Name: Matt Bain | Spark NZ
Uploaded: 2024-02-27T10:30:00.000Z
Duration: 16 min 4 s

Clips
More from MWC Barcelona 2024

Matt Bain

Marketing and Data Manager

Spark NZ

play_circle_outline Mobile World Congress Day 2: Unveiling Brain's Commercial Version for Telco Licensing and its Role in Automating Marketing Campaigns - Interview with Spark NZ's Matt

play_circle_outline The Power of AI in Snowflake: Enhanced Data Management and Analysis for Optimal Results

play_circle_outline Efforts to accelerate AI development in the country

play_circle_outline Taking friction out of the customer experience through data and AI

39:57

22:43

18:40

20:52

19:26

20:05

Info

Matt Bain | Spark NZ

Matt Bain

Marketing and Data Manager Spark NZ

(upbeat music)

>> Good morning and (speaks in foreign language). We are here live with theCUBE's four-day coverage of Mobile World Congress, action-packed day two. My name's Savannah Peterson, joined by my co-host Dave Vellante. Dave, day two's rocking.

>> Oh, it is. Wall to wall, as usual.

>> Wall to wall, wall to escalator, all things awesome. Speaking of the global nature of this event, our next guest perhaps traveled farther than anyone to be here, all the way from New Zealand. Thank you so much for joining us. Matt, how you doing?

>> Awesome, yeah, hardly any jet lag.

>> Holding up? >> I'm getting there.

I'm done up. >> You're bringing the energy.

>> I'm here, I'm here, I made it.

>> You have arrived, Spark NZ, biggest mobile company, biggest telco in New Zealand. What does this show mean to you?

>> For us, I mean, AI is my thing. So for me, coming here, seeing the scale or how people are thinking about that and just the momentum on it is awesome, I love it.

>> Absolutely. Spark just made an announcement, you announced Brain. Tell us about Brain.

>> So we've got, so Brain, just a bit of facts here. So it stands for Build Robust AI for Next Best Action. So the squads that build our data products name them, so that's how, now it's Brain. And so Brain is two things, it's an internal capability we've spent about four years building at Spark to automate our marketing function. So the thing about telcos, every year, maybe 12% of your market's in-market for a phone. So on a monthly basis, 98% of your marketing's probably wasted. 'Cause people aren't shopping. So we went, if we could build a system which sat across all of our mar-tech and could predict when a customer was in-market, we could 10X conversion rates, we could really transform the way that marketing works. So that's what we've been working at and we've had it running for a while. It runs 90% of our campaigns now autonomously without humans and when we went away for Christmas in New Zealand, which we do, for the summer, it's summer, people aren't there but Brain's running the marketing for us, for our customers. So that's the internal instance. We've just launched a commercial version of it which can be licensed by other telcos and we've got our first telco customer running that currently. So that's the other exciting piece of this.

>> I love having marketing folks on from the telcos 'cause when you read the headlines, it's all about the CapEx and how we're trying to drive costs down and we need new regulation and so when you talk to somebody about marketing, they're talking about innovation, Brain, things like that, how do you drive productivity, new revenue, so from your standpoint, what are your business drivers that are allowing you to sort of fund your future?

>> Yeah, so CapEx is still important. Don't worry, that doesn't go away.

>> But it's not your job.

>> Well, part of what I have to do but we really focus on the customer. So any CapEx we spend is around how do we improve customer outcomes 'cause we know that improves lifetime value, improves our profitability. So we've found that by implementing this Brain platform, like that's 70%, it's one seven, 17% compounding efficiencies in marketing. And we spend a lot on marketing, so that's a big number. So our focus is if we can anticipate the needs of our customers, turn up with the right products at the right time in a way that they feel comfortable with, that keeps them happy, they churn less and they spend more. So that's really the focus.

>> Well, and it feels like a personalized experience.

>> 100%. So this system works across all your existing stacks. So if you've got Adobe or Salesforce and Snowflake and all that stuff, it doesn't replace anything. But all that stuff is an industry to connect all that together. That's constant as things evolve. By abstracting this intelligence above that layer, what you do is enable those systems to do what they're supposed to do and the AI and the intelligence and the reinforced learning to work above that and allow them to understand how they should act in real time as a customer hits one of your channels.

>> So a lot of times we think about the over-the-top vendors as having all the data and all the insights, but you're a data guy. You run a data team. How are you using data to drive customer value? What insights are you finding? What have been the patterns that you've noticed since you really kind of got into the whole data game?

>> I'll tell you, it allows you to do some things. Once you're acting an individual consumer or household level, segmentation's gone. Now all of a sudden you can do things that have always been really hard for telcos to do. So for example, how do you migrate all of your customers onto the right plans so they're not on a too expensive plan 'cause they see something better in market? You can't do it at scale because that destroys your EBIT. But if you can identify an individual customer at a moment in time and go, "Now's the right time to talk to that customer." Say, "Hey, by the way, we've got a better plan for you." Reduces the uphill on that plan, but at a household level makes them happier. Now you've locked in the CLV. So we've got something called the Made For You Review, which is a real-time program which identifies individual customers in need in real time and then gives them the right message to make them feel, I guess, valued in a way that wasn't possible before. So those are the sorts of insights you get once you can operate this at scale.

>> So you're predicting there the timing in part.

>> Savannah: The churn timing, basically, yeah.

>> So the individual or the household, the actual need, whether it's a service need or an upgrade or a cross-sell. The time, so on a daily basis, if you're not in market that day, we don't talk to you. If you're in market tomorrow, now we talk to you. We can predict that. So we can predict when you're moving house. We can predict when we think you're going to upgrade your phone, all these things. And these are all inference models. So these are inference models that are constantly learning. Every time it tries something, if it does or doesn't work, it learns for the next customer.

>> And on something like a family plan, you'll increase the contract value. Maybe while the margin just drops to the bottom line 'cause you got the infrastructure. And so increase the contract value, keep the customer, and give them a better deal.

>> I mean, if you're a marketer, acquisition costs are high. The cost to acquire, to rip a new customer, market's largely captive. So to take a customer from a competitor, very expensive. So if you can lock in at a household level, large number of lines, recognizing the value of that household, giving them the appropriate value, it makes it hard for a competitor to acquire them. And at the same time, you're managing your base. And then when you're targeting a customer to acquire, you're valuing them at a household level as well. So you know exactly how much you can spend for each individual customer.

>> I mean, I hate switching. I don't ever want to switch.

>> I have had the same cell phone number and plan for since I was 12.

>> When I walk into an AT&T shop and they pull me up, they go like, "Oh, Mr. Vellante, come on in. "Do you want a water?" Right, I mean, 'cause I've spent so much money with them over the years.

>> How generous of them.

>> No, because I'm loyal and I don't want to switch. You know, it's too much of a pain. But to get me, but I've started to look, like 'cause there's maybe some interesting options out there. So how do you attract? I mean, are you saying 90% of your energy is on keeping those existing customers? But you still have to grow. You got new people coming in.

>> Once you have this level of visibility on your market, you can start to arbitrage. So marketers think of a funnel. You're not interested in the top. You're looking and shopping at the bottom. At the bottom of the funnel, you're acquisition and retention. Do you acquire a, you've got $100 to spend on marketing. Do you acquire a customer? Or do you save when you think it's going to churn? You can only do one or the other. If you've got some limitations around budgets. So the system can then understand, well, this customer has a customer value predicted of this, if I acquire them. And this is the cost to acquire. And this customer to save them. You know, here'd be the cost per save. At the extremes, you can start to arbitrage your budget across. And traditionally, those are very separate budgets, different departments.

>> Different channels?

>> Yeah, and so it will also decide which channels to talk to a customer through, so yeah.

>> Wow, that's awesome. Okay, let's go a totally different direction 'cause you have had quite a few announcements recently. One of the things that you have just set up and just successfully done is your first satellite text message for emergencies. Tell me a little more about that.

>> So, New Zealand has recently had a lot of natural disasters.

Floods, you know, storms. >> Volcanoes.

>> Not volcanoes, storms, earthquakes. So, and in those times, one of our key priorities is how do we keep the country connected. So, people are unsafe, 111 needs to work. And some of the issues around it might be something like, you can't get power in a region, so it's hard to power a network when there's no power. So, satellites emerging, it's not here at scale yet, but we see that at those times particularly now, there's real need to be able to keep connected, even if it's not 24/7, but it's like, as satellites pass over, you know, we need to be able to receive a text message. So that's where we're really trying to leave that area out and focus on how do we keep New Zealand connected at those times when, you know, disasters happen and they're happening more frequently.

>> It really resonates with me. I live in California, we just got hit with both huge waves and an atmospheric river that took down our power lines, tree took down the power lines, and also took down the cell towers. So I was literally incapable of working, and so when I saw this announcement of yours, I thought, way to go, Kiwis, that's exactly what I love to see.

>> We're not a very densely populated country, like there's a lot of places a long way out of the way, which take a long time to get to, so it's really important that we have those kind of services in the market.

>> So how real is AI to you? I mean, I know it's real, but so the new AI, we've had ML around forever, and then, you know, ChatGPT, it's funny, a year ago at the show, we were talking about chat GPT, but it wasn't as frenzied as it is now. So I'm curious in terms of how you're applying it to your business, you're a Snowflake customer, correct? Okay, so we were at Snowflake Summit in last June, Monday night, Frank Slootman was up on stage with Jensen, which is, you know, anytime you get Jensen to go to your show, it's a big check mark. What's really interesting, and Jensen was saying, we are going to superpower data inside of Snowflake.

>> Yeah.

>> Is that been your experience? Where are you in that AI journey? I'm sure you were doing AI before ChatGPT, but take us through that journey.

>> So yeah, so first of all, like, Snowflake's super important to us, because it gives us the ability to scale, you know, the data we hold, and access it in real time, in the way we need it. So I'm not a technical guy, but I know that that's a core part of what we need in place to be able to do what we want to do. So the stuff I was talking to you before, think of that as structured data, predictive analytics. That's one stream. Second bit is unstructured data, generative AI. So those are two for us, those are two different work streams for us. We have enterprise-wide allocation of budget across use cases for each of those different types of AI. And Brain is the top one. So it's the predictive analytics, structured data. We're going as fast as we can into the unstructured data piece, too. So for us, that shows up first in places like call centers. The reason it turns up there is you have a very clean data set to train the AI on. Unstructured data, by definition, is unstructured, and therefore you don't want contradictions, you don't want ambiguities. Call center databases don't have that. They have to have the prices right, the terms and the conditions right. So in a matter of weeks, you can spin up a conversational AI for your agents, which really shortens call times in call centers. That's the first application. We've moved it into HR, so now our employees are emailing an HR bot, which is an AI-generated responses to HR questions. So those are the first kind of early use cases we're seeing. But also, for applications that use the data inside Snowflake, Brain has a thousand traits. People are spinning up new ones all the time. One might be price sensitivity, one might be affinity for gaming, making it up. Developers aren't the best coming up at naming things, and they're not as consistent with their descriptions right, so you could have people working on duplicating across the organization the same trait. So we're using AI to start to manage that, and go AI should name that, generative AI should manage that, because much better than humans. It actually gets through a year's worth of work in a quarter. So that's where we're starting to see it, both consumer-facing, but also in the back end.

>> And that unstructured data is inside of Snowflake as well? Or are you using a different data store for that?

>> Technical question. I'll come back to you on that.

>> Dave: You don't know, okay, but so, okay.

>> My understanding, probably in Snowflake, 'cause that's kind of our core, the core repository for our data.

>> So I would think you're trying to do as much inside of a single data platform as possible, because that's kind of the problem that people have, is everything's stovepipe. That's kind of the value proposition of Snowflake.

>> So we're talking about system efficiency, let's actually talk about human productivity. You recently did a study about productivity and digital technology, and how it can help folks in New Zealand. Can you drop some knowledge on us about that?

>> Yeah, so, I mean, New Zealand's a small country, it struggles productivity-wise versus the rest of the developed world. And each time there's a wave of technology, like a new generation, 4 to 5G, or AI hits an economy, you see a potential uplift in GDP, which obviously helps the whole country. So it helps health outcomes, it helps people's livelihood. So we're really passionate about, as one of the biggest companies in the country, we're passionate about how can we leave that out? How do we work with government and other private sectors to accelerate the rate at which AI is developed in the country?

>> It's great that you're committing to being a part of that effort. One of the things that I find different about New Zealand than other nations, and I'm a little bit biased as an honorary Kiwi, is the collective, collaborative effort between government and institution and the tech companies to make the magic happen. What would you say is your general pulse read on how Kiwis are feeling about AI right now? Is it hyped in a good way? Is it doomers? We've gotten to talk to a lot of different people all around the world, different attitudes.

>> It depends who you talk to. It's a bit of both. It depends what's in the paper this week, I guess.

>> Savannah: What's in the Herald?

>> Yeah, I think business is interested, but a lot of businesses don't understand how to get started on it. We're a country of small businesses. A few big ones and a lot of small ones, and so part of that is how do we democratize what a big company like Spark can have and do and make that accessible to smaller businesses and accelerate their journey, 'cause they wouldn't necessarily have the benefit of the skill sets or the IP to get started or accelerate.

>> How would you describe your vision for your consumers?

>> So for our consumers, it would be to take friction out of, like if you think about a telco, I'm not sure anyone really wants to have a relationship, a meaningful relationship with their telco or their airline or their electric company. So we realize that we're a service and taking friction out of any experience is going to be the end goal. So every time you turn up on a touchpoint with us, if you have to, then we should anticipate what you're there for, we should give you the choices and allow you to make the decision really quickly if you have to make one, swipe and a click and move on with your life. So it's really about streamlining the engagement people have with us.

>> And data and AI enable that?

Is that- >> It's the only way you can do it. When you've got millions of customers across a large number of different types of products with totally different needs, like in a segment, it assumes an average person, no such thing. So the only way you can identify and target the needs of an individual customer, no human can think about millions and millions or billions of rows of data and interpret that. So it has to be AI and that's why we're so excited about it.

>> It's a really good use case and everybody wants a better customer service experience across industry, whether that's with your telco or with anything else. Last question for you, what do you hope to be able to say at next MWC that you can't say yet today?

>> I'd like to think that we've gotten closer to putting in the hands of our people, the ability to remove all of the busy work that they have to do, so they can focus on the high value stuff for our customers and our customers are getting the experience a bit closer to what I explained to you about two minutes ago.

>> Love it. Matt, thank you so much for being here.

>> Thanks Matt >> Thanks for having me guys.

>> Especially after a 24 hour travel journey to get here.

>> Matt: Very energizing.

>> And shout out to all of my New Zealand friends. Dave, always a pleasure to have you on my right. Thanks for letting me do this interview and thank you for tuning in from wherever you are in the world. Hopefully down there in Eritrea as well. My name's Savannah Peterson, you're watching theCUBE's four day live coverage here at MWC in Barcelona. Thanks for tuning in. (upbeat music)