The ability to predict demand surges and slumps in demand in real-time has gone from an efficiency booster to a necessity in a pandemic world, for everything from Uber drivers to the food supply. PredictHQ Founder Campbell Brown explains how timely data – from protests to paydays – can be crunched to forecast sudden shifts in demand.
Christina Noren: Welcome to "The Software Agents," a new podcast on how software is helping the world survive and evolve right now as told by the people making it happen. I'm Christina Noren and my cohost is Paul Boutin. Paul?
Paul Boutin: Hello.
Christina Noren: "The Software Agents" is sponsored by Cloudbees, the enterprise software delivery company. Cloudbees provides the industry's leading DevOps technology platform that enables developers to focus on what they do best, build stuff that matters. So today, we have Campbell Brown the CEO of PredictIQ as our guest. And Campbell and I got to know each other a little bit maybe I don't know it was three years ago or so and PredictHQ is a duly headquartered company between New Zealand and the San Francisco Bay area that consolidates data to help provide demand intelligence to companies based on all kinds of things going on in the world. How much foot traffic am I going to expect in location XYZ. I won't try to summarize it too much. I'm gonna turn it over to Campbell. So, Campbell, tell us about you and tell us about PredictHQ.
Campbell Brown: Hey, thanks Paul, thanks Christina. So, yeah, my name's Campbell Brown. I'm the CEO and co-founder of PredictHQ. In a real nutshell what we do is we provide a level of intelligence for business to help them unlock the why around demand. So why are we seeing a spike in Uber requests over here or why are we seeing a request for more pizzas in this particular location or why are there more Mastercard transactions over here? And by unlocking that why, we help them improve their forecasting accuracy and also their operations management as well. So a good example is helping Uber get drivers in the right place at the right time ahead of time so that they can reduce pickup times as well.
Christina Noren: Let's start by, you know, it's an interesting business and it's definitely one that when we started "The Software Agents" as I was telling you earlier, it was one of the businesses that came to mind of oh my god, location-based demand must be all over the maps since the lockdown around the world. But so how did you get into this space? What's your story Campbell? And then let's talk about what's going on right now.
Campbell Brown: We had a travel business back in New Zealand, and we were doing carrying towards globally, and it was kind of my second startup of being in. And we were sitting there one day thinking why are we seeing a spike of bookings in Paris and why are we seeing a decrease in bookings in Melbourne? And we always retrospectively found that the catalyst behind this were these events whether it be a natural disaster, whether it be a sports game, whether it be a conference, whether it be a school holiday. But there was no single source to be able to use that we could use in our demand forecasting. And so I thought there's got to be an idea there. There has to be something there. And so the first thing we did is we put this little booking list which is on our current website and in that we said, "Book now. Availability low. Pricing is high. And that little tweak on our current booking website improved conversion by 35 percent. And I knew that moment we had something.
So literally the same time we were exiting Online Republic I was raising seed capital for PredictHQ; and that was kind of how it was born. During that process of raising seed capital there was a young guy by the name of Rob Kern and I convinced him to come along on the journey. So it was just him and I coming out of Online Republic, exiting that and then just moving straight into PredictHQ. And very early on we decided that the best place for someone like myself to be was the US because one of our first customers was Uber.
Christina Noren: So I know Uber has been pivoting a lot from car rides to eats these days. So talk with us more about when the great lockdown hit and this is one of the events that affects demand more than any other event that's happened I think in our lifetimes. How does that relate to how your platform works?
Campbell Brown: Yeah, that's part and parcel, right? So we're helping businesses to understand not only the incremental catalyst behind the demand but also the decremental, so what increases our demand and what decreases. And with COVID they've obviously had a massive impact on a lot of businesses, and so helping our customers adjust to this and helping them find new pockets of demand is kind of what our business is about. And I think the really important thing to understand is once that black swan event of COVID has occurred, there's all these micro events that are triggered from that. And so let me explain that is there's like a shelter-in-place order lift in some way like Texas us but it remains constant in like California. And so that lifting or easing in Texas creates a surge in demand. A really clear example that actually happened in New Zealand, where I'm originally from and where we started this business, is 6 percent of New Zealand went to McDonalds on that Monday after all the shelter-in-place was lifted. And so this suppressed demand that COVID generates as well is something again that our customers need to be aware of and what we provide to them in the form of demand intelligence as well.
Christina Noren: And I imagine a lot of this, you know, these are relatively low margin, high volume business, so not forecasting demand correctly means you have unused inventory that you're paying for or vice versa, right?
Campbell Brown: 100 percent. And the cool thing here is if you unlock that why, so why do we see the increase or decrease in demand is it also helps with marketing and promotions, it helps you with labor optimization, it helps you with pricing optimization, it helps you with supply chain optimization. So all these different facets are related to this as well.
Christina Noren: So Paul, I want to go down the path with Campbell of understanding how this thing works and then I wanna go back and make Campbell share a little bit more about who he is a person, how a person comes to do this. But so on the how, so what is PredictHQ's platform look like right now? What are your data sources? How's the data processed? What is the set of functionality that an end user in one of your customer companies accesses? Tell us about that.
Campbell Brown: We're a developer and data science first business meaning people access our intelligence threw an API. So within that API you have not just the raw data but feature engineering plus a lot of intelligence. So intelligences looks like we have rankings. So we will tell you how big or small an event is going to be or tell you how you can filter that event based on distance from a point. So there's all these factors that go into this API that people then are taking and integrating within their forecast modeling. That's the front part of it.
Then we have a couple other platforms. We have something called Bean. And what Bean does is it, in gist, anonymized transactional data and correlates that to demand. Meaning it correlates events to demand to tell you what events have actually impacted you in the past. So you remove all that, well, I think we got impacted by school holidays but I don't quite know. It goes this is what you were impacted by on this day in this particular occasion.
Christina Noren: Is that your customer's own data or are you mingling data from different customers for this?
Campbell Brown: It's a conversion of customers' anonymized demand data without demand intelligence data. So fusing those two things together. What we realized very early on is whilst time series modeling on the surface sounds relatively straightforward, it's actually very complex. And then when you add in the additional things like doing it with a dynamic data source like ours it becomes very complex because there's actually not many data sources like ours out on the market and so recognizing both those things is we knew we needed to provide something and that's where Bean was born and was launched about three and a half months ago.
Christina Noren: So is that been, B-E-E-N or what?
Campbell Brown: B-E-A-N. My blooming accent is going to throw off a few people on that one.
Christina Noren: Yeah, so you're guest number, what is it Paul, guest number second today and you're the second Kiwi of our guest so, a little bit of family bias going on there. Anyways so your data sources, what are these data sources that are predictive of demand?
Campbell Brown: Currently we've got 18 different categories and about 400 labels and these range to anything from sporting events, school holidays, academic events, political events, natural disasters, severe weather. So it's a really broad range of the most impactful things that can happen to a business in the course of their lifetime. We are adding new ones. So we actually do, on the downlow, we actually have a new category coming out in the next months which is actually around live TV events. So the big thing there is helping us to map let's say pizza demand to when there's a broadcast of a sporting event as well. And so I guess what we recognize particularly through having Bean as our platform, is we're starting to recognize even more and more demand causal factors. And what I mean by demand causal factors is simply we're starting to identify more catalysts that are increasing or decreasing demand for customers across the different industries we work in.
Christina Noren: Let's dig into the present moment, so how exactly, in the United States we have 50 different states and all kinds of localities and we're all kind of making up our story as we go along with lockdowns and easing and second lockdowns and so forth. How does that data manifest in your platform?
Campbell Brown: So we put a lot of effort into not just tracking what events are occurring but what events have been canceled, what events have been postponed. That's actually really, really useful particularly in the insurance field as well. But I think what we're trying to help people understand is that just because you're in California doesn't mean that the same things are happening in somewhere like London and probably 95 percent of our customers are globally orientated, meaning we're helping them to understand where there are these pockets of demand are occurring because different forms of events are going off.
And so whilst California at the moment is probably a lot of suppressed demand and the likes of somewhere like New Zealand or Australia there's definitely different things occurring or even in China we saw the other day a 40,000 person conference all of the sudden kicking off. They are one of our customer's hotels. And so we were able to not only identify them and say hey look this event was postponed in April but actually was rebooked and they were able to see that it was coming down the pipeline two to three months before it actually occurred. So having this visibility on what's scheduled, what's unscheduled, what's canceled, what's postponed, what's rebooked, mainly because these are such bit catalysts or drivers for demand, it's super critical particularly now, because business are so volatile at the moment. It's not just about making the most of when incremental demand comes back; but also, look, let's say in six months' time or years' time we're through the other side of this and your business is still very volatile, mitigating any losses that could be felt by something like a flood or a hurricane, understanding that is also mission critical for a lot of these businesses we deal with.
Christina Noren: So Paul, this is newer territory what we're learning about PredictHQ. It is for you and it is for me. What's crossing your mind as you're listening to Campbell?
Paul Boutin: I'm wondering if there are correlations where the causality is not obvious? I was involved in a story where The Washington Post had an expert say look at this satellite data. Trade between China and North Korea has all but stopped since last year; and it took other people to say, um, your second set of satellite photos were taken on Chinese New Year where nobody drives a truck. But are there more obscure things like I don't know, like when it rains we sell more coconut donuts instead of Nutella, we don't know why but there it is. Are there things like that where you don't know why but A affects a demand for B.
Campbell Brown: You've kind of hit on the essence of the vision of our business is we want to predict the catalysts on any form of demand. And so our mission is to eliminate the well we actually don't know what happened. Like a really good example is one of our retail customers was they were saying, "We cannot for the life of us figure out why we saw demand in this one particular store and this one particular time. Can you let us know?" And what we're able to quickly identify is that there was a climate day protest that was occurring on the same day as pay day, and so what that meant is there was a large proportion of people that were in that particular area who also had been paid, but that increased sales by about 26 percent. So it's just understanding not just the events that occur on an individual level but collectively is really powerful, and I think that's our mission is collecting as many different of these catalysts and not just looking at them at an individual level but looking at them at an aggregate level as well and using things like distance from that particular location and other factors to help determine that kind of level of correlation strength is kind of what we would deem it to be.
Paul Boutin: That's a fantastic example.
Christina Noren: Let's get into a little bit into the software factory but we are CDIC company so we care about them to be nitty-gritty. So what does your software delivery organization look like and I'm curious to know how you are delivering new functionality? Like how frequent is your release schedule? What supporting tooling and processes culture you're using? And then I'm curious as to how, if at all, that system of software delivery had to respond to any changes because of the pandemic.
Campbell Brown: Okay. It's a bit to unpack there, but let me kind of break it down to kind of a couple things is, first and foremost, our platform or platforms have been built on Python. So that's kind of at the heart of that, reason being it's super flexible for not only being for development but also for data science. The way in which we kind of deploy is we're actively deploying with two week sprints. And I think just to kind of touch on how that's changed in terms of COVID, I think that if anything our deployments have increased and I think a lot of that has come down to probably the lack of distraction in the world which has allowed for a lot more focus in terms of a couple of things that we look at. So if we look at our pipeline, we've broken our pipeline down into different micro-services so that we don't have a reliance or we don't have one breakpoint that could shut down the whole entire data quality pipeline. Then we've broken it down into micro-services so we can jump in, fix, and then come out without it impacting the delivery of our data to our customers. So that's just some of the things we do. Obviously we have specific software and platforms we use as well. But the data quality side just to touch on that is probably the unsexiest sexiest part of our business because it is just so bullen and tough. There's so much to it, it kind of blows my mind.
Christina Noren: So let's dig into that a little bit and it's funny because like 20 years ago I had a startup in the arts space where my data pipelines looked like shipping art dictionaries to Pakistan and the Philippines for manual data entry of facts and so forth. We were building algorithms to compare the data entry of the same data dictionary, our dictionary from in the Philippines and our dictionary sent to Pakistan. It was the same one and if the facts roughly matched up, we just let it go to the site and if it didn't it went to a human curation. So it was like messy qualitative data. And I can only imagine the data pipelines for the kind of data you're dealing with so take us into that a little bit.
Campbell Brown: Jeepers, the amount of machine learning models we've had to deploy just within the pipeline is immense. The reason – just to quickly go over, zoom in on that – is the data that we're ingesting, we have upwards of 40 percent error rate on that data. So incorrect GO codes, spam, duplicate. So we have to fix this on the fly and at scale, and the reason we have to do it is our data has to be forecast great. Now imagine if we go, hey, look, quality is not that important. Let's just all let it go through. And you've got someone like a Starbucks or an Uber using our data and they're trying to forecast accurately what are going to be some peaks and troughs in demand, if you have all those false positives in there, they can't get the accuracy they need and they'll make bad decisions. And so everything for our business is underpinned by quality. That kind of flows through our veins and our business. It's something I think a lot of people forget about. They want to talk about, you know, the next biggest and best thing but effectively for AI to work properly, it needs a quality data source and it needs to understand. And so we took that challenge up from the very early days of this business.
Christina Noren: Let's dig into that. So what are a few – and I'm not going to ask you to reveal too much special sauce but just what are a few of the data sources that you're receiving on a daily basis and what does that data pipeline look like and what's the intersection between the traditional notion of a data pipeline and a continuous integration continuous delivery pipeline into your software?
Campbell Brown: I think if we look at just one of the providers that we bring in, so we're probably bringing in somewhere in the vicinity of, you know, 200,000 to 300,000 unique events a day. But of those unique events – and let's just say it's a sporting provider. We actually just don't bring in one sporting provider. We bring in about 20 because we're looking for that one true instance of that event that's occurring. So first and foremost, we've got to clean this data that's coming in and so that's kind of the frontend of our pipeline. So that is removing duplicates and spam, et cetera. But then we got to do this matching job and we're going to do this comparison job of what is actually telling us the truth about and what helps us to verify the instance of this event. So that goes through a whole raft of other, I guess, parameters. And this is just constantly happening all the time, every single second. And so I think that's a real unique thing. Well, maybe not unique for our business but certainly something that requires a level of scale that we probably didn't first anticipate when we built this business. But now that we've kind of built it in there, it's helping us to not just bring in more event data but now we've built out this massive big knowledge graph. And so we're starting to understand a lot more about the world. And so our knowledge graph is built up of venues, celebrities, sports teams, all sorts of location information that helps us to provide intelligence back to that event or that demand causal effect that we then release to our customers.
Christina Noren: So to get more topical again, and you know, I'm not a sports ball person so you know I'm out of my territory here, but I do live next to L.A. Live which is, you know, probably one of the biggest sports complexes in the world. How does your system know the difference between the Lakers are playing in person and with a live audience versus right now they're trying to play, I think, again I'm not a sports ball person, they're trying to play without an audience which I imagine is a very different thing from the demand perspective for the many Starbucks around here.
Campbell Brown: Yeah, exactly. And that's a really good point. So we pull in not just attributes on the event itself but also is it attended, is it not attended, ticket sales. We also have people movement data that we bring in as well. So all of this is kind of helping us to understand how big or small this is going to be and whether that's in a COVID world or whether that's in a post-COVID world. And I think that's largely part and parcel of why we're building this live TV events category because what you got to remember is there's that locally impacted demand that happens when it is an attended event but all events aren't just about attended events. There's non-attended events. There's broadcast events. What we've actually found is that the broadcasting of these sporting events drives pizza demand, it drives grocery demand, drives beer demand, drives even Uber demand because people aren't wanting to drink and drive if they're going between places to watch the event. So it's kind of that I guess the beauty of being in this situation is it really forces you to focus and really forces you to expand and understand what are those additional demand causal factors that probably weren't around or probably weren't that much of a focus in 2019 but certainly are here and now in 2020. And also just to remember as, look, we're not going to be in this situation forever. Look, I hope it's way shorter than it's potentially going to be. But understanding that when these attended events come back, then how are they going to adapt? And so what we're really telling our customers at the moment is that there is a massive information deficit out there. You can either choose to be part of that and solve for now, and then when those additional factors turn back on again you can make the most of that also. So it's this really interesting moment in time for us as a business and also kind of having that overarching view of other businesses or customers that we're dealing with as well.
Christina Noren: So on all this, like attended versus unattended, I'm going to guess that a year ago that wasn't an important dimension for your knowledge graph. So has your team, how does your team arrive at these new dimensions that matter and how does that get shipped in terms of both your knowledge graph backend and the parameters of your API and whatever interface? You know, if tomorrow we suddenly have a new dimension, how does that happen?
Campbell Brown: It first starts with our data science team and so they are the ones that are building out the models that can be used. And then those models are then passed over to the development team for them to integrate within our API. So there's a bit of a follow-on process and I think one of the other big things we've learned is to make sure our staging environment is on point so we can break a lot of shit without impacting our main environment. And I think like if we just think about the non-attended/attended and how we're dealing with all these things is we spotted back in February that we saw a wave of things happening and so we decided to act really early. Some of that was internal changes. So how can we deal with cancellations/postponements at scale? So we built this massive model Amer around auto-linking and then we also thought, well, public facing, what are people going to be looking for? And we knew that they wanted to be updated as quickly as possible if an event was cancelled or if it was postponed or rebooked or whatever may happen. So that's why we released notifications last week. So this is a frontend SES platform where businesses without any experience or any code necessary can quickly set up configurations to be notified of impactful events that are happening anywhere around the world. And when we say impactful events, it could be a hurricane but it also could be a conference. So it really, really depends on what the customer is looking to do. And what we're doing now is we're looking to build out from that an API-facing notifications product as well.
Christina Noren: Let's dig into that a little bit. So if I'm a large retail customer of yours, I may subscribe to this notifications API and it may drive my communications to customers?
Campbell Brown: So first and foremost, if you're a large retailer and you're on an ops team, you can set up configurations around any of your stores anywhere in the world and you can understand by distance from a point, is there anything occurring that could impact the store. So that's first and foremost. Now, the next step on from that is in providing an API and that API could be used at the store level. So then not only can the operations team set up configs but they can then start releasing these notifications through to store managers so they can adapt. It could be staffing levels, stock levels, pricing levels, all sorts of different things. And I think what we know to be true is that in the next two to three years, timely and impactful information is going to be mission-critical because there is so much volatility in market. People have to make the most of pockets of demand and they have to protect themselves and mitigate their losses from any of these other decremental impacting events.
Christina Noren: So if I am a seaside Starbucks store manager and a tsunami is coming, I probably don't want to buy as –
Campbell Brown: I think that is the most extreme use case I've seen but, yeah. Look, that's an extreme example. We'll go with that.
Christina Noren: So Campbell, we really try to get to know our guests a little bit more on this show. So we know from what you said earlier that you were a founder of another company that we started to see these needs earlier. But let's go further back. Are you a software developer by vocation or where did Campbell come from?
Campbell Brown: I came from a little place called Hawke's Bay which I think you and I remember talking about a while ago. Look, I started as actually training to be a pilot. So I started when I was 12 years old and started flying planes out at Naper Aero Club. And then went solo about when I was 16 and then about 18 I just lost the love of flying for that particular time, probably because it cost so much so I went to university and did GIS or geographical information systems. So it's been a super weird line to get me into startups. But effectively, I'm a product person so I'm not a developer. I love to solve problems and work with people to help solve them. So I guess the way I'm kind of hardwired is every single problem I'm faced with I see an opportunity and so what I'm trying to do is bring people along the journey with me to help. If we can provide a solution to that particular problem, then we can create an amazing business. That's what I saw when we first started this business is I can't believe no one is solving this. It seems so freaking obvious that British Airways or Starbucks or Uber or Dominos, they should know this, right? They should know that there's a school holiday happening in this particular location, but they don't, and they don't know it to scale.
Christina Noren: Other people have tried to deal with this location data. Like Foursquare famously tried to create a data business. But it seems like, you know, tying the data to the logic and the use case around predicting demand and providing that kind of thing seems very unique with what you're doing.
Campbell Brown: The nuance there is I think with someone like Foursquare is they're trying to tell you, hey, there's going to be people in this location. What we're trying to do is like we're telling you why. And if you can unlock that why, that just unlocks so many use cases. And I think that's what's really exciting about what we're doing. And the other cool thing is that all of our businesses are using the same intelligence except they're augmenting it for their own use cases. And so it's just we have a different view on the world and I think, if anything, being in New Zealand probably helped us form this business because if you're in the ass-end of nowhere, you want to know what's going on around the world and you're not going to have that knowledge of what's happening in Paris or what's happening in London or what's happening in San Francisco. And so we had that foam of like we've got to solve this, like how can we not know what's going on? And I think that's the real – that was kind of that real burning fire in my belly that made me know that this is the thing to chase. And I'll tell you what, moving our family over here to the US was, you know, one of the toughest moves I've ever done in my life; but I feel like it was completely justified in terms of where we are today as a business and some of the customers that we have.
Christina Noren: A lot of Kiwi entrepreneurs sort of have relocated themselves and kept software development in New Zealand. Right now with the travel restrictions we all have, it must be much more of a burden?
Campbell Brown: Look, I think the cool thing is so our office is probably – in New Zealand is probably 80 to 90 percent full five days a week and we've given them the choice of working hybrid as well. So I think that's kind of unique. Look, it is tough for me not being able to get back to New Zealand to see friends and family; but I think I'm not alone in that. I think everyone else is in the same boat. But I think if anything with this, it's made probably Silicon Valley realize that businesses can operate with teams around the world because when I first came here, I don't think they really felt like that. And I remember a number of deals or a number of – I think it was when I was doing my Series A I had a partnership meeting, the deal was pretty much done, and one of the partners spoke up and said, "I fundamentally don't believe in having teams in different countries" and completely tanked the deal. But now, everyone is coming to me going, "Um, wow, you got a team in New Zealand. What's it's like?" I'm like, "It's great. They're awesome. Super talented. Amazing place to live. And they're pretty much COVID free so they're having a great old time.
Christina Noren: Our other Kiwi entrepreneurs who are aerospace entrepreneurs, they're duly headquartered in New Zealand and the Netherlands. They were on one of these with us, yeah, two days ago and it was almost obscene seeing people _____ together in a conference room. It was almost obscene. And you know, it's funny, in the three years since I met you, you know when I met you, I was working at a company that was very much thou must be in the office and I was commuting. I was commuting almost two hours each way every day in the Bay Area. And I've been at Cloudbees for these last couple of years plus and Cloudbees we have people in 17 countries, and we work totally distributed, and it's just natural, and it was almost no disruption. I mean a few of us liked to work in an office but it was kind of a personal choice. And there was a lot of skepticism about Cloudbees relative to other DevOps companies. I think DevOps is a little bit more distributed than most sectors, but there was that little bit of skepticism of how effective can you be. And the truth is now with the pandemic, we are far more effective than companies that are doing this for the first time.
Campbell Brown: Look, what I am going to be fascinated with is to see how quickly face-to-face meetings bounce back and it's interesting watching what's going on in New Zealand and what's going on everywhere else because it seems to me that you look at everyone – it's just going to be super interesting because I think if people see other people are doing face-to-face meetings, does that create this catalyst for like, well, we've got to be doing them? And that's just going to be this really interesting thing to watch as we move to the kind of the next normal, I guess, globally.
Christina Noren: Yeah, I think, you know, for us it's been, you know, we have ten years of history at this point at Cloudbees of doing this totally distributed. The company was founded by people on three continents and coming from an opensource development background which is very asynchronous and distributed. So I feel like we have lost something by not being able at critical ritual points to go back to being in the same room and forming those bonds and having the shared experience of working through a problem in real space; but as you alluded earlier in the conversation, that in some ways you've been more productive because there aren't as many outside distractions. So I think we've also learned that there is a lot of places where we were doing international travel or big events or social events as a company that were probably counterproductive and we can do less and be more productive.
Campbell Brown: I think the cool thing is there's going to be a big inflection point for a lot of businesses and I think there's a lot of positives that can come out of it which sounds – I don't mean it to sound dismissive of the situation we're in because it is a horrible situation. But I think every business leader out there needs to find for their business what does the next normal look like and how are they going to adapt to it and what do they need to build for now because I tell you what, waiting six months to a year, that's too late. What I worry about are these businesses that are just hunkering down but actually not moving forward; whereas, you've really got to look at what can we do now and what can we do to set ourselves up for success in the future and I think that's kind of the big narrative that we've been talking to a lot of our customers about and it seems to be working really, really well.
Christina Noren: My takeaway, Campbell, you know from this is I think with a lot of these conversations there are businesses like yours where the mission you were already on, and in your case the mission of understanding that events out in the world impact demand on a localized basis and businesses need that information, has become much more intensely important through this pandemic. So you're just doing more of what you were already doing, and the reason that you exist is more proven in the world, and I think that's a lot of the narrative. So that's my takeaway. Paul, what's your takeaway of what we've heard so far?
Paul Boutin: Well, I'm already familiar with a lot of this sort of predictive stuff but I didn't realize the level to which Uber makes it reliable for me to get a sandwich between Zoom calls by keeping the driver nearby. And I also want to point out because I know for listeners, a lot of people see these things as just capitalist wringing more money out of the system or something. Two things that people have pointed out that I'm going to throw in is, one is L.A. traffic, being able to predict your drive time between important meetings and events is always something we wish were even better. The second of a largescale global technologist said, "You Americans, whenever you go to the supermarket, do you ever worry they're going to be out of food or your favorite item?" It's a rare thing. And that's not because they have an endless supply of it. It's because of logistical planning that uses machine intelligence and we still have the failure of several unexpected items in the supply chain in March. Toilet paper is the one everyone knows. So the best thing I heard that was totally a surprise to me was the phrase next normal because whenever anyone says new normal, I think we don't know what normal is now but there is something that will be emerging. And as you said, people are dying right now but it's also an opportunity to say, wait, where have we been really wanting to make changes and we've felt like this wasn't the right time and now we have to?
Campbell Brown: Both of your comments, I don't know like feel me with joy because I'm glad that you've got it and you get it and the way you've articulated it back to me is on point. You know, like we, as I said before, we're in a situation where there's a massive information deficit. That can be a frightening thing, but it also can be a thing that you can use to understand in the future how are you actually going to adapt? If this is ever to happen again, heaven forbid, what are you going to do? How are you going to adapt? What if it's at an epidemic level and it's only happening in Australia and you've got resources, you've got stores there, how do you reallocate spend or how do you help out staff there? And I think there is obviously that overlaying, you know, people, the capitalist side of things as you mentioned before; but I think ultimately, if you can help navigate or be the guiding light for your business and help them get through the situation and keep these jobs, keep people in jobs, and keep you're making the most of these pockets of demand and also mitigating those losses, then, you know, that for me feels me with joy as well because I know that we're part of that recovery. And I think if we can be part of that recovery and get these businesses going again is a huge thing.
Christina Noren: I'll put a point on that which I was joking with someone whose mother is a medical historian and the other day which is 2020 is the new 1348.
Campbell Brown: My goodness.
Christina Noren: Fundamentally, like why do these kinds of plagues and pandemics have such a horrible impact on people? It's because they move everything around and people are starving in one place and food is going to waste in another place. So this kind of demand intelligence is a fundamentally humanistic thing. It's not maximizing capitalist output. It is making sure that human society is able to take care of itself.
Campbell Brown: Oh, 100 percent. Look, one of the biggest use cases is actually not putting produce in a store because they know the demand is not going to be there and so that food's not going to go off. And I think that for me is just, you know, look, we all want to create businesses that are going to save the world. I think I'd be naïve to say that we're that aligned, but I think understanding that we can have that impact and we can force that positive change is a good thing for us as well and hopefully a good thing for our customers and their customers as well.
Christina Noren: Well, that's amazing. Okay, well, I think we have a point on it. What do you think, Paul?
Paul Boutin: I think that's great. And I'll throw in that it's not the biggest thing in the world, but my niece who is working on a COVID cure in a lab, I'm sure that she's glad that she doesn't have to skip her Starbucks because they knew there would be a line that morning on the way to work.
Campbell Brown: Okay. Okay, that's good. And more power to her because that is awesome. I mean, look, we need everything we can at the moment in this country. So I'm glad there's someone like that working on those things.
Christina Noren: Thanks, Campbell.