 Thank you for my interaction Hello everyone. Thank you for for coming So we're gonna want to talk about business and we're gonna talk about location intelligence First I wanted to ask the audience. Can you raise your hands if your background is more business than technical? Okay, given is the business track, but we wanted to check a little bit of the audience Cool, so let's talk a little bit about ourselves. My name is Miguel Angel. I'm the city of Geoblink. My background is technical I'm a software engineer and most of my career. I've been a backend engineer working with different technologies and then moved on to the technical management track Yeah, so Hello everyone, I'm Rafa Polido. I'm leading product at Geoblink My background is also technical. So I did computer science a long time ago, but I got into product management early on So what I do is I What I know is about building products or the right products for the market that also Makes business value for the company So what do we do at Geoblink? Geoblink is a location intelligence startup our product helps retailers Real estate companies and also FMCG companies to make any decision that has to do with location Much better and much faster So how do we do that? Well, we start from from the problem So we try to understand what are the problems within those industries and then we look at location intelligence so data and also technology to provide a specific business answers that solve those problems and All of these is presented into our product, which is a cloud-based SAS product and that is easy to use and it's very powerful So any company can take advantage of location intelligence Thank you. So in this talk we wanted to provide a little bit of background about location intelligence This is a term that is still kind of recent some people have been talking about it for a few years But it's still definitely not mainstream So in this talk we're gonna talk a little bit about the history of location intelligence where it actually means How companies can use it and then the second half of the of the talk We're gonna go more into practice details about challenges that we have found building this This platform location intelligence platform in the tech side and on the product side So in order to Talk about where location intelligence comes from about the history and about what it actually means We're gonna cover three concepts that came before location intelligence and actually shaped it So these three concepts are They are mainstream and all of you are gonna know about it And the first one is business intelligence. Okay, so I wanted to talk a little bit about the origins of business Intelligence because it's relevant to this talk and It all comes from the 19th century This guy called Richard Miller growed something called encyclopedia of business anecdotes And in there it talks about this bunker that always had an edge Against the competitors because he always managed to get information faster and he talks about he having a train of business intelligence and he always knew first Compared to anyone else when the wars ended or what the king wanted and he used that in his own benefit Later in the 20th century the fathers of business intelligence these two fine gentlemen Hans-Peter working at IBM and a Howard Dresner in the 80s They provided the first definition of business intelligence that is more or less how we know today And it is concepts and methods to improve business decision-making by using fact-based Support systems and I want to stress three parts of this sentence the business decision-making Fact-based what would be equivalent to actually using data and Objective data and systems or computers. So In these days business intelligence is very common and pretty much every company out there uses some way of business intelligence Big corporations have huge platforms with tons of teams. It's even a small companies They have some way of business intelligence and is it's all about using the data to understand The different variables that impact your your business and what to do about that. So With our business intelligence tools you can retrieve and share data and we saw the data you can identify business opportunities gain insights You can generate reports about key metrics to actually understand the the things that are important for your business It helps improve the decision-making process based in that on Objective data and then you can generate analysis that are real-time to know right here right now Which is actually important and what should I worry about my business? So this is funny and good, but these VI platforms They don't actually take into account location location data or Special data is very different from the other data that your business generates It's not just numbers. It's actually tied to coordinates and also in order to visualize it You cannot use standard VI tools You need to use maps and you need to use special visualization. So Let's move on to the next topic. I wanted to cover and this is GIS or geographic information systems. These are systems that are basically related to maps So I'm also going to talk a little bit about the story because it's actually interesting the first person that Is considered to have used GIS in some way is this guy called Jonas know and not the Jonas know you are thinking about but another Jonas know and At the time 19th century there was a big Spread of disease of cholera in London in the Soho neighborhood in London And everyone thought that it was being transmitted by the air and they didn't know how to contain it So what this guy did is he used a map and he put the different points or where people dying were living and identified a pattern It turns out that people dying were living around a specific well He figured out that it was actually being transmitted by water and by cleaning the well they they cleared the disease and Moving on in the 60s. There is this Canadian called Royer Thomas Lund he had to do an inventory of land in Canada and You know if you have to do inventory of land in a country like Canada that is pretty large You don't want to do that manually. So He actually came up with this whole GIS Concepts he was the first person that actually figured out how to use computers to input data linked to coordinates what we call geolocated data Also stored this position data and then query and put it in a map using graphics something very advanced for the time During the last few decades or years. There's been a huge yes explosion Now we have cheaper and faster computers Computation in with GIS are pretty expensive because you have to use polygons. You have to do intersections You have to do a lot of mathematics. So it's computationally experienced. So it helps and now computers are cheaper and faster You also have new technologies both hardware and software that been developed to manipulate the special data And most importantly now there's a lot of data is very rich and it's everywhere. We have data coming from satellites We have Everyone here and all around the world using GPS. You have the internet with smartphones and internet of things And also you have this concept of public data smart cities where governments and institutions are making all sort of Data related to the resources of a city public so that private companies individuals or other public organizations Can use that data to actually Make cities more efficient and we're talking about the word resources in Transportation education all sort of things and all this data is just there for people to use it and so This all has brought amazing products that Everyone is using this day any person can buy a smartphone and look something about Google Maps or order a taxi But is it's kind of crazy to think that only a decade ago These systems could only be Implemented and used by specialists and now anyone can can actually do it but when you actually understand the concepts of routing and maps and Zooming in those are things that are embedded in our day-to-day slice But only a decade ago. It was actually pretty complex and it was in common for people to understand these concepts The last concept I wanted to talk about is artificial intelligence And I'm not going to cover the history of AI in depth because Everyone in the room is probably familiar where it comes from But I wanted to cover two specific concepts of AI the first one is machine learning although machine learning per se is not actually Artificial intelligence, but is one of the building blocks to build some of the key functionalities And I'm talking about forecasting and and catheter organization The other concept I wanted to talk about is deep learning or neural networks and that is something that you need to to be able to use in order to implement something called computer vision and This is actually very relevant to this talk computer vision is the technology used in self-driving cars Basically to understand what's in an image computer vision is just about a computer being able to identify What is represented in an image so when when a Tesla is saying, okay? This is the road and that is a person That is a traffic light. He's using deep learning inside the car in order to identify all these different objects It's also being used. I've seen another talk today about a computer vision identifying satellite images And that way you can see the buildings in a city Not only that also what kind of building if it is a warehouse if it's offices and what this means is that now we can automatically understand the data in images and Put a coordinates in that data because you know who's taking the position of who's taking the picture So you will know the position of what appears in the picture. So now we can massively Gather information that is geolocated With that introduction Let's talk about location intelligence So so location intelligence analytics is all about location And it's kind of closing the gap between GIS systems that are all about cartography and location and maps then business intelligence which is all about grouping the data and getting the insights and There's also the artificial intelligence. So you can actually make sense of the data and get very powerful insights So reality is the location intelligence is so powerful providing business solutions to specific problems when there is a location involved So when I talk about better solutions, I talk about Data data enrichment. So one thing that is quite powerful is when you take your business data So imagine you're a retailer and you have your network and you know your customers because you have a loyalty program and you know the performance and and all the details about the revenue of each one of the stores and then you cross this With information about the location all these different data points Advanced analytics competitors attractors Consumer spending all the population flows when you mix all those things you have many different data points that are very powerful But how do you go from data to actual insights? Well the the enrichment and the data is just the foundation. It's just the the base now You apply artificial intelligence techniques to actually get actual insights from the data And then you present these in a very map centric And so you can do the analysis very much focused on location as a business So I just said that we went from data to insights. How do you go from insights to business value? Well location intelligence is working in three different directions. So let's let's deep dive Descriptive analytics, so it's answering the question. What has happened? So this is very powerful when summarizing a situation So if you have data around a location the people visit an area or the disposable income for the specific area You can and you cross that with your business information with your stores. You can actually It can actually help very much to identify problems and potential solutions So if your stores are split across the country and your target Demographic is not close to some of your stores. Maybe there is a problem there and you need to optimize your network So I'm gonna give you a couple of examples One is just by visualizing the distribution of your customers Imagine that you are an e-commerce and you know where to deliver your products because you have many clients And you can look at that and even enrich those data sets with for example information about the disposable income Another example is the spending data If you're able to identify areas with high spending in the category that is very relevant to you That's telling you a lot of information. But again, this is just describing the situation It's not telling you anything else and what's there We take us to the to the next area, which is predictive analytics Here is slightly different because it's answering the question. What could happen in the future? So it's actually anticipating future scenarios for your business So, you know, what's what's happened in the past and now what could happen in the future? So obviously there is much higher business value because it's not about telling what happened It's also doing some simulations in the future and seeing how those things will go for your business So a couple of examples, I think a very popular one is the sales forecast So once I know everything about your business and we know everything about the location And I have the history of your sales performance Then we can anticipate in the future if you open a store, how much money you're gonna make And that's actually quite accurate. So it's it's kind of telling you future scenarios so you can make better decisions and And this is a closing simulation is kind of the other way around Imagine that you went through an acquisition or a merge and then, you know, you want to reduce the retail footprint that you have So, you know, you don't know exactly how so you can do prediction and see if I close this store How many customers I'm gonna keep how many of them are gonna move to the to the online channel, for example And finally another area that location intelligence is helping quite a lot businesses is the prescriptive analytics So that's what should we do as a business? It's very interesting because that's providing advice on on a specific outcome So if I know your your end goals as a business what you want to do and I know all the information I can proactively suggest do that change that open their clothes there It's just more proactively telling what's the solution for your business Obviously, this is the highest business value Because it's like you don't even need to think almost I'll tell you a couple of use cases so redesign of the network if I know the business goals I would you're trying to achieve as a business using a location intelligence platform We can redesign your strategy to achieve those goals or For example on the marketing side of things if the location intelligence platform knows that you're targeting a very concrete Target for for a new product for example Then it's going to be able to identify those and look at the flows and how what are the changes on their behaviors to actually? Target marketing campaigns much better But let me give you a couple of more specific examples. What do you see in this image? This looks like high street, right like a city center. There's many people there, but do we know how many people there? Difficult right do we know their profiles? Are they visiting are they passing by maybe they live there? So for businesses, they have these questions, you know, there are many questions from geo marketing to offline retail To even like residential real estate. There's so many questions that are very difficult to answer So how do they do this? Well, they actually it might be some Funny, but they actually send people to count and to look at people and figure out the profile if it's rich or not And you know, that's if it doesn't really scale and it's not really You know helpful for for the business So here is how location intelligence actually looks at the same problem So this this is coming from our database. Those are signals GPS signals from the mobile devices There is no map behind or anything. Those are just the these signals So if I link these to the The business value that location intelligence can provide and then you have on the descriptive side I can tell you for a specific period of time for a specific day How many people were in a specific location? Where did they come from? Where did they go after that? If I go more into the predictive and then because I see the patterns I can predict how many people will visit an area and I can even take into account things like events and all other different Location scenarios that can impact that prediction and Finally in terms of the prescriptive side if I know your business goal and I have all the information about your business I know where the customers are, you know, I just can't tell you exactly what what to do And it is very very interesting because You will see that in north Spain is kind of some popular road around there That's coming Santiago and you can see that from data Great another example if you have kids, maybe you might be familiar with this product and this is from a large corporation This product is targeting kids What are the questions that those businesses the consumer goods business have when bringing this product into the market? Well, first is where are the kids right where are the kids and where are the parents and where are the families? Where are the schools where are the parks because if I have to bring my product to specific point of sales And I want to optimize and maximize my performance and need to know those things, but they don't it is very difficult for them to know This is how location intelligence actually solve the same problem. So what you're seeing here, and this is coming from from our app This is the the map of Madrid all those dots are point of sales and the different colors are the Potential for that product depending on on the point of sale. So we can analyze around each one of the stores If there is a school if there is you know, like a park maybe and assign a score So we can describe, you know your network and and tell you Using a scoring. What are the points with the most potential? If we look at the prediction then Looking at your historical data. I can even tell you how much is going to be needed for each one of those and on the prescriptive side What I can do is I can Identify where your customers are and define the strategy and then just design how your sales force have to work and just Without you telling me anything apart from the objectives Great so Those were like a couple of use cases, but there are plenty of them across many industries That's just not retail and consumer goods is real estate is banking and there is government There is pretty much everywhere And the point here is that more than 80% of the business data Has a location component and that impacts the business decision. So if you think for a moment, you have stores you have Clients located somewhere you have deliveries and you have vehicles you have competitors So location intelligence is very relevant for most of the business and data has to do a lot with location But not everything is that Cool and that is it. So let's talk about the challenges behind Thank you. Yeah, so here we wanted to make it a little bit more practical and talk about some challenges The first one I'm going to start is obviously the data that is where everything starts and it's probably the harder So if you try to bring this into an analytical level where we're saying here is what trying to represent the real world The real world is very complex and has many different aspects so probably the first thing that you should do is you should divide the real world into data layers and That would represent the different layers in in the real world, right? And the thing is these layers they would bring data in different formats from different sources That represent different things that need to be treated in different way So how do you actually process that from an engineering point of view? Well, that's that's a challenge and that's probably means that you're gonna have to use a Bunch of different databases that allow you to manipulate the data in in different ways Now we don't have all these Databases in production, but we have used either all of them or have been prototyping with them Or we don't have it in production. So that's that makes it this an interesting problem And so now you have the sources of the data and you know where to put it But we're talking about a lot of different sources and I do blink we work with more than 60 So if you try to do that And in a following manual processes is gonna be to use and it's gonna be very time-consuming So how do you automate that? Well, you should use and something called data pipelines and these are tools and We use Luigi from Spotify and Airflow from Airbnb that allow you to automate all this process of capturing data Manipulating it putting it somewhere Okay, so now we have the data in our databases. Well Actually, it might be data that you cannot really use because you have to check the quality You have to understand if it really represents what the source is telling you and you have to Actually spend time and taking the quality maybe cleaning the data and or maybe you cannot actually use it at all because the quality is not Good enough and remember that this part is Really key because if you start with back data then the final numbers of the final statistics that you're gonna present to the user Then just not gonna be real. That's just not gonna be truth And you want to show numbers that are as close to the truth as possible And so this is something that is like the foundation of the whole Product and also you have to take into account the the bias of the source and the dais regarding data source means how this data is only applicable to some a part of the population and it can introduce a lot of noise and Distortionate the numbers that you are showing and pretty much every source has some bias You just have to be mindful of it and know your way around it Or if there is no way around it, you just cannot use that source and the final aspect that I wanted to comment on is Data regulation and protection things like GDPR. This is also very important And very very relevant and something to have in mind for every single source that you use It actually doesn't impact location intelligence that much because you are not storing personal data Every all the data that you're using is anonymized so you cannot identify individuals or anything like that But you have to check that all the sources that you are using They actually comply with GDPR and you have to always and Take into account that whatever you are producing cannot be used for that purpose, etc Another challenge I wanted to talk about is And when you get data in different format from different sources and Maybe it's not the granularity that you need. So this is a map of Paris of Data related to different areas of Paris and the center of Paris is treated as a single Neighborhood so this is not good because probably your user wants to have something like this Which is very granular and it goes all the way down sometimes maybe even a street level So how do you go from one to the other? It's a problem about statistics and mathematics But sometimes is is quite difficult Okay, so now we have the data and we have the right granularity Okay, so now you have to put that in the real world the real world that has streets and houses and schools and crossroads So that request a different set of problems and a different set of technology that we solved using Graph databases in our case Neo4j where the the nodes of the database represent street crossings and the edges represent the street and You can now mix this with all the the other data that you have This is also useful in order to show a routing or build models to understand how people are moving around the city and things like that Right, so now we have the data and now we know how to represent it now You have to build a front end and building the front end for location intelligence Is also interesting and challenging because you have to put a lot of data there and your user. She's not Using a mainframe in the cloud. She is using a laptop with four gigs of memory And this has to fit in the in the browser So actually one of the first things that we do when we talk about inserting a new product feature in our tool is thinking how much Information we're gonna have to send to the front end and if that's already about the threshold that is considered Natural because otherwise you can easily freeze the laptop of your user and for that we use View js in the front end and we use extensively node js to do a lot of precal questions in the back end and quickly send Stuff in parallel to the front end and then the final tech challenge. I wanted to talk about is well big data and Technology is like spark or hadoop. They've been around for a while They've been used for many different purposes in many different industries But they don't actually go very well along with location data and graphs because they're fundamentally based on map reduce and divide and conquer and You cannot really apply that to a graph this being some Advanced with your spark and some tools, but this is still a challenge that needs to be solved We're working on it or the companies are working on it and I'm sure there will be a lot of development in the in the upcoming years Great. So in terms of building a location intelligence product, there are of course many challenges The first one is is the user right is the user experience So I mentioned a few things like predictions and then you have graphs and then trends and then a map So how do you put all those things together in a way that your user is going to get it? It's going to get the value and not there's all So it's not that there's one user you have many different kind of users So you have the GIS experts those are the pro users They want to know all the in and outs of the platform and the location on the data and everything else all the options And then you have the business users that maybe they don't have any analytic experience. They just want to get the answer So all that complexity needs to be you know taken away From the user experience and just create create the right experience for the users So they can get the value Also in terms of user experience, I'm pretty sure that if you were to build a mobility app I mean Maybe you miss the interface maybe but as soon as you put a map like some kind of Information below and a massive button so you can order, you know, if you copy this you are okay more or less But there are many location intelligent solutions out there So it is very challenging to create these kind of solutions these kind of products So sometimes you have to lead the way Make sure that you are very customer centric You sit down with them and you validate how they want to get the value and In terms of the value and that's another point related to your experience Because there are so many different use cases that you can solve with data and you have the data You might have the temptation of you know fixing all those problems So, you know if you try to solve all the different problems at once you you're gonna end up with a product like this So this product it does pretty much everything, right? It's a phone is a fax is a printer is a copy It's got internet and Facebook. It's got everything, but nobody knows how to use it, right? So there's that that one person in the office who knows how to use it and produce the busiest in the office So you don't want to do that with your location intelligence product And then the the other problem is starting from the data. I see this problem quite a lot from Especially large corporations because they have these data. They think that they can bring value into the market So develop products and and everything else based on the data and that's just not the solution You should start from the pain point. What's the problem? Because most of the time The solution is not only one source of data is plenty of them And it's tons of technology on top to actually bring the business value to the business When talking about the data, then we need to talk about Partners and data location partners and Miguel Angel mentioned this before but this is very important You need to bring the partners on board where the quality is good. It is consistent So it's not just a provider is a partner. So it's very important that you're careful with your data partners There's so many memes in this presentation So one thing that is also related to the features is because you have all these different data points You might want to show them all right all these, you know ratios and penetrations and all those different things But then you use it's gonna be like this. Just you know, where is the value for me? Like they're gonna get lost And finally in terms of challenges, I think this is an important one There's a cultural change involved with this. So it's a it's a mindset So many companies they've been operating in a very specific way for the past 25 years When you you're telling them that they don't need to count people anymore that there's a better solution You know, that's change and people don't like change So that that's a that's a cultural challenge that that you need to face And also the users the users when they think a tool like this a location intelligence tool like this And they're thinking should I look for another job because you know this tool is gonna do pretty much everything So, you know, but that's not true. So you really need to be aware of the cultural change Okay, so kind of a few takeaways First When looking at business data, there is a lot of location involved with business data Actually, there is a lot of location in many of the business Location intelligence is now from the technology and data standpoints. That's mature enough now to provide direct business value That's that's already there And more importantly, it's becoming mainstream and it's it's growing and it's gonna solve more and more problems for the business Thank you very much We have seven minutes for questions Sorry, you probably you were planning to repeat the question. Yeah, no So the question is if here in Spain and the local market, although we operate in Spain France and the UK but the question was if in Spain a business are aware of the geospatial systems and kind of the location intelligence what the location just can bring Into the those business Well, there are different of different levels of expertise So you have very large corporations with the specific teams that they have their own tools and they kind of know about them And then you have the small medium business that they are not really aware But from our experience, there is a lot of experience and gut feeling involved in many processes Where data can play an important role. Yeah, I can take that one I was talking in general about the VIE systems and also I was talking about not just being able to host The coordinates also so meaning about specific analysis of data based on location I mean, I was I was talking in recent years Obviously location your location intelligence been around probably for like three four years So, I mean, I'm I don't know exactly what systems you have used. Maybe they were actually Right, but isn't Tableau is not really location intelligence. It's just a way to visualize Data, it doesn't allow you to actually analyze and extract insights from from what's going on In any case, just taking with a grain of salt I mean, we're talking about how initial business intelligence involved into location intelligence Those those kind of the message that we were to transmit More questions. Yeah, I'll take that one Yeah, there are many different data sources. So you go from public data sources that are available, you know from the government But there are many data sources that they come from providers So sometimes, you know, you can buy the data from a provider Sometimes you have to partner if it's a banking company, for example But it's never one source of data. In fact, some of the data we generate ourselves We have our own indicators To solve the problem if the focus is on the problem And you know when you're looking at the data you you gotta work with different partners and different data providers So it's not easy. It's a tricky part. Yeah, so When I was splitting into things so you have the business data that you can get from the business, right? The historical sales for the past two years, for example And then when it comes to location data, it really depends on the source But you know, you try to have like at least a buffer of one year one year and a half of data So it actually the prediction makes sense otherwise But again, it depends on when the data was available and how much data you are collecting But yes, it's very important very relevant question Yeah, from the technical side you just need to Figure out how to manipulate a stream of events at different points in time With that you can build a model What's going about just like simple machine learning And the problem there is that with location intelligence you have a lot of different Variables what's looking on of the hundreds? So the the tricky thing here is how to find senior model with all those variables and extract the ones that actually are impactful Well, thank you guys very much. Thank you