How Data Keeps El Al Israel Airlines Airborne

Nate Nelson: Hey everyone, and welcome back to the Data Masters podcast. My name is Nate Nelson. I'm sitting as usual with Mark Marinelli from Tamr. He's going to introduce the subject and the guest of our episode today. Mark, how are you?

‍

Mark Marinelli: I'm doing well, thanks. Today, we're going to hear from Ido Biger, the CDO of EL AL Israel Airlines. He will talk about what data lessons he applied to the airline after working at a major Israeli TV station. He'll talk about how embedding himself in different EL AL departments made him a better data leader, and what he teaches the next generation of data leaders as an adjunct lecturer.

‍

Nate Nelson: Okay. Here's my interview with Ido.

‍

Nate Nelson: Ido, thanks for speaking with me. You went from yes, which is one of Israel's largest TV providers to EL AL, which is Israel's largest airline. What was that like? Were there any challenges that you faced in making that transition?

‍

Ido Biger: Well, yes is the most advanced television company or telco company in Israel. And now, lots of what I brought from the telco industry to the airline industry are things that were not necessarily connected. And it allowed me to think out of the box when I came to a traditional industry, such as the airlines and bring new ideas towards them, such as I was working on recommendation engine while I was working at yes, right? Like such as Netflix. So you could choose your video on demand assets that you wanted to watch or programs, and immediately it will recommend the next one, that you could find relevant for you, according to the person you are, et cetera. But in flights it can be the same, suggesting you the next destination or understanding what would be the best relevant route or ancillary for you.

‍

Ido Biger: So you could bring more ideas from those experiences that I had in telco to the airlines. Another example is what we did with all the field of set top boxes. So predictive maintenance that I was working with yes, we could implement it directly at EL AL as well, because there is predictive maintenance to set top boxes but at the same time could be predictive maintenance to airplanes. Once I understood that there are many, many opportunities for us to work with the different business units, one of the biggest gaps in the airline was connecting between the data of the different business units. So for example, you can't really divide or treat separately the data of aircraft's problems or issues and the customer satisfaction. You can't really disconnect issues with on-time performance to the customer satisfaction. So by collecting all different types of sources, such as I did in the telco industry, it allowed me to think again, out of the box and not in the traditional way of the airlines and bring those ideas directly to a quick and very good implementation at the airline at EL AL.

‍

Nate Nelson: So would you characterize this as an industry or a company phenomenon? Like, is it the case that airlines just in general, happened to be more behind or traditional van telcos?

‍

Ido Biger: In the telco industry, first of all, the data analytics was far more advanced, for sure. You can think of, I don't know, American examples of AT&T and Comcast that are doing... Or even at Netflix, as we all know, are far more advanced worldwide in their analytics capabilities. So in telco, again, it was very part of the business or the day-to-day routine to be data driven. So no matter what data you collected, immediately it became something that was inherited within the operational systems themselves. At EL AL, the operation side was very disconnected from the analytical and maybe commerce side. So, only when someone came with understanding that there is no... As mentioned before, you can't really divide between the different organizations because an airline is a huge company that's constructed out of those different companies, right?

‍

Ido Biger: There's the maintenance and engineering company, and there are the air crews or crew assignments and all the different around operations. There is all the thing about food. It's a huge catering company and a food company, and there's of course the service and commerce and they all treat themselves as different companies within this huge conglomerate and what data allows us to do very fast is to connect all the different dots of this organization. So once again, because analytics in the telco is a huge part of the operational processes, immediately when I came to the airline and I understood that this is where I want to be. And sometimes when you work in a single organization for most of your life, you can't even imagine things that are already out there, but in different industries. So my main goal was to get to a situation where data analytics within the airline becomes operational acts as implemented in the past two years. And I can give several examples if you'd like.

‍

Nate Nelson: Yeah, how did you pulled all this off and bring all these data silos together?

‍

Ido Biger: At yes or at the telco, most of the data, or most of the operational systems had a certain reflection in the single data warehouse, right? When you see a single source of truth, et cetera. So, whenever some analysts wanted certain information, certain data to be extracted to him or to use by him, he knew that it's probably is going to be in the data warehouse and sometimes you need to add a few more sources. At EL AL, for example, 30% or maybe 40% out of the organization data was there. So 60% was not. So when someone wanted to connect different sources of different types of data into their analysis, we had to come up with a new way of methodology because to say to the organization, please wait for a few years, we'll collect everything to a single data warehouse, and then we're going to start. We know it's not going to happen. It's not even relevant to stay relevant with this some sort of methodology.

‍

Ido Biger: So what we created was a system called Rapid BI, where we could collect all the different sources into this huge, let's call it a data hub. And on top of that, we created the semantic layer with business entities. And whether it's views or whatever, we could connect ourselves with the visualization tool. In our case, it was Tableau. We connected ourselves to the semantic layer that behind the scenes was quick and dirty, right? So, we collected immediately all those different types of sources and all the different types of data to this single semantic layer and by that we could implement, we call it data products in a very fast way within a week or two. So immediately the customer could see final his product, even though behind the scenes, it was quick and dirty. It wasn't really structured well in a single data warehouse as we are used to.

‍

Ido Biger: So by that, we could kind of close the gap there, the huge gap between the situation where a lot was in terms of data analytics with other companies. So within a year and a half, pretty much all of the organization units got their data products and started to work on them. And not only that, in a matter of a week or two, we could come up with a new version because the changes were very, very easy for us. And only when we got to a situation where he really loved his product and used it on a frequent basis, we then started to engineer it behind the scenes and make it solid, like a solid, robust solution. And by that we could really implement new products in a very fast way. The first example for that was... For example, the on-time performance model, where we could connect 14 different systems that weren't represented in the data warehouse, when within two and a half weeks, we came out with a very good solution for the operations VP.

‍

Ido Biger: He of course showed it to his other peers in the management, and immediately it became something like everybody wanted it now, especially that we use the mobile version, of course, so we could walk with the data product and use it not only when you sit in front of the desktop. A lot of our pilots and people in the operations are using tablets. So one of the benefits was that it was presented very well in tablets and not only in desktop and it started to became operational. So we could, for example, prepare to the station managers, a full overview of what's going to happen to them, not just what happened to them, but what's going to happen to them in terms of disabled passengers and people with... Again, where wheelchairs, children, special needs, and prepare their flights or their preparation to the flights that are going to happen from their station in a very, very fast way, for the company to act very well up front.

‍

Ido Biger: So it became a huge part of the operational system itself and not just dashboards and data visualizations. And only then, I could really understand that we're not just implementing a way or methodology that will help the analyst to figure out better what happened, we're preparing models and preparing tools and data product that will help the company in the operational system to behave very well and to prepare to the near future and sometimes the far future in a much better way.

‍

Nate Nelson: This brings me to a bigger question, which is that when you go to EL AL, your job isn't just to introduce newer and better ways of doing data, it's that you have to get everybody around you who is used to doing things in the old way to get on board with your new plans. So how was it to gather all of the people around you, all of your colleagues behind your ideas, and importantly, did you experience any pushback?

‍

Ido Biger: Well, that's a great question, but the thing is, I'm not the issue. They're the issue. My job is to make them kings. So if people that are utilizing data becoming better at what they do, I did my job correctly. Because I'm not going to be the issue here. I'm not going to be the one who's going to push them to do things. I'm going to be one who is going to show them how by using data, they're going to do their job better, but they're the domain expertise. I will never be good in operations as the VP operation. I wouldn't ever be good in service as the VP service. So I'm going to be the one who's going to be behind the scenes for them to be able to do a better job. And that means that I need, first of all, to understand what they're doing. So for example, my first two months was just spending time on learning the business. And I joined technicians dealing with, taking care of their routine, procedures of the airplanes.

‍

Ido Biger: I joined the pilots and air crews during those flights to understand what they're doing. Flight attendants and pilots during their flights. I joined the station managers to understand their day-to-day routine. The catering guys, I helped them prepare meals for the airplanes, et cetera, the cargo guys, et cetera. So that means I'm there to be able to collect the relevant data for them to get, or to understand better decisions and maybe to confront their pains. So once I got their trust and understood better their business, understood the data that I had, that might be able to help them and of course, find use cases that are really painful for them, I was able to gain their trust and make them become better.

‍

Ido Biger: So whenever someone came with a new data model or a new way of presenting his improvement, thanks to the data, my role was to just maintain that I should keep feeding him with more and more relevant products for him to get better, for her to get better. And it practically like a train that started... I was the one trying to push this train, but once it started to roll, everybody wanted to get on the train. And on top of that, I started to create a culture of self service and we called it a data literacy program. And by that we understood that in order to grow exponentially, I know as a term being used now in the COVID aspect, but to grow exponentially with using the data, we had to come up with a training program for the business analysts to make them better, to make them good users of the BI products.

‍

Ido Biger: And by that, we could focus on things that only we can, and they can focus on analytics, they can focus on examining the data, preparing data and the right relevant sources for them and of course getting the insights. And by that, by having this data literacy program and even someone to lead this program, our head of data literacy, which was a unique part or unique role in Israel, we could easily work with a thousand customers, internal customers on making them better according to the goals that they set up. For example, I called... Let's talk about SQL. So red level of SQL is someone who just knows how to write, I don't know, select on tables and do nothing, well, just get a table. Yellow would be someone who can join tables and do something on top of that, like complicated work loads and someone who's green in SQL is someone who knows how to write procedures and really knows how to work in difficult means of SQL.

‍

Ido Biger: That means we came up to an analytical unit manager and told him, "How many people in a green level do you want by the end of the year? Let's help you build a program for you to get there. How many people you have enough in the yellow level or red level for you to work with?" Because not all of them wanted SQL experts, right? And by that, we did Tableau, same with colors and BI [inaudible 00:15:47] of SAP, we worked on Excel, even Excel. So maybe you want someone who knows VB or to record macros in your team, analytical team. And sometimes it's not necessary. So, we created this program to help all those hundreds of analysts or hundreds of people that are data consumers set up their own goals and I was, again there to help them achieve their own goals.

‍

Ido Biger: So if you ask me about what is the main success factor that I see in this implementation at EL AL is that, I wasn't the issue. They were the issues and I was there to help them become better at what they do and their success of course, is my success.

‍

Nate Nelson: Yeah. This brings up an interesting point, because on this podcast, we've talked to a lot of CDOs, CIOs, and most of the time we're talking about technical matters. How to bring data silos together, how to effectively master your data, this kind of stuff. We spend less time talking about the people component of these jobs. Ido, if you were to say, how much of your success can be attributed to the technical expertise that you brought to EL AL and how much to just your people skills, your ability to manage people and get the best out of them?

‍

Ido Biger: I think a lot of organizations are confronting with issues of a chief data officer and what kind of skills they should have, or he or she have. I would say there are three different types of approaches. The first one would be, let's take someone who's very good on the business side, make him like a senior responsible for the data side. He will work with the IT, get all the information that the organization needs. He will have a good analyst with him and he will support the organization needs. Most of the time, especially in Israel it didn't work. And why, because he just became like a premium customer of the IT and not necessarily all the people saw him as this kind of a new position that wasn't really needed. But in some organization it might work because it's someone who really understands the business and really understand the domain, he is a good domain expertise in the business and he's now focusing more on the data products on the business side.

‍

Ido Biger: The second approach is the more technical one. And that means take the BI leader or the BI manager and make him a CDO where he is in charge of the data platform, the data governance, all the issues around it. And of course, dealing with all the different regulatory aspects of the data. It happens a lot and I always used to work with a lot of CDOs in the banks and their main role was that, work on the data platforms, choosing the right technology, understanding the data management needs, working on the data governance program and in the banks, it was again, data lineage. And that was a huge issue that needs to confront, but it was more a technical senior level management.

‍

Ido Biger: I have the privilege on working on both sides of the equation in this case, because in order to understand the business and fully focus my efforts on the right things for the business, I need to work with the people, with the people on the business side and that's a very, very, very critical aspect of my work. To be their best friend, to be their ally on whatever they're doing. So I got to a situation where I have, let's call it colleagues that nobody before me could reach, right? The pilots and I don't know, whoever in the airline or in the television company. No BI manager has ever, before my role joined the technician in his daily routine, fixing problems in the customer's house. No BI manager before got to on a plane and flew to Italy and back just to understand the work of the pilot while he was taking care of a flight, the captain of a flight. And nobody spent two or three days with the station managers to fully understand how they're dealing with passengers in their own domain.

‍

Ido Biger: So by that, I could really, first of all, again, got their trust and really understood what I should focus my efforts on. Now, on the other side, I'm very proud to have 17 years of data engineering background, and I really understand what they need to have in the terms of data platforms and everything, to be able to deliver what I want to bring to the business side. So for example, my condition to come to EL AL was to take the BI department under my supervision. Now, if not, I was just another customer dealing with analytics. So whenever an organization thinks of hiring a chief data officer, I would suggest that it would be both a technical person, someone who really understands the bits and bytes of data engineering and of data analytics, if possible data science, because he's going to lead the data science team probably. And of course invest a lot of time in his domain expertise.

‍

Ido Biger: So I'm less afraid for example now to move to a different industry, because I have a very solid background of data engineering and analytics. So I need to invest most of the time in the new industry that I will go to in order to understand the domain, but to have this solid technical background, for example, let's talk about insurance. If someone is expert in insurance and he becomes now the CDO, he won't be able to fill up the gap of 17 years of data engineering, right? So it will be very, very difficult for him. So if an insurance company wants for example, to hire a chief data officer, I would suggest a very, very strong data engineer with maybe some background of financial industry. But the most part would be again, the technical side of it, but the people's approach. As you mentioned, it has to be a people's person, otherwise it will just become another IT guy that needs to support someone and sometimes they even build teams that should translate business needs to IT language.

‍

Ido Biger: If the CDO doesn't talk business, he can't be a CDO. If he's not talking IT fluently, he should not be a CDO. I believe in both aspects of the equation here and that's like the fundamental part of being a chief data office.

‍

Nate Nelson: Let's strip away all the details for a moment. What lessons can you share about making a successful transition between data roles in different industries? So how can other CDOs make similar jumps to what you did?

‍

Ido Biger: Start with a very strong background on the fundamentals, right? If you have good fundamentals of the data engineering part and data analysis part, continuously work on that. And I teach my students, the most important characteristic for someone here in our days is to learn, to keep on learning all the time. You never stop learning. So in terms of technology, keep investing your time in learning new technologies and being better at what you do professionally. Second, understand that no matter in which industry you work on, you will never be the best in each and every one of the domains. And by that, I mean that no matter which organization you're going to find yourself at, learn the business, invest the time, especially in the first few months, joining the people and do what they do.

‍

Ido Biger: Sit with the customer service person and listen to his call or her calls. Join the technicians, join whatever industry you are at, join the professionals. Work next to them side by side with... Whatever it is, you have to be part of the business. You are the first line of business in their IT, and you are the first line of IT in the business. And by that you can't divide it and just focus on one area or the other. And the second thing I would suggest is get a very good BI manager working with you, because he's going to be your best friend. And by the way, he's going to be the one you're finding with the most with, because he's going to want to maintain the solid structure of data, et cetera. And you will come up with this new business ideas, et cetera, but you understand his world very good.

‍

Ido Biger: When you are a chief data officer, and you're not understanding the technical side of it so much, you will come up with this business aspect of things and you want to really understand the language that he speaks with you, or she speaks with you on the BI side and learn both... So to make long story short, be very good at what you do in technology side. This is the thing that you will take with you to wherever you go. Invest time in the people and the business that you're joining now. That's the most important part. Don't just focus on where it's easy, focus on what you don't know, especially in the beginning.

‍

Nate Nelson: My last question, you also teach data to graduate students. So, Ido, what makes you passionate enough about data that when you're done at your data job, you then go and do your other data job?

‍

Ido Biger: Well, now you sound like my wife. I really love teaching. I'm a basketball coach in my profession as well and I really like to do something well and then teach the new generation how to do it. And of course, the most gratifying way is to see someone who becomes better than you. In basketball it won't be very hard, but in data, it is a challenge but I'm very happy to see people that are really becoming very good at what they do even much better than me. What keeps me motivated is two things. First of all, again, seeing the success and seeing the eagerness and enthusiasm in the eyes of my students. When you teach, you immediately see, even if it's five students out of 10, even if it's 50 out of the 100, you see the ones who are taking it, one step forward. And I'm really enjoying it. It's like, again, investing in the future.

‍

Ido Biger: And the second aspect of it would be something more selfish. Because when I teach big data technologies, I have to understand them. When I work in a certain organization, I only know the technologies that are within this organization and I work with on a daily basis. So for example, I had to teach about [inaudible 00:26:42] for my students, even before I implemented [inaudible 00:26:46] myself in the organization. So I always have to keep in touch or keep online with the new technologies and new methodologies for the university first and then implement it sometimes in my organization, whether it's good or not. So you always keep aligned to the latest news in those industries. I think it's wonderful and it's worth the time that I invest in it because it's not for the money, it's for mostly for the internal part of the soul.

‍

Nate Nelson: Ido, that's all I got for you. Thank you.

‍

Ido Biger: Thank you very much, Nate. It was a pleasure.

‍

Nate Nelson: Mark, that was my interview with Ido. Could you maybe provide us a parting thought to leave us with?

‍

Mark Marinelli: Sure. A point that Ido repeated is the importance of learning about the business as a data leader. As we heard, Ido took a very hands-on approach to learning about his colleagues' problems. And then he thought about how the data could be used to solve them. He made allies on the business side of EL AL, and that was what really made him successful. Other data leaders should take note. They may not need to be as in the trenches as Ido was, but they have to figure out how to speak the language of the business and understand the business in order to be effective.

‍

Nate Nelson: Okay. If that's all, then thanks to Ido for speaking with me. And thank you, Mark, for speaking with me.

‍

Mark Marinelli: Take care Nate.

‍

Nate Nelson: This has been the Data Masters podcast from Tamr. Thanks to everybody listening.