 Good morning. Good afternoon. Welcome. Hello to everybody. I'm very happy that we are starting our next, I don't think it's the third or even fourth innovation talk. And I'm very happy that Olivier Boussard from the UPU is with us. He is the one who is making all this possible. And I'm as well very happy that Minat Jakobsen is with us. He is our presenter today and he is our guest in this innovation talk today. And I can assure you it's a very, very innovative topic we have today. But we are going into that in about a minute. Before that, Olivier, the stage is yours. Say hello to our guest. Hello and thank you, Martin. Hello. Good morning to everybody. Bonjour tout le monde. It's a real pleasure again to see so many people connected to our innovation talks. For your information, for those who are connected for the first time, the innovation talk series is the series of webinars that are being organized by the Direct Marketing Advisory Board of the Universal Postal Union. The Direct Marketing Advisory Board is a group of postal operators and private sector companies that are interested in developing and promoting direct marketing. And we do a number of activities, including those webinars, where we bring the views of external experts, of people that are not from directly from the postal community, and that are bringing different perspectives, different flavors, and innovation and innovative thinking to the discussion on direct marketing. The session today, I think, is extremely interesting. I'm super excited about it, because it's basically the merge between what we've discussed in January and February. For those of you who are with us in January, you might remember that we had a fantastic webinar on how to use social media to promote B2P marketing. And Martin, who was an expert in that field, did a fantastic job in explaining how to use LinkedIn in that regard. And we had our session in February, early February, around the importance of data in building direct marketing strategy and data-driven marketing. And specifically with our colleague from Dutch a post, we touched upon the importance of the quality of data in driving direct marketing. So I think today is really the merge between those two discussions, the use of Internet and the importance of data and data quality. And I'm very excited and looking forward to my presentation. My name is Jakobsen, he's a data entrepreneur, if I may say so. And we're very happy to have him today with us. He's going to share his views and experience and perspective on the way we can use the Internet in driving and harvesting data on the Internet for B2B business. So I'm very excited, very, very happy to have mine up on board. Thank you very much for taking the time with us this morning. And I wish you a very successful and very helpful webinar. Thanks Olivier for that nice introduction and as well for making this whole Innovation Talk series possible. I think thanks a lot to you from everybody. I'm seeing that a lot of you are saying good morning. So I see somebody from South Sudan, from the Arab everynates, from Algeria, Maldovia, from Slovenia, Cambodia, Namibia, I think we are already again all over the world, which I personally really like a lot. And I'm very happy that I can present you mine up Jakobsen as our speaker today. I know mine up for about, well, I think about 22 years now. So it's quite a lot of time. We know each other. And he is a very, very interesting person, to be honest, because he is an interesting combination of a statistician and a real person. So you normally know these statisticians are very strange people. They are in their data. They are not really in the normal world. But he is a good combination because he is as well a statistician, but in the normal world and has a great sense of business. And some years ago he had a fantastic idea. And that's what he is talking about today, about using the web, the internet as we all know it in a totally different way as we all use it. So we all use the internet in some way to look for companies and to look for information. But he looked at the same internet as we are all looking to, but in a different point of view in a different way. And that is what he will be talking about. And I really like that way of using data and using the internet in an interesting opportunity and making a business opportunity out of that. So very welcome, very warm welcome to you, Minot. We are all interested and I think the the stage is yours and Olivier and me, we will be now away a little bit and give you the stage to you. So have a lot of fun everybody and it will be interesting. I can assure you that. So see you later and Minot, the stage is yours. Fine. So I will try to share my, I have to, yes, allow everything. So I will share my screen and I will show you the presentation of today and thank you very much Olivier and Martin for the chance to present our small solution we have now in Germany and how we harvest the internet or what we are doing on the internet, new ways of data collection and what we are doing this way. Some small words on me. I'm 57 years old. I'm the owner of Mavankon. This is an agency for analytics in Germany in the dialogue marketing area and in the e-commerce area and I'm the founder of B2B smart data. This is a startup in the B2B area where we look for the internet for the company websites and what we can do with those information. So we have a little bit of big data, a little bit of artificial intelligence and but I will show you with the examples what we are doing there and so we will go here and so what about the content talk about the data where we are, what about the companies, how can we look for the companies in the big data area, what is the web intelligence for business, what we are doing there. We have a lot of examples to show you, the methods and best practices we are doing and last but least we are talking about our strategy for our company that we founded five years ago. Now we are seven people and we are working on this topic with seven people with the stuff that's all we have. We have a lot of computer, we have a lot of data but we can work with a small number of employees on this thing. We look at big data, what is the big data and so on. The big data is for components, we have a big volume of data that's very big at the end, we have set a byte of data, we have video and we have a variety of data. We do not only have text and numbers, we have text, we have video, we have chats, we have all this information that is not that good structure. The variety of data is a very big problem to get this information to be analyzed, to be understood at the end of the day. We have a big velocity, we have a very great data flow and we have a variety that means that the data, the uncertainty of data is the correct data, we have some other data and so on. These are the four Vs of big data, there are some more Vs but they are the main big data Vs. If we look at our world now we are talking about setabytes. Setabyte means I think it's a pitabyte, exabyte and then setabyte at the end and we'll have eight times more. In eight years we will have eight times more than in 2017. That means a very big data universe around us. The data is there and we can use this data to get more information about what we want to know. If you look at what's happening every minute at the moment in 2020, we have more than 200,000 Zoom hosts participating. We have Reddit, maybe you know about the game thing about in America, we have Reddit, we have 500,000 people to engage with content, we have Netflix, 400,000. DoorDash gives more than 600 meals every minute, every minute. In Instagram we have 350,000 stories. YouTube 500 hours of video every minute. Every minute, 500 hours of video is uploaded to YouTube. Twitter we have 320 new users. We have in marketing consumer spend one million dollar. We have Facebook uploads 150,000. We have WhatsApp billions of messages and WhatsApp in a minute, not in a day, not in an hour, in a minute. At the end we have a lot of people making video calls like we now do at the moment, like we do it as the system we have on Martin's. We have Microsoft Teams 50,000, we have Facebook and so on and so on. So the data is getting bigger and bigger and bigger and bigger. We get more and more data, but how can we use it? If you just go down to us and look okay, how many websites do we have around the world? How many websites do we have? Almost every company has a website. So we have a lot of websites that are private, but most websites are from companies. And if you look at the websites, we started in 91 with one website. It was a CERN. I will show it afterwards. And then came Yahoo and came Google and Facebook and came Twitter. And at the moment we have one about 2 billion. You can say one about 2 billion websites all over the world. These are the websites you can use. These are the websites of the companies. And every company has a website profile. And we can use this information that is concerned with this information, but over the website. This was the first website. This is a CERN website. On the right side you see okay, this is an experiment. The VVV is a WWW is an experiment. And they talk about this experiment, what they are doing there. You can look it in the old edges of the internet. And this was the first website, as we know, which started in 91. Every company has a website. So what we are looking, what we are doing, this is, these are my colleagues. No, I'm just joking. These are the robots we have. So we look on all those websites and want to make a structure from all those websites we have. Okay, at the moment we have websites from Germany in Italy. This is Germany was the base and now we are starting with Italy. But we can do this strategy all over the world for every company and for every market at the end. What we are doing in Germany, we have one about 17 million websites in Germany. We are looking on the website. This is a website for a company or for nearly a company I would show later on. And what we are doing then we call the data. We look for the data and put it down on our service. That means we have a customized web crawler. We do not use those standards. We have a customized web crawler because we have so big amount of web information we want to collocate from the websites. We take those data and put it in the database. So we have all those HTML source. We have all those pictures. We have all those programming information. And we have one about more than 3 billion words website combinations in our database. It's a big database with AWS where we located and we make it near time. I would say near time. So we look at every website in Germany once a month and some websites we look daily. And so we have a big database of all those websites. And then at the end we have an analytic framework around those websites. What we are doing. So we have all this information. What can we do with this information that we have in the database? We call it down and what can we do now with those information from the website? So this is what we call web intelligence solution. So if you look, if you want to Google or if you want to be, I would say, if you want to search on the internet, you say Google in Germany. And if you want to Google some information, you just say, okay, I want to Google some information for, for, for, for cup producers, maybe. And if you have those information for cup producers, you said, oh, the biggest one and so on. And we do this automatically. So we have all this information. We can give you all the cup producers in Germany if you look. But we do not only have the words, we just look how relevant is this word for this website. That mean, if you look at the Warrior Mail website, you'll see, you know, it's, this is the website of Warrior Mail. This is with pictures and all so on. But we are looking for the words and looking how relevant are the words for Warrior Mail if we compare it with all other websites in the web. And this is what we get from the Warrior Mail. So we get, for sure, we get parcels, we get tracked, we get missed, we get compensation, delivery and so on. Oh, that's funny. We get aftershaves. They talk about aftershaves on the website. They talk about horses on the websites. I think it's about the history of Warrior Mail, where they, where they talk about the horses on the website. They talk about the products, their batteries, not allowed and so on, packaging, customs, retailers, their workforce and so on. So we get a fingerprint for the Warrior Mail website. And this is the fingerprint we use. And this is just only 200 words, the most important 200 words. But Warrior Mail has more than 100 million, I would say 100,000 words on the website. If you look at the UPU website, yeah, this is this one. I made it just, I think, yesterday evening. Yeah, this is the UPU website that we're talking about. And this is what we get from the UPU website. So we say, okay, you have some, some actual information about Bywood, Abidjan, you see it here from Pakistan and so on. Yeah. And, but we have also surveys. So UPU made surveys, competition. Yeah, they talk about careers. They talk about the congress. They talk about stakeholders. They talk about postal. They talk about flights. They have a footprint that is, I think, the digital footprint and so on, useful and so on. And so we get a fingerprint for every company, for every company in Germany, for every company around the world. If we have this fingerprint, this is the collocation of the words, we can make analysis on this. And I will show you how we do this. So we have this fingerprint. We have the fingerprint. This is the count of the words on the website, but in relation of all information in the country, or if you're on all information of the world, and you can use this fingerprint to do what? Okay, we have a single fingerprint for every company, but we can also make a fingerprint for a target group. And I will show you this on the first thing. So if you have a target group, we are, we can find digital twins on your data set. If these are your customers, you have 231,000 customers in your database, they say, okay, they use this product. Who is the digital twin of this website? This is a question we want to answer. So we take your database, we take the information we get, and we take the ULs, it's a website information, and we make a data enrichment. If you do not have the information, we find the correct URL on this. And so we have the customer's websites at the second step. If you have the website on the second step, we are making an analysis. We're looking, okay, what about the similarities about these fingerprints? What is the DNA or some of that stuff? What is the common fingerprint of the whole target group? Yeah, this is what we want to do. And then we have an analysis like scoring or weak AI, I would say, artificial intelligence that we use. Okay, we have this target group. How can we find the DNA of this target group? If we have the DNA of the target group, we can use it and look for similar websites that are similar to your target group in all of your country or all of the market you want to reach. And then we find the most similar company website to your target group. And we can give it to you. You can mail it, you can call, you can make LinkedIn, like Martin does. And we also do it, you can make all those marketing information you want to. So you get to the target group and you expand your target group with this. Afterwards, you can sell your activities and you can use this to get more understanding of your target group. And at the end, you take those who react and you make go on and so on. You make a model extension, get better and better, because if you want to get a learn sample, you get a next learn sample, you get better and better in this enrichment process, because you're getting more and more precise in the analytics of the algorithm. But how does it work in detail? I can show you an example, but my computer is frozen. Martin, can you hear me? I can see you. Okay. It was just my small computer here. I'm not in the AWS. So a small example, we have a target group. It was a company, it was a distributor for electronic companies, making 1.1 billion euros a year on revenue. So they have all those components on those how do you call it, boards. Yeah, you have those semiconductors, you have those sensors and so on. This is what you find on those boards. And they sell billions, billions of components and they gave us a target group of 400 potential leads. They say, okay, this would be 400 we want to talk to. Can you find some similar companies? You said, okay, give us a 400. We found some similar companies. The question, can we identify more interesting leads in Germany? And the result was that we get a tag cloud for the customer and we find some very, very interesting things like technology, industrial integrated, some German words like component, this is components in German, but a lot of English words because of those companies that are in the target group are a little bit more international orientated. And we found this tag cloud and got all the information. We said, okay, this looks very, very cool and it seems to be plausible and then this is the target group we have. So let's have a look. We transpose this target group to the German market and we found another company. It's not working. Just a second. And we got some more dresses and in the more dresses we find the best practice, Gardena. Gardena is a seller of garden lanes. That means these things where you water the garden and for garden tools and so on. But why do Gardena need components that are on boards? That was the question we had. And they are in the branches like plastic industry or some of that stuff and they are also on the gardening industry if you look for the branches in the database. But if you look on the website of Gardena, the company we have in Germany, you find that Gardena is trying to get the smart garden. The guy sitting in the mountains looking on his smartphone and looking how his garden is watered at the moment. And because we find Gardena is making the shift on the website, we can give this address to our customer and the customer says, oh, that's pretty cool. I will never find Gardena with any other method because your method looks on the website. What is now the interesting thing for the company? And this is one thing and my customer went or some salespeople went to Gardena and said, oh cool, here we have some problems here. So our project office and so on maybe can help us with our components and they make a deal with this. So it works very well at the end and our customers are very happy with this result. And here what we find on the Gardena website was systems, was automation, solution, system, this is German word, some word intelligence and so on. These were the words we found at the Gardena website. Why Gardena is very, very near to our client and how we can expand the target group at the end. So we get a target group, we make a model with this role of words and find the next best offers and next best companies you have to contact for your business. On the other hand, we can also take this information for market insights. We can analyze the websites to profound deep insights in the market. One example on this, one back pass that was for buses, it was before Corona or before SARS, we had a customer that produces buses. And if you produce bus, you want to have some information how many buses to drive. And in Germany, the Federal Transport and Authority because of GDPR does not publish every address they just only give for some regions. Okay, in this region, we have seven buses in one month. That's all. But my client wants to know who is the owner of the bus? Do they still have the bus? And what we did, we get the information of the buyers of the buses for, I think, 1000 addresses. And we looked, okay, how many of those addresses can we found information about the bus fleet, the coaches they have on the website. Because the theory behind was that the coaches of the vacancy agencies that use buses will promote those buses that are very external, very big and very nice and very comfortable. They will show it on the website. That was what we did. And we found on 58% of the website, we found those coach information, those bus information, because they show us the bus fleet on the website. And on the 50, on the 42, where we do not find, there was something like airport Munich, okay, they have buses, but they will never tell about the buses on the website. And here's one small example that you can see on the right side. This is from my hometown near the Baltic Sea side. And they have every bus on the website. And they have it also with the plate information and which, which area it is, I think it's, it's, it's, it's time to answer this case or say something like that. They have all the buses on the website. And so what we can provide to our clients, we can make a database with all those bus information. And we have all those vehicle information and we have the changes because when they, when they sell a bus and they get a new bus, they will put, they will promote it on the website. And so we get a very, very near term information, how many buses does this client have. And so we can make a database and optimize the sales with this information. On the other hand, that's a very, very cool project was with dog breeders, especially in Corona times. Now dog breeders are getting bigger and bigger because everybody wants a dog in Germany, especially in Germany. And so we have those small, small puppies. And what, what's happening on the, on the website of, we have in Germany where the dog association or dog breeders association, they have one about 300, 3,200 breeders. But when we look in our web intelligence, we found 15,263 breeders websites. And what we found on those breeders websites is very cool because we have those litter announcements where those small puppies will come. And we have one about 30,000 litter announcements a year. So they say, okay, we have two dogs and they are paired. And now we are expecting puppies. And the owner of the website, the breeders says, okay, we expect puppies from Donald and Daisy, if you call the dogs this way. And approximately 30,000 litter announcements a year. That means that there's more than 200,000 puppies we expect in Germany. I think in Germany, we have all about 3 million dogs at the moment. So this is very small. But if you, if you say, okay, 10 years old, then we have one about market share, I think 60% of all those puppies that were born in Germany. But we have to think about the process of dog breeding. If you are a dog breeder, you get those puppies and you want to sell those puppies after eight to 10 weeks. If they are eight to 10 weeks old, the puppies, you will sell it to the new dog owner. So we take this information of the, of the, of the litter announcements on the database. We said, okay, they will, they expect dogs in April. Then we call the breeder in April and said, okay, is it going good? Do you have some, some puppies that say, okay, I have six puppies. And then we say, okay, wait a second, we will send you information about food, wet food, dry food and so on. You will get a blanket for the small dogs and you will get food bowl for the dogs and so on. And we give those packaging, those parcel to the dog breeder with our, our food. And then we get one parcel. Yeah. So here you can use that with one parcel. And after seven weeks, we sent another package with six parcels because now we know they have six puppies. And this is the parcel you overtake to the new dog owner. Yeah. And you give it, okay, here's the dog. And here we have a parcel. This is from a dog, from, from food engineering company, blah, blah, blah, and so on. And you say, okay. And what do you have? You have the blanket. You have the football. Yeah. The blanket is very important because the little puppy leaves mom, dad, brother, sisters, family, yeah, and gets a new family. So he's very, very nervous in this moment. He's very young. He's very nervous. But the blanket smells for the old family. So he's a little bit more, not that nervous, he's a little bit more calm because of the blanket. And the dog and the football also have those logo on it. So we have a combination of this and we have, and as a new dog owner, you will change everything. You have to change everything in the, in the, in the puppy's life, but you will not change the food. Yeah. So we would take the food because he's very nervous with a stomach and so on. And you will never change the food in this process. So what our client says, it was Pray Natal CRM, because it's CRM before the dog is born. And we know this because of the information we have on the websites. And so we can combine it with getting more parcels and more packages at the end of the day. Another example, this was for, for hotel stars. I'm sitting in the hotel now. And we made a map, web monitoring. It's also published in German and also blue angel certification. So we look on the web is the information we find on the website, the correct information. So for, for the yoga, this is a company in Germany that give the hotel stars in Germany, they want to understand, okay, do they make some advertisements with hotel stars? And they're not allowed to because in Germany, it's under law that you have, you have a regulation of the stars and so on. And we found 1228 cases of illegal style advertisement. So what to do? They get in contact with the hotels and company of the whole guy and said, okay, you have start on the website. You can have our certificate, you have to pay for it and so on. It was, it was a big, big for surprise. And they were very, very successful with this. So they get a lot of companies, a lot of hotels that checked their websites for the stars and made a great deal on this. So this is also published in German, sorry, it's German. But also for, for any certificate, you can find, you look for the certificate on the website, look in your database is correct, is it not correct? And we can use this information. So we can find all those information on the website. That is what we do. We have all those words that we can use it with the tag clouds you have seen, and with the modeling and with the twins. But we can also find employees for, for instance, on the team websites, how many, how many, if you're going to, to the doctor website, you see so many doctors, so many, so many assistants or so, so, so many women looking in the practice, we find who we are, the founding year, maybe like this one, if you look for the founding year of Microsoft on the website and in the official papers, they are a little bit different because they were founded in a garage some years before they were an exact company. We found all those products, we found the media information like software, maybe, maybe some e-commerce software, do they use e-commerce software on the website, we made some projects with just some pixels, which we can find, and also social media, we can all find the information of YouTube or Instagram or LinkedIn or whatever they have. So we can company and we have the correct LinkedIn profile of these companies and what we do. If you look at employees, this is personal information, but on the other hand, we are just only working with companies. So what's our profession? At the end, we have a low requirement for data because it is available in the company, so we can enrich this information. We have a very fast implementation guarantee, it's just only weeks, it's not months or years, it's weeks because we have a lot of data pre-processed in our database and you can use our proven and pre-trained AI. So we have a lot of AI around this topic, we have this big data and AI, and so we have a very high quality analysis. We do not find some blogs, we just only find companies, for instance, if you look at this case. And we have identification of customer signals, you can find this very good, and because of we only have business data protection complaint, it's data compression complaints and the implementation. It's not personal information, it's only business information for the companies that what we are doing in this area. And now, 20, yeah, 30 minutes, I'm in time, sorry for this short technical problem. Do you have any further questions on this topic? Thanks a lot, first of all, my nut for that really interesting presentation and all the information you just gave us. I see that the first questions are already coming, but let me start to really understand from a post to company point of view of what you're doing. If I understand it right, you have a database of all information on all websites of all companies in Germany or some other countries. And then you start to use all this information in a different way of things. So sometimes you might just use the address of the company so that I can send a direct mail to that company. And other times, you're using the information on the website, which is more detailed as in the dog breeders where you said, okay, a company is trying to get to dog breeders as customers in the B2B segment, and you found the best number of dog breeders. So there are a lot of companies out there on the market who are selling B2B addresses. What is the difference between your approach of big data and AI to these approaches which are already out in the market? Do you have an advantage or is it the same or is it just different? What would you say? Yeah, if you look for the for the Gardena example, for instance, you will never find this company with classical information on branch size or revenue or something like that stuff you have usually in the database. So the first thing is that you find your wife's target group. The second thing is that we have the website in this month or the regular website on this time. So we do have actual information. If you look at the databases, you have information two years old in the database because they upload and they call and sometimes I get a call. How many revenue do you make? I'm from Dunham Bread Street or something like that stuff, but they have to collect the data and the data when they collect the data next day it's old data. We have regular data on the same day. So we have actual data that's very near time. So we had some tests where we combined our addresses with old postal addresses from the databases and they had a rejection of about 10% and we just only had 2% because the addresses were very very correct at the end. And you can look on the difference between the websites in time. They changed the information of the address, for instance. There's this new people coming, new CEO, some of that stuff that is published on the website, especially in Germany where we have those rules of impressive. It's typical for Germany that every website has an impressive where we have to write down who's the owner of the company. Okay, I understand the point of more current data of course, but personally I like the idea of getting the right target groups much more interesting because what you're saying is if I go to let's say Dunham Bread Street or some of the other address offers in the market, I get data like the size of the company turnover, number of employees and maybe an industry sector like whatever, dog breeders. But if I want to have dog breeders of a special kind of dog, I could get them with them. But you could say, okay, I only get toodles or I only get German shepherds or whatever I want. So you have much more data because you use the data which is publicized by these companies on their own websites. That's right, yeah. I just got a question, is that GDPR compliant? It's GDPR compliant because we just only use the information of the company and it's published all over the world and you can analyze this. So it's allowed to analyze the data. That's a German law. It's a German law that you are allowed to analyze it. But I think it's probably European law as well. Okay, and if I now I am a postal operator, I'm the German post or whatever, the Italy post or the Cambodia post doesn't really matter which country. How could I well have an advantage of that offer or don't say that offer but that idea of crawling the web and using that data. How can I use that as a postal company? What would be my advantage? If you are interested in direct marketing area, you can find all those companies that use direct marketing that mean e-commerce companies or any other companies that have a big amount of direct marketing. Put it in the database and look what they are doing and find those profiles, those companies that are a little bit more in using postal mails than emails and so on. If you put in the database, you can say okay and if you are also a parcel deliverer, you can follow the website of the e-commerce companies which parcel deliver what they have. So I could use that to find my own customers as a postal company. So I can identify which are the companies in the market who use direct mail for example which would be my customers or they might use a competitor or something like that. Is there something else in it that I would like to offer a service to these? So I'm thinking indirect. So if I find somebody who is let's say dog breeders, he is selling to dog breeders, I could offer him as a postal company to give him better addresses of dog breeders and have some more direct mailings getting out of that. Yeah to get more parcels, to get more direct mailings out of that and you can, if you are looking for the e-commerce sector for instance we made a project where you, if you are a warrior mail, you can find okay which e-commerce company is using warrior mail or also UPS or Federal Express or something like that stuff. You can find those information on the website. Okay, so you could get competitor usage as well which is interesting. Okay, sorry, yes please. You wanted to say something? No, no, okay. We just got a question from Marco Provasi. He is asking whether the approach of data analysis is different if you use it to identify services or versus products? No, it's the same. It's the same algorithm for services and for product. It works in both areas. Because the company is describing what it does on its own website and it uses words, describe it, whether it's a service or a product doesn't really matter. Okay, that's what's interesting. Okay, another question. If I am a postal operator and I would like to start to do something like that, to use data from the web and to offer a service like that to my customers in let's say Cambodia or let's say Namibia or wherever on this world, what would you recommend this postal operator to do that? Yeah, start with your own database. Yeah, start with your own database. You have a database of your clients expanded to the whole country. So in Namibia, we find all those companies in Namibia and make a database of all companies with websites and then find the correct information for your products you have the postal office and who is nearby, who is a twin for which product in the database. Okay. And then get the right contact information about the people that are working in this area and contact them maybe on LinkedIn and like you make Martin and so on. Okay. Oh God, I cannot really read that name. Shung or Minghan is asking for some data on data privacy and the legal framework in Germany or the EU. So well, I can answer that one because I'm president of the German DMA. So I will try to answer that one. We have a European regulation on data privacy, which is called GDPR, the General Data Privacy Regulation. And this one works in all of Europe. But this is only for personal data, meaning per data of individuals. So it's not about data, which is open available about companies. And if I get it right, what minor is doing here? Sorry, he is collecting data about companies and not data about persons. So that is the reason why you're doing this in a B2B sector and you're not doing this in a B2C sector, right, Manat? That's why, yeah. That's why we do it in the B2B sector. So that's the reason for that. So the next question we are getting is how do you filter out the good websites or put it another way to do? You only look into secured websites like HTTPS. So do you make a difference between HTTPS or normal HTTP websites? No, there's no difference. We just take the information from the website. There's no difference between HTTP and HTTPS. It's the same information. Just another protocol. We have some problems with flash, for sure, for flash websites, but they are not common anymore. We run about 2% of websites we can't call because they are too slow. They have some security we can't use, but we have 98% of websites we can call. So how do you get the websites, the domains? Because how do you get this information about which websites are out there? So if I put up a new website saying, well, B2B, UPU, B2B, smart data approach.com, you don't know that website probably in your database. How do you find it? Because usually you make some digital things and we can find your website. Maybe if you are getting on a map service, using a map service where you can point your address, we find you there. Or if you go to the yellow pages, we find you there. We have some different sources for this. So you build up, let's say, a library of all websites in the specific country which will use them. In Germany, we have 70 million websites at the moment. I think 3 million are linked to companies or maybe 3.9 at the moment because of the dog breeders. This is not a company. This is a part-time business for the people. But for us, they are business addresses at the moment because of our algorithm. Olivier is asking whether you collect the email addresses as well. Sometimes we do this for our clients, but in Germany, you are not allowed to use those email addresses without double opt-in. Other countries like US or not Europe, but I think outside Europe, it's allowed to use this email information. We have some clients that use email information without double opt-in because they say the email address is interested in the information. Okay, but it's possible. What else information do you get from the websites? You made an example that you get this fingerprint from all the votes. You had an example that you used the photos or the description of the photos with the buses, I remember, if I remember this right. You just told us that you used the email addresses. Are there other data you scrapped from these websites that you are using? We scrapped everything. We have some programs, some algorithms that scrapped the owner of the company, the postal address, the email address, and everything, the social media accounts, and so on. We have some different algorithms on those topics that use this information. Okay. How many people are working there and so on? Who is the doctor? Who is the sister and so on? You collect all this information about the website and then I can use it in a different way, depending on what I'm looking for. If I'm looking for employees, you could deliver the number of employees. If I'm looking for whatever else you start to collect or not collect but use that data, but you're collecting everything you can get. Okay, I get that. Any more questions? I have to look whether there are more. Yeah, we already had that one from Olivier as well. How do you can use that data to boost postal direct marketing mailing? It's saying that you would start to offer that as a postal company for their clients who can use this data and then use more direct marketing in the B2B sector. I think we had an interesting webinar on LinkedIn in December with this group, and I think you're working together with LinkedIn in some way as well. Can you go on that? How you combine your data with LinkedIn? Yeah, because when we look at the data of the company, we can also find which LinkedIn profile is linked to the company. If you go to this approach, you can use, we just wrote a book in German about hitchhikers guide through LinkedIn, where you can just click on and find the correct people in the company you want to contact. So you identify saying, okay, company A might be having the same fingerprint as your target company, so it might be interesting, but you don't know which other people in the company. So you're linking that one to LinkedIn, and then I can use LinkedIn to identify who is head of sales or who is head of marketing or who is doing the direct marketing in the company as well. So let's say from the company level to the employee level, I go to LinkedIn to use that to go down on the employee level. To make social selling, to get in contact with the people, say, okay, are you the right person for my product and so on, to get in contact. We are using it also in our company, it's very very successful at the moment. So I'm just getting two more questions. So one is from Mohamed, he said he's using yonus.com to check the website. Is that good enough to lay on it? I assume Mohamed that you are using yonus.com to get some information about the website. So how fast it is or how often it is made current or something like that. But probably you won't get the data from yonus.com. Do you have your own crawlers miner to do that or what are you what are you using? We use similar things like yonus.com to find okay, do I have a redirection or some of that stuff and so on and our information and all those things. It's good enough if you do so, but if you want to get a little bit more deeper in it, you need to quality information about it. So crawling means you have some kind of computer program which goes to that website, looks what is on the websites, download all the information from the websites onto your service or your virtual service, and then you can use the data. So you're going deeper than yonus.com is doing. The other question is from Olivier, he's asking about B2C. Is your approach being used in that space as well? No, not at the moment. Because of GDPR restrictions in Europe, we do just only for business websites. You could probably do something like that if you have the full access to all the social networks. You could probably use the same methods, but it's just not allowed to do it that way, right? We are not Cambridge Analytica or some of that stuff. But in a way, it's what Cambridge Analytica did on voters in the US. It's probably the same, but it looks a little bit like the same methods, collecting all the data of the voters using some AI to get some predictions of what they might behave or what they might vote, and then you can start to address them. So in a way, it's the same approach to use data, but the difference is that what Cambridge Analytica did wasn't allowed, and what you are doing is very much allowed, and it's not a problem in the B2P space. And I like the idea to use open available data from websites, where companies are talking about themselves and telling what they are doing, and just use that data which is open available in a different way than you normally do it. That makes sense. Okay, someone is typing, so there are some things. Oh, there's another question. If the conclusion is as good as your data, what are the main pitfalls and accuracy you had to deal when you're processing your data sets? A common mistake and misleading conclusions, we should be careful, but that's interesting questions. Thanks, Daniel. So what are the main problems you encounter when you do this data analysis? Yeah, sometimes it depends on the target group. If you are advertising company for pens, for printing on pens, everybody needs pens, so the target group is very wide. We made some projects there, we were not successful, because our method is too expensive to compare it with the addresses you get when you go to business. So if the target group is too broad, it's not specific enough, then you have a problem because well, you can only, well, it's better to find specific target groups. Okay? That's right. And in former times, in the first one or two years, we had some problems with blogs on this information because blogs are not companies. But so we have to filter it out and we made an algorithm for this. And so this pitfall is now a way. So it depends on the precise of the database. And it depends on not to count every word on the website, but to compare it with the whole area of your country. And this is the main point why we are so successful in Germany, in German markets, because we do not count how many words are on the websites, but we combine it with the whole area of Germany. So yeah, I think that was one of my questions actually when I read that from Daniel, there are let's say common words like the or or whatever one or something like that. So you get rid of these common words in combining or in comparing what is on the website in comparing these two others because everybody is using the you don't use the word the because it's just element noise is what Daniel is saying. That's actually what you're doing with this comparison. Okay, that's interesting. The noise is the HTML language. We have a special algorithm for this to get all those information. So we only get the clear text on the website. Yeah, this is what we use for our database. Yeah, makes sense. So thanks a lot, Meinat for not only presenting this but answering all these questions. Thanks for the ones who asked questions for for helping us in this dialogue. So making a real talk about that and dialogue out of that. And I think Olivier wanted to tell us a little bit about the next one. So Olivier, I'm just giving you the possibility to open your microphone and your camera. Here you are again. Hello, Olivier. Thank you, Martin. Hello, everybody again. And thank you very much, Meinat for that very, very interesting presentation. And I think you have provided some very, very interesting elements for our viewers today in how to maybe go to the next level in terms of, you know, using data analytics and building up marketing campaigns for clients. It's really about the whole point of that discussion is really to bring new ideas to the participants on how to use that gold mine of data that is available there and how they can enhance their value propositions for their direct marketing clients. So that's really the point of that discussion and the point of our innovation talks is to bring a different view, different perspective and to go to the next level. So thank you very much. And I hope that was really interesting to everybody. The next session will be in April, 8th of April. We're going to take a little bit of a break in March. And we're going to meet again early April. So everybody, please take a note in your calendar. 8th April, you will receive an email from Abby and or myself or Martin or the three of us together on the topic. We're still working on it, but we're sure that we're going to bring you a new perspective again on the way you do direct marketing, on the way you use your channels to propose and offer a new type of campaigns to your clients and how you enhance your value proposition to your clients. So that's the whole point of our talks is to bring new dimensions and new perspective to your daily business. We hope that you really enjoyed that session. Again, thank you, Minut and Martin back to you and see you very soon. Bye. Thanks Olivier for that final words. Thanks Minut once again for being our presenter today and that interesting topic. I think it's really interesting to see what you're doing with data. We are, everybody of us is going to a website every day and we just go to one or two or five websites. But what you are doing to 17 million websites is totally different. It's the same data, but it's a different approach. And I really think there's a lot of in that approach for direct marketing business. So thanks a lot for that. I see a lot of thanks to you from the audience. I'm saying thank you to the audience as well. Have a great day for the ones where it is still morning, like in Europe have a great afternoon or evening for the ones who are still already on the other side of the world and when the day is going to an end. And I hope that we all see you again back on April the 8th to the next innovation talk. And bye bye from Europe to all over the world. Bye bye.