can do one of two things. You can, god forbid, ask the customer which channel they would prefer. Not very hard, why not do that? Or you cycle across the channels, which is what we do, until such time as they respond on one and then we carry on the conversation on that channel until such time as they choose to tell us that they don’t want to communicate on that channel.

Or we can do some really interesting things with language, this is lovely, where someone will have opted in for say a text message conversation and, we communicate in any language of course, and then we write to them in English and they respond to us in say Spanish. We can recognize that. Go back to them and say, would you like future communication, and we’d say it in Spanish, would you like future communication to be in Spanish? And if they come back and say, si, presumably. Then we … Englishman display is terrible language knowledge. Then, then we can carry on that conversation in that language. So it’s all about customer preference. Either ask them or cycle across until such time as you discovered. If you do that, people will be happy.

Jean:
Awesome. And thank you for that. But before I let you go, I just have one little fun we have with our guests and I get to ask what are the three things that you use the most on your phone? And be quite honest here.

Mark:
Right.

Jean:
Almost friends.

Mark:
Okay. I’m, right … Lots of layers of answers there. So I’ll give you a straightforward answer. Start with music, right? That’s the most important thing to me. I spend vast amounts of time listening to certain types of music very loudly to annoy people on trains. So music first. I’m afraid next up is email. Okay. I’m a great believer in correspondence that has more than a few characters. And I think well articulated arguments normally take a few sentences. So email’s very important to me. And then I actually quite like voice. I quite like … It is said that listening to someone speak, 80% of the information that a human understands is not the words they’ve said. So that can be the, there’s a word called prosody, which is all the sort of underlying subtlety in voice. I’m doing it now, I’m slightly slowing down, I slightly accelerate, I’ll change the volume and pitch and tone. And you can tell tiredness in voice and you can tell happiness of voice. Salespeople are told when they speak on the phone to smile when they talk because humans can tell if someone’s smiling when they speak. They really can, right? So there’s all sorts of value in voice that I find tremendously helpful. So it’s those three things. I don’t do social media. I have, I have all the accounts very early on in all social platforms. I’m just not very social.

Jean:
Yeah, but you’re a heavy user of communications tools, obviously.

Mark:
Yes.

Jean:
Email and voice.

Mark:
Yes.

Jean:
So there you go. So simply, awesome. I thank you.

Mark:
Pleasure.

Insights Team
insights@contactengine.com