 And now let's talk about the next generation data warehouses. The new normal is a world of social distancing and contact tracing. Spatial data analysis is becoming more critical than ever for governments, for businesses and even for our own very survival. We are lucky to have with us our next speaker who will take us through the examples of how big location data is helping respond this crisis to effective location intelligence. He is Javier de la Torre, founder and chief strategy officer at Carto. Welcome Javier. Thank you, Helen. How are you? Lovely to see you. Lovely to see you. Javier, you are the first one. So the level is already very high. I know when Carto was Carto de V, they were the most successful ones at the beginning of many years ago, now that you are super well known. So I'm sure you have a lot of fans in the audience and I'm going to remind them that they can ask you questions at the end of your talk, if you're going to talk around 35 minutes, more or less. The last five minutes we can dedicate to questions. So make sure you send your questions with enough time because Javier lately, these past two days, the audience send the questions a bit too late when you're gone. So before Javier leaves, send your questions either in English or in Spanish. Spanish is as good as it's English and take advantage that he's with us. So Javier, when you are ready, we're looking forward to listening to you, all yours. Thank you very much, Helen. Well, it is, first of all, it is my pleasure to be here with you today and I didn't realize that we were on the attic floor. It's actually, I mean, the attic at home. So I think it's quite convenient and as Helen would say, maybe you know, just as a little interaction. And I found it carto with many others. And I'm very happy to be here today talking about one part of analytics that some of you might not know. And I think, you know, it might be interesting on your day to day, but also understanding how it's being used as of today on responding to this crisis of COVID-19. So in fact, let me start with that. I know this sounds bold, but probably there's never been a more important time for geography. And I mentioned, you know, just because of two particular things that are happening, the most obvious, you know, is COVID-19, where, as you will see, location is a fundamental part for, you know, like managing this crisis, not only understanding how and where it is happening, but also kind of like, you know, like coordinating our response to it. And you will see a few examples of that in my talk, but also we cannot forget there's a number of other crises that are coming right now, you know, that still are going things like climate saints or biodiversity that are there. And, you know, they haven't gone away and we're understanding geographies also critical to things, you know, like where, you know, where is the sea level is going to rise with cities are more vulnerable to climate saints is going to be very, very critical. So obviously, I mean, like we, I love geography, that's my career, but I think you're like, many people don't realize about the importance of it. And if we put it on the context of this conference, geography today is a study, it's actually kind of like analyze using what we call spatial data science. So think of spatial data science as a subset, right? Of data science that is specialized on understanding geography. So, but before actually kind of like the best way that I'd like to describe spatial data science is also by giving a few examples or, you know, like, or like I'd like to say, explaining the difference between understanding, knowing where and knowing why, let me actually call I dig a little bit on that. So if you have in your, let's say on your company, you're selling a particular product and you have the locations of all your customers, you likely can, you know, use a BI tool such as Tableau and make a map out of it. And that will help you to understand where things are happening, right? You want to see where your customers are. But if you want to understand why your customers are there, you're going to need to create a model, you know, to model the characteristics of the places where your customers are so that you can understand why they're more likely to buy your products from one place versus another. So essentially you have to create a model to understand the geography of your business. And for that, you're going to need a location and taking this platform, you're going to do a spatial analysis. And obviously, you know, you can use a platform like Cartoon, right? So that's the difference, right? So this very straight way there, you know, like one thing is being able to see where things happens, but that's not enough to understand why things are happening there. And how does this work? Obviously, I mean, like, believe it or not, our, our field was studied through, through the study of, of, of, of, of crisis in terms of, you know, like immunology and so on. And, you know, one great example, of course, you know, for explaining how our type of analysis works is thinking on the spread of a virus. It's very obvious that the closer that you are to the virus, the more likelihood that you can get infected, right? So there's many things like that that can be a model and a study in the same way. So think of, you know, like customer behavior. If all your neighbors is stolen a lot and alarm in their homes, the likelihood that you will install an alarm at your place increases dramatically, right? So this is what we call essentially that, you know, in the, in the first law of geography, that everything is related to everything, but things that are close by tend to be more, more related, right? And that defines an entire field of analysis that we call spatial data science. Um, these techniques, uh, just to put it in context, you can think of them as another type of a spatial modeling or machine learning, right? So you have, you know, like for vision, you will have like, uh, you'd like driving, doing like self-driving cars. You have like vision type of, um, machine learning. You're doing like a speed to have like natural language processing. Well, if you're working with location data, spatial modeling and spatial data sciences, they feel that color does the spatial modeling. It is actually what you, what you will be using. There's a few pieces, uh, technical here. If your data scientist asks you, why do we need to understand, you know, why do we have to take in consideration location? So the first thing, you know, just a two of them, and there's many more, but you know, two that are very obvious. Just think that if you model, you know, where your customers are, um, customers that are in year one to each other, like we're saying, are more similar than that are those that are farther away. So it's not only the characteristics, it's also if they're clustered, if they're close one to another, because we influence one to another. We call that a spatially auto correlation, right? The other thing that is important to understand is that, for example, customers behave differently based on their location, the regional variation. So we say like, a model might work in a place, but it might work differently in a different place. We call this, those models are non-stationary. And if you don't model, and if you don't study your geography, uh, taking in consideration location, you will be missing these things. So, um, there's many organizations that haven't gone yet the way to, to call like, uh, to move to, towards the spatial data science, but location is a fundamental, uh, it's a fundamental dimension that most organizations are going to have to work with. At the end of the day, everything happens somewhere. So it's suitable, you know, for studying it from that perspective, right? Okay. So, uh, so with that being said, um, obviously 2020 has given us an incredible crash course on location analytics. I think we all have never seen more maps and analysis about areas and zones in any other year, more than in 2020, right? So we are all familiar with maps about, you know, like, uh, which areas are more affected, you know, like, which is, you know, the restrictions that you have in your areas. So there's been an incredible amount of data scientists working on a spatial problems because of this and organizations are also working with location data. So in fact, only a cartoon, we, we made the program to offer cartoon for free to any, um, COVID-19 type of, uh, project. And we got 165 grants in a matter of a couple of months. So it's been really, really kind of like an explosion of analysis when it comes to, um, uh, COVID-19. I just want to show you a few examples of the things that you can do with a spatial analysis on COVID-19. So think about, like, for example, calculating the risk of demographics and health factors. So we know that COVID-19 doesn't affect everybody in the same way. So you know, like your age, your color, your health, and many things define your vulnerability to the virus. So if you want to see then how it's actually, how different, how different communities are vulnerable to COVID, you're going to actually need to start doing this type of modeling. Right. So that's a very classical one. And that's used then later to call a provision where the resources government is going to put, you know, which areas are more likelihood or in a bigger danger because of the virus. And no one that we've seen, unfortunately, also a lot on the news because it tends to be a very big, you know, like political debate is this association between human mobility changes with, you know, with with lockdowns and the socioeconomic patterns. So understanding that's locked down actually, you know, like changes, you know, like effects, you know, the infection rates and what is actually the patterns into what is actually the impact on the things, you know, like the economy around it. So that actually, this year, we've had the opportunity to do a lot of these analysis at the scale by, by, you know, by looking at different strategies in different countries or different states in United States, where we could see, you know, how was the different policies was actually getting, you know, like an impact on on on the crisis on both the economy and on the health parameters. Even more, kind of like, I would say, you know, I'm not mundane because very important, but you're like, particularly in Spain, managing things, you know, in this new normality, how do you how many people feed in the beach? Right? That sounds very obvious, but you know, like you do actually need to do quite a bit of a spatial analysis to color perform this type of analysis and, you know, and to to open them, right? So this is just a few examples of that. The last one, I'm going to say like, saying like it just couldn't be more, more personal in a way, I'm in an attic, we're in the attic loads at home, I'm actually and I'm in a lockdown area here in Spain. So in the case of Spain, depending on on the region that you are, there's different kind of like lockdown strategies. And the one that I am is defined by a small region. So you can see it in here is in the town called Mahalla on the north of Madrid. And the way that you actually define those areas has a very a very big impact. Obviously has a very big impact on the people that live there. You know, you're like restricting, you know, what they can do and what they cannot do. But also, you know, like on the effectiveness of the measurements, right? Is this part of the city, you know, like people in this part of the of the city move more towards this other part of the city, do they go somewhere else? So how do you split the territory? How do you split the regions to maximize in a way, you know, or to limit the connections between between citizens while minimizing the number of people that you have to lockdown is a pretty heavy space of problem. So those are the type of things that, you know, like a spatial data science enables you, right? Now with COVID-19, and not only COVID-19, you know, like climate change and over the last five to 10 years, it's fair to say that we've seen an incredible new set of requirements for a space analysis. There is, and I would like to call like a walk through a few of them, but it's only because you know, like some advances on the cloud, that we actually right now able to do many of the studies that I was showing you. And just so to set a little bit of the requirements, but what are we seeing that has changed, right? What is right now that we're doing when we were with this type of data? So the first thing, obviously with COVID-19 that we've heard is the immediacy, right? The idea of like having data always available, but you know, like be able to run analysis and get responses, results very, very quickly. It's very important. It's very important because, you know, we cannot wait, you know, like for weeks, you know, to get the results in order to take consideration. Like you might in the past, you know, like if you were a company looking at your where you're going to open a restaurant, you might, you know, take weeks to get that decisions. But in the case of you know, like COVID-19, I mean, you're going to need to act much faster than that. The other thing that we see is the freshness of the data. So now I mean, data is the world is changing very fast, as we know, I mean, with crisis like COVID-19, now we see, you know, that suddenly we have a we need to look at the population and, you know, and demographics in a very immediate way, in a very fresh way. And things, you know, where, you know, like how you have characterized population in a place has changed very, very quickly. So we need the data to be more fresh than ever. In fact, we used to call it work in overall thinking on, you know, like data being updated once a year was the most common call like timeframe for for the upgrades updates on the on the data and allocation data. And now we're looking at data sets that gets updated, you know, like weekly, monthly, even weekly and sometimes daily or even hourly. So the data is becoming much at a much faster refresh rate than before. It also has to be multi source because of the need to color get things done very fast. And we need to actually get them, you know, like with very fresh data, no single source of data, it's becoming enough to explain most of the things that we need to model. So for example, census data, gets done in most countries every 10 years, it gets updated every year with samples. Well, that's not going to make it for some of the analysis. So so therefore, you need to start looking at, you know, like mobile phones, data, strict, great car transactions, data, you're going to need to have many different alternative sources of data that you need to compare and you need to use to complete the picture of your analysis. Also continues. Now we see that most analysis, you know, become effective when you are like in a kind of like in continuous integration mode, where you don't come and do it once you come, prepare the analysis and let it go for time. Right. So we see right now in COVID-19, it's very clear, you know, with the seven days color windows for some of the KPI, some of the metrics that they use. But also, you know, it's the tendency the importance is not in seeing a one time snapshot, but to see how it is evolving. Right. So you need to color have your models always be running continuously. And finally, you need to automate it for all the previous reasons. I mean, like if the, if your analysis cannot be done in an automated way, you're likely not going to meet, you know, the immediacy, the freshness and the continuous requirements. Right. So that's actually pretty heavy new set of requirements that we didn't have before. But luckily, one of the things that, you know, that have happened, not only to space a little science, but overall to analysis, it's been the cloud, as we have seen in this conference a lot, and you are all aware. So in our world, we say that, you know, like cloud native space analysis, and I'd like to call like decalibrid on to what that means cloud native space analysis. It's actually the solution to a lot of these new requirements. If it wouldn't be for the cloud, it will always be impossible, will be cost prohibitively will only be accessible to very few organizations to the type of analysis that we've done that we are doing these days. So again, like with many other things, you know, like we said, like, what will happen? You know, like COVID-19, if we didn't have Netflix? Well, we don't know what it will have been your COVID-19 if we didn't have cloud for doing some of this spatial analysis. So let me tell you a little bit about what this space analysis actually cloud native space analysis means. So it's pretty much based on what we like to call the next generation data warehouses. And if you're not familiar with this, I'm sure you know most of you are data warehouses, such as Google BigQuery, Snowflake, Athena, Delta Lake, Patsy Drill from an open source perspective. So those are kind of like a set of data warehouses, you will see, they have some specific kind of design principles that works very well for the type of analysis that we're doing. Some of them have the common particular BigQuery Snowflake and Athena, sorry BigQuery, Athena and Azure Synaptics, you will see that they're in the same category because they cost actually the same cost for the same, they have the same price on their different clouds. So it's $5 for processed terabyte. It's becoming kind of like an standard in that. And all of these data warehouses are in a way adapting spatial capabilities to it. The first one was really BigQuery in July 2018, but since then Snowflake in February 2020 and Amazon Redshift in November 2019 have already added spatial capabilities to the products like a spatial data types and functions. So it's pretty great for us. And why is this very interesting? What is the special about these type of products? So first one that you, I'm sure many of you are very aware, is this computing storage separation because you pay separately the storage, which is normally on S3, Google Cloud storage, in cheap storage, cloud storage, separately from when you actually do the analysis, it becomes very effective for, essentially, collecting a ton of data, but only pay for the analysis that you're going to make with it. In fact, if we wouldn't be for this computing storage separation, the bills of running some of these analysis will be profitable. It will be so expensive that it wouldn't be possible. So that's a key component. The second is the scalability. As this data warehouse liberates the cloud computing, they can necessarily open up many instances and run on the processing in Paolo. And that enables us to do things like, in this case, in this map, what you're looking at is around 14 billion points in a map. Now, in order to visualize something like 14 billion points in a map or do an analysis on that, you're going to need to partition and you're going to need to process the data in a very, very effective way. So actually in Carta, we created the full technology for that. So if you're familiar with formats like Parquet or Apache Arrow, there's another way of structuring the data that is more optimized towards location. And it's kind of like in what we call in tiles in a pyramid. So that makes it very effective not only for visualizing data, but also for processing and for doing analysis on the data. And that type of format, the type of indexes, run incredibly well on this type of data warehouses. And last, the third one is what we call data multi-tenancy. The fact that most of these sources because they live on the cloud, like the separation between the users since it's a logical separation. So all the data is under the same place in the cloud. So I like to say, we all live in a single database. So that makes effective things like this, where you have a SQL where you are essentially doing the joins between data sets provided by many different providers, by many different sources, that they create on their own, but that you can join life inside your queries. Which is an amazing thing. No need for ETL, no need for replicating the data, all that extra cost. That's incredible. In fact, actually Cartel does benefit a lot from that. We do, as part of our product, have a product called Data Observatory that essentially provides a very comprehensive list of data sets to use for spatial analysis. Things like demographics, human mobility, all sort of like the most common data that you need to do a spatial analysis. We already provide it. We have more than 10 categories in 33 data sources. All of these right now is available in BigQuery and coming to Snowflake and Athena and Azure Synaptics in the coming months. That means that we have all the data already prepared so that you don't have to import it into your system. It's already, like I like to say, a join ahead of you. So it's a game changer on the way that we distribute data. At the end of the day, what we actually do at Cartel with all of this is we like to say accelerate the spatial analytics by leveraging these next generation data warehouse with its own characteristics and adding a spatial layer on top of it. Our goal is that more organizations can discover the power of a spatial analytics for their own business. With that, I mean, it's very clear. I mean, spatial analytics is going cloud native. That's just really not going anywhere, anywhere else. So if you're interested around it or if you've got any questions, I'd be happy to answer all of that. But for that, I would just like to leave it there. If you are not looking at the location component, there's likely something that you are missing in there. I'll be happy to actually help you look at it. So thank you very much. If there's any questions, I don't know, I'm just now looking at how... Okay, I guess I have to... I wasn't ready, Javier. You have so much more time. You were supposed to finish, like, in 15 minutes. So I don't know if the questions are coming in yet or not. Actually, so Javier, I don't know if you could go more in depth into some of the cases. For example, of the success... Some of the latest cases that you were... You actually published in your blog, in Carto, in the website, you have some examples of clients, of the use of these with clients, actual case studies, no case studies, but cases, examples, for example. That's what I mean, with your clients. Yeah, yeah, this is great. I mean, I tend to speak very fast, so I'm sure that's what happens here. No problem, that is fantastic. Usually, we have the other problem. So we have another 10 minutes, if you want, Javier, if you want. And, you know, for the... No, I'm happy. Actually, you know, like, one of the things I love to actually serve with you is some of these visualizations I was talking about. Like, if you're interested on visualizing or handling hundreds of millions or billions of rows on the map, if you have that type of data, we'd love to actually work with you guys. So I'm just actually going to give you a demo. I think that's probably the best thing here. And tell us a bit more about... Javier, tell us a bit more about... I think it was very interesting what you said about the... You only pay for the data that you analyze. Tell us about costs in general, or how is it easy for any company? Because your clients are obviously... We were talking before, MasterCard, super-top big companies, but is this supposed to be used by any small company, any retailer? You said that you said something like everybody's going to be needing... I've been taking notes. Everybody's going to be having to use this, because location is everywhere. We are always... Everything is somewhere. So in which sense this is available, you said the computing storage separation allows people the access to many clients, but to how many? I mean, an old retailer from the street, you're talking about the most huge companies. Well, maybe that's... Yeah, let's dig in that. So if you haven't checked out products like BigQuery or Athena or Synapse Analytics, you'll see this is an example on BigQuery. The difference of the way that it works is that you used to pay for the amount of data that you store on your database, on your data warehouse. Now, you do pay for that, you pay for the storage, but at the same time, you do pay separately for when you actually do queries to it. And that means that in the case of BigQuery, we're just going to show you an example query that I'm doing here. I might have a data set... I'm just going to show you an example data set in here. So for example, a public data... So in this database, I'm going to find... We're okay with timing. I'm going to show you here, for example, here, there's many tables, there's many data sources already available. I wanted to show you one that is called OpenStreetMap. OpenStreetMap is a database of Wikipedia of maps. And you have here, for example, the entire... You can find here almost any, let's say, bar retailer in the world, just by looking at this data. And if I look at this database, if I look at the details and so on, this is, for example, a table that it has around 870 million records. It's in size, it's around 300 gigabytes. If you needed to have this table available for query in a regular database, you would need to have a very, very large machine available for computing. You want to run this in a Postgres and so on. So that makes it really expensive. You will have to have a very large server just for being able to query this table. Now, the difference on this type of data warehouse, the way that they work, is that when I... What I'm going to do here, I'm going to query this table. We're supposed to be seeing your screen, Javier, because we are not seeing... Are you sharing? Sorry, I'm not seeing it, but yes. I'm not seeing apparently everybody else does. And actually, I have a lot of questions already for you, Javier. So when you're ready, go ahead. Let me just finish this, right? So when I run this, like now, this is actually not a very good query, essentially now, in this case, BigQuery is only going to charge me for the amounts of, in this case, gigabytes or terabytes of data that it needed to process the query. So it means that, you know, I always had this table, I almost paid nothing for it, but when I did the query, I paid a little for it, right? And that changes dramatically, like I said, like the way that, you know, like you... Essentially, now everybody can have a data lake without actually paying for it. This is almost free, the storage, and you only pay for when you actually use it. So you only really pay at the time when you're getting the value out of it. And that's been a key, a big key in here. And that is, to your point, means that now organizations with much smaller capacity have now access to do analysis that before just wasn't possible. I mean, processing like this 140 billion, 14 billion dataset will require an amount of infrastructure for any organization that was only available for a few of them. Now, that's actually, you know, like accessible to anybody. And in terms of that, and you can see, you know, from organizations that, you know, we work with that, you know, work at, you know, like, they might only have, like, five locations, you know, like, it's very small organizations to, like I said, like to talk to MasterCard, ModoPhone, very, very large organizations. So there's a huge democratization on the capabilities. Excellent. This is very good news for Carto and for the world in general. Javier, they ask you, as Funcion is asking you, she says, Hi, what are the opportunities for special data in digital market marketing use cases? Well, that definitely, you know, like it's, I mean, like in the case of marketing, there's been obviously a very big interest around all the ATEC, I'm just going to stop staring at the screen here so I can see you. So one area that, you know, like we've seen obviously with the targeting, right? So understanding audiences based on location is being pretty critical, right? So it's not the same, I mean, not only the country, but, you know, understanding the population, so the possibilities to target individually, you know, like based on, you know, where you are and you're on the characteristics is being one of the most successful usage of location intelligence traditionally. Now that's actually changing quite a lot these days, you know, because of privacy and for good reasons. So there's not, you know, so much of, you know, like targeting now with locations since, you know, like most of the data these days is very well locked down in that sense. So the love opportunities are moving towards, you know, like connecting the physical space with the online experience. Like, for example, can you actually connect, you know, like your visitors that come into your store in your shopping mall, you know, with what is actually their behavior online? Those are actually the things that, you know, like location also is pretty important, understanding your customers and, you know, like providing their experiences. And those are some of the frontiers, I would say, that we've seen on the marketing side. Excellent. So Azum Piong, take note of those tips. Lourdes is asking you what about the future on this or of this? Do you have already real references and achievements? Well, I think we do have a few. In terms of me, I mean, like, if you talk about, like, the industry in itself, I mean, like, it's the location industry and it's been existing for 40 years. Now, I mean, on our journey, I mean, like our company started a few years back. We have now more than a thousand customers, you know, from all over the world. I think last time with countries was like 45 different countries around the world. So the need for understanding location is fundamental. Now, how fast are organizations realizing of the importance of this domain? This is still, you know, like various from industry to industry. But, you know, from obviously the public sector and governments has already been there, I mean, like for a long time. Now we've seen a lot on, you know, like on insurance, on retail, on telcos. If the organization is not taking consideration location, it will eventually. Yeah. And I've read that apparently there's obviously an exponential growth forecast for 2026, 2027. I don't know which, why those dates particularly. And I guess this year, 2020 has as well, if you want advance or press the throttle into that growth which was already there, right? Yeah. I mean, on one part for sure. I mean, like there's been an incredible acceleration of digital transformation, as we all know. And as many organizations, you know, have done that transformation, location intelligence is one part of that path. So you definitely having that. I mean, like the projections from a business perspective, I think is on the other like $70 billion industry, you know, like in the next four years. So it's a very large industry. And it, I mean, it varies a lot, you know, from, from sector to sector. And one way that I like to describe it is that most studies say that, you know, around 80% of data has a location component. And if you think about it that way, you know, like that's an incredible amount of data that yet have not been used from a spatial perspective. So the growth in terms of, you know, like how much, how many organizations will be enabled, it's very, very substantial. I think it's a double digit growth, you know, year by year. And it's been, you know, for a number of years now. So there's a bright future ahead then for, for us and for Carto as well. He was telling me before Javier, the amount of work they have. This is fantastic news for a Spanish company that is, is based in a way in the States as well. Javier, this is usually New York. Now he's enjoying majada onda, which is fantastic. The attic. We have no time for more Javier, but as you say in your website, your slogan or your motto is moving from seeing where to understanding why. So unlock the power of spatial analysis. That was fantastic having Javier to open this third day at big things conference. Thank you so much. It was lovely having you Javier. I hope to see you around and to see you soon in big things next year. Okay. Thank you very much. I love you. Big, big kiss to you.