We're still exploring the early steps of how a company can embark into a digital transformation journey. And to discuss this in more details, we are fortunate to have with us today Antoine Gourevitch. Antoine is a senior partner in BCG Paris office. He has run multiple digitization projects and he's actually now our global practice area leader for data and digital platforms. Antoine, welcome. >> Thank you for having me there. >> So Antoine, in the course, we have described how managers should rethink their customer journeys. How they should run that newly designed process. But when we step back and we think of incumbent companies who have a legacy business to run, who need to continue running that while thinking of how they digitize different journeys and different core processes, how should they organize? Where should they start? >> I think that's a very good question. And as we know, it's much easier to create a startup than to run a company that does a huge legacy. I guess the customer journey and starting with the client is probably a very good start. And then you should start with how do I increase my loyalty? How do I increase my conquest and choose really a few not pilots, because I guess for me, the main, I would say trap would be to go into pilots. And in every company today, you have dozens of pilots but bringing absolutely no money. What is very important is to choose a few area where you're going to go at scale, and really get money and the value. So for example, if you decide to go and increase your share of wallets, for example in after sales by mixing data from different sources, then you should deploy that in all countries at once, to really go to tens of millions of value. And I would say companies should begin to go to think big, choose specific areas of value, and definitely avoid what they call the pilots or proof of concept trap. >> And when you talk about scale, what makes it difficult to get to scale? Because you're right, actually a lot of companies just go into the infinite pilot phase. What is difficult in getting things to scale? >> So what is difficult for scale is the fact that when you do a pilot, you're taking data and cleaning the data for example from different sources, I would say in a very artisanal way. And it's extremely difficult to replicate to everywhere else. For example, we did the pilot for one of the automakers in Belgium. And we took the data from the finance departments, after sales departments, from the sales companies. We put them together and so we we're able to increase market share by four points. Now question was, how do you do the same in Brazil, in Russia, in China in US? And if you have not sold before but you know where is the data, what in this scale platform are you going to build to put the data in place, you need to replicate country by country which is an extremely costly process. So you need to do both things, innovation to prove that it can work, and think before running to, how will you industrialize that. And today, it's easy to do it, for example, in interdata space use of cloud. We use cloud-based platform but it's not a magic solution. So you need to go into the legacy systems. But I would say, for me the difference between companies that successfully digitized on its user, as the people were both compromise between innovation and. And it all starts with a data platform. I mean, it's companies like Uber, Amazon, Google, Facebook, they have natively built with digital platforms. So it's kind of easy, and they can organize their own data. For example, Amazon is updating its price 25,000 times a day. And you can do that only if you have a very clean and very, I would say, efficient data platform. If you go into banking, or into energy, or into alternative, nobody has a data platform that can do that. So you have build a data platform, which is a bit contradictory to what people think is that digitization is about changing the front office and building a normal bilapse which is very easy and true that takes a few weeks. But you still have to work on the legacy. Now the other trap is, if you listen to some IT people who are a bit old style, they would tell you, okay, let's do a transformation of the legacy for three years, which is also not the answer. So between three weeks and three years, you have to find the good middle ground where we say, okay, how do we decouple the mobile application to be extremely simple and provocative so that turns legacy to make them all work. And this is for me, how you can industrialize and change. And what is fascinating in this is that when you speak to a CO too there, he will speak with you about technology in that time in like that is something that will never have happened before. I was working in the bank and actually we're doing a project for them. And then we had a CO meeting where the CO said, tell me what options I will not be able to exert if my architecture is going in this direction. So, sets a new type of questions and those are these questions that you can never see you today is when they go to China and meet, for example with Jack Ma of Alibaba, the guy said, I've put all my IT into the cloud. That's what he said, Alibaba Cloud, and then I'm not speaking to any IT people any more. I'm just concentrating on those set of people. So you see, we came back to US or to Europe, asking us a question, should I put all my IT into the cloud and do the same thing? Which of course is a good idea, but it's a bit more complex to put in place. But anyway, I would say the key is to really have in mind the fact that you will need to be much more faster in the way you are delivering your applications to both Internet, mobile, and everything through the multichannels. And the way to deliver it for clients is every months, or every two or three months, but also do that on the solid foundation. And companies will win-win. And I believe, by the way, that companies will win over startups. I think that companies who are able to digitize will be much better than startups. Because they have the data, they have the assets, they have the capabilities, will really bring the compromise and win the game. >> And one core idea in here is, how do you manage to run at two different speeds? I like a larger idea of two-speed IT that you have introduced actually in one MOOC that you guys should check on at some point. How do you manage that? How do manage the two-speed if you have the big legacy that you need to run while creating the new platform? >> Yeah, so two-speed IT is of course needed but should not last for too long. Because of course in the beginning you have, I would say, your legacy which is moving at a certain speed. And we know the speed of the legacy is very difficult. One of my favorite CEOs said, whenever the banking industry said, whenever I open the hood to change my legacy system, it costs me 1 million. And it just takes three months to understand the impact before I can do any coding. So with this type of legacy, you have a speed which is more in a year time frame. Of course when you are building a digital, that platform, to be able for example to increase market share by investing better customer moments of truth, for example, if you want to understand if your customer likes a product or whether he buys a product and so you want to adjust that, you should do it in weeks or months. So you have two speeds, and so you need to adjust. But the key is to decompose legacy to the frontal phase, so that at least you can go at the same speed. And then suddenly you can start deducing. So the way to do it is to take out the data that's in the legacy, put it into a data platform, which going to be a new skeletal platform, could be in the cloud, could be not in the cloud. And then I would say the modern application will tap into the data, and will make the legacy work only on the traditional functions. So by doing that, you can have an evolution of over, let's say, five, ten years to simplify the legacy. But you will be building your digital platform better than the data platform which enable you to and compromise. So I would say modern companies, very old company, use this modern approach, which is working extremely well. And it's not rocket science, but the key to that is to have the capabilities to be able to do it. So you need to have software engineers. You need to have architects who understand deeply the new technology and deeply the old legacy architecture. And so we're about to do that. It's extremely difficult today when you have lost the architecture, I would say, capabilities, which happen in some places especially in retail goods, in energy, a bit less so in banking. Because in banking you have less software package. >> All right, that's a fascinating talk, Antoine. Thank you very much for taking the time to share with us all that knowledge. >> Thank you very much.