 I'm the CEO of Making Science to give you who we are, basically. So we are an integrated marketing and technology company, which is kind of an interesting animal, because we have like 80 software. We are in total 250 people here in Madrid, and we have 80 software developers, data engineers and data scientists, 80 creative and designers, and 80, which are marketing specialists, right? So we have access to a lot of data, and since we have the marketing inside, we are capable of basically putting in production a lot of use cases, you know, and actually measure the models. Of course, we are able to build all the ETLs and all the data pipes and a lot of models, and we are able to test a lot of things because we have basically the full value chain inside, right? And today, basically, I'm going to be talking to you about all the ATEC and MARTEC ecosystem, right? And how you come, we are building different models, you know? Of course, of advanced analytics and a lot of machine learning models to do activation, right? So as I said, you know, we are 250 people and we integrated a few companies, and we provide all the services to SMO, you know? Again, so we do it from software developers to digital marketing, to SEO, to design, to UX, UI, and we are able to go to do very quickly everything with, yeah? And we work with very large clients, and we have more than 120 clients, specifically in the area of advertising technology and marketing technology, right? And I would say, like, in terms of... I'm a computer science engineer and I finished a long time ago in the school and my thesis was on artificial intelligence in 1993 and was very interesting. I was working about the expert systems and all those kind of things on the first neural networks. And it's interesting, like, the new wave of AI and advanced analytics is coming from the UX, right? The UX and how we can interact with users and the reason, the basic reason is because that's where all the data is created, you know? Of course, there is a lot of... There's more areas where a lot of data is created, like IoT and many other things, but the interaction of companies and of consumers with companies in the different assets and with a different platform is basically which is creating the big wave of change, and of course, with the Netflix, with the Spotify, with Amazon, with Google and so on, right? And our thesis when we work with our clients is, like, the shift, you know, from traditional companies to companies that are doing the digital transformation is moving very quickly what we call smart companies. And if you were measuring a traditional company by the size of their fixed assets, of their traditional assets, you know, how many houses in my balance sheet, you know, how many plants I have, how much machinery I have, a digital company would be measured by how many intangible assets you have, how much software you have, how many servers you have, how many web pages you have, how many apps, how many users, what is your monthly average user. The smart companies are going to be measured by something which is different, which is, you know, how much data are you processing, how many models you have active, you know, in the different parts of your value chain from your supply, from, in your value chain, from the supply chains to your warehouses to how you interact with consumers, right? And that's what's going to change. That's what's going to define what are the winner companies of the future, you know, how many models you have active, how much data are you processing and how much data are you processing, right? And that's basically what we do. Of course, you can have data in all, in all the parts of the organization. We focus on the marketing and sales data. And it's interesting because marketing and sales, if your predictive model doesn't work very well, it's not a crisis, right? It's not like you are pretty, in a false negative, it's not a problem, right? Because you have an accuracy better than 70% is good, 80% is good. So it's a very interesting field for testing as well and be able to iterate very quickly with the data and with the models and with the consumers, right? So it's not by coincidence that I was saying before, like the oldest revolution is starting by consumers interacting with companies. One of the reasons is because the problem of failing is not big, right? If you are doing a model for an Airbus to fly or you are doing a model for a satellite to send to the moon, if the model is not good, it's a big problem. In marketing and sales, it's not a problem. Maybe a CMO is far, but that happens every day. So it's not a big problem, right? And the reason of the very fast pace of progress in that field is the fact that we are able to iterate very quickly with the data and that there is a lot of data which is produced. If this is, probably you are familiar with this, this is the McKinsey research, you know, of the McKinsey Institute of the opportunities of advanced analytics and machine learning in terms of what are the areas of the companies where AI and advanced analytics are going to have bigger impact. And you see like the biggest one is marketing and sales, which is interesting. We focus on, it's a very big area and the other one is supply chain management and manufacturing and you have all the different areas of the company like risk, service operation, pro-development, right? And it's not a coincidence as well because companies do spend a lot of money on marketing. So if you are on a sale, so you are able to optimize advanced analytics and machine learning is going to help you to make better prediction of the most models, you are going to be able to save a lot of money, right? So that's where we focus. I'm going to be telling you about different things we are doing, right, in the ecosystem, right? Also, this is the Garner CMO survey that is published every year, and basically Garner asks the CMOs in the U.S. is basically where are they putting their dollars, right? And you see like in 2019 and 2018, and we have the new one but it's not here, but basically the area where they are investing more money is in the marketing technology, so they are going to be spending less money in people, in agencies, they are going to be spending less money in media and they are going to be spending less money in people, internal people that is going to execute the campaign. So they also see the opportunity that automation and be able to use the data to get better return on investment on their dollars and it's happening. Interesting thing is what's going on in the advertising and marketing technology space. Maybe you are familiar with this slide. This is the Luma chief market view of the ecosystem and basically what you can see is the different players in the advertising technology and marketing technology, different companies that are doing things. So you can see here companies that are doing mobile marketing, companies that are doing mobile apps, creative optimization, video ad management, search ad management, social media, landing page, microsites, blogs, all the different technologies that you might use if you are a company, a CMO that wants to go to the market and you need a marketing strategy, landing page, a campaign and so on. So the interesting thing of this is all these are platforms and systems and all systems generate a lot of data, so there is a lot of opportunity. But I would also ask you to focus on the date, so this is 2011, and you already can see a crowded ecosystem. So at this time, all these systems and all these platforms were generating a lot of data. And this is what happened in 2016. So the ecosystem gets much more complicated and it's probably not by chance, so the fact that there are more and more platforms that companies are using and so on, it demands from you much more other platforms. So again, these are systems that are generating data, and in this one you can see organized by areas like advertising and promotion, content and experience, social relationship, commerce, data management. So more and more platforms, more and more data. And this is the Martik 5000, so this is 2018, which makes it a nonsense, I would say. So 5000 companies is a lot of companies, especially imagine doing a few RFPs to understand what they do. So I mean, why is this happening? This is happening also because of the previous slide of McKinsey. There is a lot of dollars, so there is a lot of money to be made if you are doing advanced analytics, data management, machine learning, machine learning projects. So big opportunity in terms of if a company which is focused on that, which is our case. But at the same time, marketing problems are the same. Right? If you ask a CMO what he wants to do, I mean, I would say, how can sell more? What is my optimal media mix? How is the consumer journey of my customers with this visitor buy or not? What pros do I recommend my consumers? If you were back to 1960 and ask a CMO what were her problems, it would be the same problems, right? The thing is right now the tools and the solutions are different, but there are new questions, right? Because we have all these platforms that I was explaining before. There are new things that you can do, for example. Should I bid for that visitor? Because now the visitor is a visitor which is visiting your website or is using your application and you have all that data and the question is should I show an ad to that consumer or not? And that was not a question that you could ask 20 years ago, right? Because when you were doing, for example, television, everybody would see the same ad. Right now when you are in an online platform based on data and based on predicting models whether you should show an ad to that user or whether you should bid for that user in the future or what is the propensity of that consumer to buy a product, right? Or what will be my conversation right next month, right? So I mean the platform sees, in one side, the problems of the CMOs is the same like it's been in the past, but at the same time the platforms are going to be helping you to solve new problems, right? So I'm thinking a bit what, let's say any client is doing what do they use information for and what are the kind of use cases of problems that we are solving for them and basically let's think about how marketing and sales use information and basically, I mean it does marketing and sales is everything, basically we use information for three things, right? To understand what happened, to understand what's going on and to try to understand what will happen. The what will happen in the past was, you know, basically I think my smell says that it's going to be like that so the what will happen was a very dangerous territory in the past, you know, and basically you have all the models built on your brain out of experience, right? And the good ones were able to predict the future and what is happening was also a problem in the past because the data was not ready. I remember like 20 years ago with business objects that the data warehouse, you know, the cycles were a month, you know, it was a month before you have all the data at the beginning, right? So what is happening, it's been a problem up to let's say 10 or 10 years but now it's a reality that you can really work on what is happening, right? What happened, you know, has been traditionally the domain of the decision making, right? Of information because you could see what were your sales in the past, what was the segmentation of your consumers in the past because you have all the data and reporting on data coming from different data sources, right? So in terms of what you do, you would do reports and dashboards, right? The past is being, let's say, a problem which has been solved by technology in the different waves of technology. The thing is now you have much more granularity, right? Because you have much more granularity on the data, you are going to be able to understand more things, get more insights and maybe probably make better decisions. What is happening, I think there is a big revolution right now in terms of what the things that you can do if you are working in marketing and sales problems because you have a lot of data pipes coming to different systems and you can design things that are going to be able to react real-time to consumers and be able to make decisions and to take actions on what's going on. And that also, the big revolution is what will happen, right? If we combine the prediction with the real-time, then you can be ahead of the game of your competition, right? And we see what Google and Facebook and Amazon do in terms of when you are able to combine the real-time with the very good predictive modeling what you can do for consumers in terms of everything. We will go later into different use cases. What you can do in terms of optimizing the user experience in terms of optimizing where you spend your dollars and so on. The reality is like all this data like most of the data is still unactivated, right? I mean, we do some surveys of our clients, you know, what do they do, they use their data from coming from all these platforms and even many times they don't use the basic reporting capabilities of all these platforms, right? So even though we are generating a lot of data so very few, very little data is still in marketing and sales analyzed and very, very few very little data activated, right? The problem is like it's a lot of right now where we see a lot of our clients is like, there is like the hype in terms of the data is a bit down because they are spending, okay, spending a lot of money in different platforms, right? And the return on investment is not getting to me, right? Because basically it's a lot of complexity and you only generate value as you can see when you do an activation of that platform, either without decision or without predicting model which is which is online. Just with introduction, let's focus on the different data sources and the different things we can do in marketing operations, right? So ideally, I mean when you think about the consumer journey of a consumer let's say a consumer that you don't know at all in historically, you know you can have basically a consumer which is around that you want to target and still you have offline and online impact or opportunity to see to impact that customer. You can see below you have the offline above you have the online and your customer will follow a process of considering your product from the awareness, consideration phase, purchase service and then on the federation, right? And in theory with in online, we should be able to track everything we do with them, right? And we are going to see there some examples of how we can do that because online platforms with the app server, you can track from the very first time you touch a consumer with an app, with the first app with an impression of the app. There are some limitations now with GDPR, but basically you should be able to track what happens with that consumer from the very first app you saw him to understand if you saw him another second app, if he visited your website, if he bought then if he came back because you have your analytics platform, if you saw him another app and so on. So as you can see now, it's a problem that could be solved because basically in terms, you have data from everything the customer did, right? So it's a known problem. It's a lot of data, but it's a problem that you can solve, you know, by different techniques, right? Which was a problem that in marketing was unsolvable in the past only using like very generic statistics, right? Because in theory from the very impact you have to that user you could create a profile of that user with a give him a profile in one of your platforms, right? And from there make different predictive models work for you, right? So I mean the accuracy of the models of course it's going to be getting improving as you get much more data from that consumer and he visits you more as you show him more ads but eventually you can create a propensity to buy your product from the very first interaction and then moving on to the charm probability of course it would be going to be zero, right? At the beginning because it's not your customer yet, right? Then you can understand you know what is their next best product, product recommendation from them and as any new touch points, a new data is coming around that consumer, the different predictive models should be able to optimize, you know, what kind of products are you showing that client. So in terms of what we compared 30 years ago of what is the kind of things you can do it's basically you can do everything, right? Because it's coming from zero data to basically all the data, right? Today if you are especially for online you can approach the challenge of doing marketing on sales, assuming that you have all the data which is typically a very good thing if you're solving any mathematical problem, right? Having all the data helps you. So what is the ecosystem, right? So basically I've divided here like three main blocks of platforms, right? That you can use to basically are platforms that are generating logs and generating data and eventually you can create models on top. First you have the ad tech ecosystem, right? Which basically you have first party, second party and third party data, right? And we are going to see which kind of platforms you have there, right? Okay. Then you have the marketing technology ecosystem which basically CRM or marketing automation on those kind of platforms, right? So basically mostly it's first party data and then you have the internal system that you've always had that you eventually can connect to the other systems, right? Of course in terms of with GDPR especially now the first party data is okay. I mean it's easier in terms to be lawful in terms of complying with the law and tracking having all the information. Second party and third party is getting complicated and in fact Google, Facebook, Amazon are more and more cutting the availability of second party and third party data in terms of what you can do in terms of the data, right? There was a big cut from Facebook last year in terms of the kind of segments they would offer you, like in terms of third party data and there is being a new cut from Google in terms of what can you do with the second party and third party data in the platforms, right? So to give you a more the idea of what kind of platforms you have in the different categories, right? So in the ad tech you are going to have your ad server which basically tracks everything you do with a consumer with the impressions, with the clicks in all the different platforms in social in search, in display and all that interaction will get locked into the database, right? So you have completely traceability of a consumer, then you have the search and social engines, you have the demand side platforms then you have the DMPs, right? And you have a few of the guys there, you know that kind of data that is generating all that data you can collect and you can attach to an ID you know, of course anonymous of a user and you already will be able to track so with that basically with the ad tech you are able to control everything which happens outside of your domain, right? If you define, like your company your domain is your website your application, right? So everything which happens outside of that work you can track with the ad tech ecosystem but you can track everything, right? You can track everything what is happening and you can attach all that information to a specific ID, it can be a cookie and so on, right? Then you have the martech ecosystem, you know, with your CMS your testing, your CRM, your information internally, but you can track also what happens in your domain in your application, in your website and there are ways to connect that connect external ID of your user of your user outside with the internal ID when they are visiting your website or your app, right? And then you have your internal system, right? That there are also ways to where you basically come up the different ideas of a different platform, right? So with all this information you can you can basically dump all this information into a common database, right? So you are to be able to track end to end the life of a consumer from the very first ad impression that it had with you to the very first bit in your website, to the very first time it bought your product to the recommendation you give to him and basically when it eventually it's not your client anymore, right? So in terms of optimization problem is very good, right? Because if you have a lot of data probably you are going to be able to fix the problem, right? So in terms of platforms and you are the typical suspects of things like different ad servers providers different social engines providers different DSP providers some of them there is integration of them right now change name and so on some DMP providers, right? So if you have a strategy of integrating all that information and connect it to a single ID basically you are able to track everything which happens outside of your website, right? Same things with your martech the tools, right? So basically you are able also then you are let's say all the interactions which happen with your consumer when you are inside your application on your website you can connect as well, all of them so you can understand, you can do analytics you can do predictive model on what happens within your app or within your website, right? And the last one what happens in your internal systems, you know that also you can connect that information with with the first what happens in your marketing technology systems and then with your advertising systems, right? So and then you start what I was saying and activating the data, right? If you think right now let's say with consumers where we have this setup basically we are getting I don't know gigabytes or terabytes of data, you know because for example I mean for us we are very experts in the Google analytics in the Adobe ecosystems, right? So we are in terms of the at-tech and martech so we have a lot of clients and a lot of integration and a lot of use cases which are very specific to Adobe and to Google analytics but basically in every and all the situations we have we are getting like gigabytes, terabytes of data every day which are which are fitting the different advanced analytics models and the different predictive models, right? So what you can do with this information and basically we build a platform that does all this in fact, right? So if you think about in this slide the different phases of your go-to marketing marketing, right? So basically you have a digital marketing phase where basically you are launching campaigns and you are launching ads to people, right? Then those consumers will interact with your digital assets you know, your web app your application your web, right? Then eventually you will sell something and they will deliver something to that consumer and when you sell something you go into the loyalty phase, right? So with all these data sets that we are collecting from the ad server from the DSPs, from the DMPs from the CRMs from the marketing automation tool to all of them but you can release lots of models right? and a few of them that you can do and all these models if you activate them, as we said I mean of course you can make decisions on them but if you are able to activate them online with APIs you are going to get a lot of savings of money, right? So what you can do in digital marketing for business strategies, you know for example you can do that in the platforms basically is getting all the data about the interaction of the consumer to create automatically ads for those consumers, right? So if I know that you visit a specific web page then the ad I'm going to show you is going to have the text automatically which is going to do and if you saw different images I can create that image for you as well in real time you can do lead scoring propensity, right? You can do purchase propensity to buy a certain product or to revisit to understand if that cook is going to revisit you or is going to buy and you can do personalization you can do clustering and you can do dynamic pricing, right? So those are and the good thing is like since these platforms are a standard those models are quite standard as well because when you get the Adobe data set and when you get the Google Analytics 360 data set and when you get the double click data set and when you get the cracks data set with the API there are some things which are non-standard but most of the things are standard so the models, the ability to put in place the models very quickly is easy I would say if you go into your website, you know with all the data which is collected you can do of course pro recommender you can do MBA you can do pro scoring you can do dynamic pricing you can do category listing order the good thing is like with all the basically with the information if you use Adobe Analytics or you use Google Analytics you can do exactly to 80% of what you could get into building a good predictive model for pro recommender or for cookie scoring then if you go to sales delivery, right? so you can do demand prediction you can do stocks, logistics product enrichment you can do alerts and we have different use cases of listings and when you go to loyalty you can do also MBA you can do personalization what is the difference when you have all these platforms which are online versus what's happening in the past is that basically the data is coming continuous to you and I'm an engineering background what's happening with the data compared to the other systems is similar, it's more like an industrial process rather than a traditional IT process when you think about a chemical plant or you're using about an electric plant what happens is that the flow is real, right? and you need to be able to manage the process of different machines and the flow of materials so basically the data products are more similar to industrial than which is to a traditional IT systems which is more like even base and I think that's going to be one of the big changes which is happening right now when you are doing about activating predictive models because all these platforms are generating the data in real time in a standard way so you are able to normalize them and feed the data to the different models and activate them in real time because the nature of the platform is online, right? so you don't need to do a lot of data processing so then activation so this case for example this is journey optimization, right? so when you have all the data coming from in this case from the Google or Facebook of your analytics so what is the journey in terms of how your consumers are buying your products so CPC which will start the journey CPC is cost per click so basically they click in a Google source or they want to organic or they want to other source on TPRG so once you have built the model basically you can understand the attribution and the different models and basically the data you have it in BigQuery and you have it in another in another database and the models are feeding continuously and optimize continuously, right? we publish this in this case with Berthi Germany, right? so coming back to the beginning why is CMOs are spending money on Marthecan on data is because it's going to save them money, right? so in this case in Germany what we did is of course an insurer spends a lot of money doing marketing and remarketing so remarketing when the banners are following you right? and booking.com when the hotels keep following you so if you already bought the hotel or you don't want to go to that destination anymore it's a lot of waste if that company is showing you more so it's spending money so in this case what we do is we will by the way I said we are very expert in Adobe and in GA360 all our models go in Google Cloud Platform we build everything on TensorFlow and so on so in this case what we collect is the Google Analytics 360 source the AdService source the Google AdWords source like four or five sources, you know, coming online from different platforms the model originally was a neural network then with an XGBoost but anyhow again in marketing accuracy is not a big problem the important thing here is like the model is running in real time so every time a new visitor visits the website or the app it gets an score of probability to purchase the product and what happens is that that information goes back automatically to all the buying platforms it goes back to Google AdWords it goes back to Facebook so what happens is like automatically you stop and start campaigns and stop and start bidding based on the purchase propensity of that user and of course the data is coming the model keeps feeding and we keep retaining the model with more hits for that consumer so the results so basically we save first 50% of investment right away so they were investing 200,000 per month so the following month they were spending 100,000 and the sales were the same okay so when you think about what is the ROI of investing in marketing and sales is very important for investing 200,000 to invest 100,000 the following month to sell in the same thing of products so what they did then is since they had the budget they invested the same 100,000 in other strategies so they end up selling 20% more right so this is probably in the web in Google if you want we've done this model with more than 20 clients now and it's showing the same results so basically applying these data models to optimize in marketing is saving all the time between 30% and 50% of investment right and then another one we just published this is in Spain also with Altamira in this case it was with Facebook so it's the same process so basically what we did is in this case you need to the data processing because Facebook doesn't allow you to pass information in the passing information with an API you need to do it in session in the web or in the application in this case we use the same of model we combine here with the city of visitors but also using lookalikes using Facebook models in terms of lookalikes and in this case we reduce the cost per action of Altamira 25% right so if they were getting a visitor to houses for 100 they were basically getting the same one for 75 so as you can see I'm finishing basically now marketing and sales it's a very interesting area in terms of building models because the martech and the ATTEC revolution is happening there is a lot of data and again it's not mission critical and there is a lot of money and basically that's what we're doing and that's all thank you very much I don't know if you have any questions no questions