Name: Metadata Masterclass: Scaling Across Global Enterprises
Uploaded: 2025-10-29T19:00:03.880Z
Duration: 30 min 56 s
Description: Metadata Masterclass: Scaling Across Global Enterprises

Transcript for "Metadata Masterclass: Scaling Across Global Enterprises": Hello, and welcome to Context. I'm Stephen Goldbaum, field CTO for Financial Services at DataHub. I'm happy to welcome Mikhail Matsou, a seasoned veteran of the data industry. Mikhail, perhaps you'd like to introduce yourself? Hi, Stephen. Hi, everybody. I'm very glad to be here. So I'm bringing more than fifteen years of experience on data management. I have been doing data strategy, data management, first in consulting and now in the financial industry for the last ten twelve years, actually. And over the year, I have developed my expertise around four pillars, strategy, innovation, data, and apply all that applied in education sector as my social side, but professionally in the financial services. So I'm very glad to be here and to discuss Meta Data AI in general in this transformative time where we are seeing how AI is shaking the industry. Great. Let's dig in. So the financial industry is a pretty conservative industry, and you've managed to affect substantial change within that industry. What strategies have you implored in order to do that? Yes. Thanks, Stephane, for this question. I think this one call, made me think about success element, right, recipe for success when you want to change something in this industry. And I have three things really that come to mind. First one is leadership. Second one is sponsorship. And then third one is value. Let's talk about the first one, leadership. Leadership, when you want to transform in our industry, from my experience, come from the followership. It's about identifying very early followers that understand your vision and then find a way as a leader to follow your first followers. Right? And to me, that's very important. So that's become a collective journey more than, a single hero trying to transform or change the the, the the organization. And on the sponsorship side, this is essential to have a senior sponsor, someone which has a lot of authority very high in the organization. And it's also essential to not have just one, but multiple. And in my experience, the way I've done that, usually, I try to have chief data officer on my side because I have had, those head of data role in in different division in my organization, you know, large banks, financial in the financial services. They have business that have almost their own chief data office. And what I found is, as a head of data or head of information architecture for a division, it's always good to strike this partnership, this allyship with the CDO. And, and for them, it helped them push the agenda. And, for me, it helped me find resource and also support outside the the organization. So this type of sponsorship are key on top of the sponsorship obviously in your organization, but also sponsorship with from all the business around. The more you build multiple sponsorship, the likely your your your project is to be successful. And the last one is probably the most important. We have an industry which is very much value driven, and it's not a a place where we come for dream. Right? People are there to make money. Right? It's things need to be successful. It has to bring results very quickly. So it is always important to scan the horizon and find an opportunity, a big problem that everybody is aware of, usually in the form of project. And, I found many of those opportunity. One of them was, when in in one of my, for one of my my employers, we needed to go through a transformation of our clearing business. And another one was when the organization was was doing front to back transformation on his data. And, and another example, was when we had a regulatory requirement that we had to push with a specific hard timeline to hit. So this kind of opportunity that is important to find and then demonstrate that you can bring value to resolve the problem, and through that, bring the change and the transformation. So in summary, really free point there, the leadership in term of followership, the sponsorship, and the value. Great. Yes. So that's a very top down approach and certainly there's a that's a major part of the strategy. There's an interesting other side of that as well as that on the day to day lives of the people who are actually going to be touching the systems that you're responsible for, there needs to be some value there. Have you developed a strategy for dealing with that as well? There is no strategy, that will touch ground if it's not affected by the people that it will impact. Right? And, that's why, very early on, the focus on the followership is very important. Like, get the people that are close to the business and, identify ally that, believe in the transformation that you are you want to do. Don't try to broadcast all the time. You need to develop those one to one relationship, and then listen. Right? Just listening and flexing really the strategy to adapt, right, is the best way to get to something which is really relevant for for the business. It's very, hard, even after many years doing data to understand just on our side, data people, really what's resonated for the business or the very specific best way to address the problem that they are facing. It's only for the partnership. And, and to me, to to respond directly to our question is is is about taking the backstage very quickly. Have the vision, convince people that you can bring value, but then take the backstage so then the real people that will be affected by the transformation that you are bringing, really and, really drive the show. One of the way that I I I do that is by very quickly have a communication plan in my project. Very early on, there will be a communication going very regularly. If it's not weekly or biweekly, it will be at least monthly. And this communication is not about talking about the good work that we are doing. It's about talking about the good work that all the people are doing, contributing to enable the strategy. And having those regular communication in place, talking about people and their contribution, celebrating the progress that we are making, really shift the dial and, and progressively make the work just easier and easier and faster and faster. Great. Yeah. So interesting. So it's a very cultural approach to, to change, which obviously is very effective. Let's shift over to to metadata. So metadata, I think, in in watching your your past, speaking, you've mentioned that metadata is an important part of data change. Can you elaborate on that a little bit? Yeah. I do think that, there is, no, data architecture without data. There is no more than data architecture without data, and this is just reinforced by, AI coming. I will give an example maybe to elaborate on that. And I will talk about the clearing transformation program. So first, what is clearing? You know, business, they exchange asset, right, from your end to to the other, and those asset need to exist, and they need to move. And, so when you do a payment, it has to be settled. Like, the money need to arrive. And sometimes it takes time, in business like the River City where there is an exchange, you need to inform the exchange that they you are buying this asset. The exchange need to free the asset so then you can you can get it on your on your on your portfolio. So this is clearing. It's really central and very important for for the business. And, I was fortunate to be part of a program that needed to transform the the the whole business process around the clearing of a major bank after, using the same platform for seventeen years. And there were really no data intellectual property. So this organization was talking about its data in the language with the the world and the structure of the platform that they were using for the for for clearing. And when this came to move moving to another application, there were IT challenge, intellectual property challenge, legal challenge of exposing, some aspect that the the the the existing platform were considering as secret, right, to their competitors. And the what method that I gave to this organization in this project was really to accelerate that with clarity. So immediately when we started, we took a very risky pay and very bold decision was to to to spend the first six months of the project, putting the the whole energy of all the business analysts, project manager, data modeler, data architect, data engineer, focusing on building a canonical data model and everything around data and get a strong clarity. And all those things around data, it was real were really metadata metadata, describing the content of the data, but also describing the context of this data. And what has happened after this bold step was that the rest of the project become very smooth. Once we had the foundation of our metadata architecture, we cut the time of the delivery of the project almost by two. So the project that was expected to last thirty six months compared to, I mean, what we're seeing from other organization of similar side, we completed that in twenty two months, and we completed that with a lot of certainty, a lot of clarity on the on on every decision that we we were taking. So from that point, it's really demonstrated, to me that you can't do a proper data strategy and particularly data architecture if you don't focus on the metadata aspect. And today, it's even more relevant, this this approach when you see AI coming and when you're talking about data mesh strategy or when we talk about data fabric. The foundation of all that to happen is really strong and and and clean metadata. I mean, we we are not talking about transformation in in small startup. The financial services based on that the heart is working in in in large organization and, doing program that will impact multiple businesses. And in in in this particular case, we had 36 businesses that we are using the clearing the central clearing platform. Right? And, and this metadata work has driven a complete business process transformation in infrastructure simplification. So so that's an impact even beyond, data itself. Yeah. So interesting. So the lesson here is that a project that's sharing data or a project that's migrating data is no it's no longer just about the data, it's also about the information about that data and that needs to be a first class part of that project in order for it to be successful. And I think that realization has really come come to its own in the last few years. And certainly, that's a component of AI. So let's let's dive into AI a little bit. The our metadata is is valuable in two ways with AI. I think it's it's in terms of giving information to AI so that it's able to make intelligent and and, informed decisions. And then it's also useful in terms of understanding that metadata as you've just described. So in maybe in a legacy situation where you haven't documented all the information about the data, a lot of times that can be an accelerator as well. Do you have any thoughts on that? Yes. I will take it from I've been applying some first principle of thinking here. So the real problem we have for data bring since data is a major topic in a large organization is trust, in fact. Trusting the data that you have so that you can use it confidently has been a problem for long and remain a problem. This was a problem since when we are talking data warehouse, since when we are doing business intelligence, since when we start doing big data, data lake, etcetera, till today. And AI I mean, that that's why metadata is even more essential because it's resolving the trust problem by bringing transparency. Transparency, right, creates trust. Right? And that's something that the financial industry has learned very quickly and for long. The more the market is transparent, the more we know the the the the transaction that are happening, the more the market is behaving as a proper market. Right? The the the whole point of the market is to bring transparency. And data need that. Data need this thinking market thinking. And, that's where AI can play a major role, right, by coming into our work and, help provide this observability, help really automate data governance at scale so we can provide the the largest transparency as possible on data. We can enforce control. We can identify problem even before they they they they they become problem, I mean, as data is is is moving through its life cycle. And AI, doing that, we also benefit from that because the more we trust our data, the more we we are confident on the compliance of the data and on the performance of the data, the more we can automatically enforce policies on and control on this data, the more we'll be able to expose this data to AI. So we'll bring AI in the organization and give AI data. And using the same metadata, AI will be able to navigate the data landscape, understand what can be used, what can't be used, how every data can be used, and, and create this, this this virtual I mean, this, circle virtual circle, right, whereby AI is helping data to to to be of higher quality, and then data of higher quality and higher transparency also helping AI to to be more used and more impactful for for for for the industry. Mikaela, you've you've talked about strategy and cultural strategy. Another strategy that I think is is picking up a lot recently is the discussion of ontologies. Do you have any thoughts on that? Oh, you can vary directly on the on the on the key topic of the moment, and I'm very passionate about this one. Look. If we I mean, it's absolutely essential to me that we shift into a world where AI can navigate the data landscape. Right? And, we need to enable that. That's where our focus need to be. And I have not seen anything which is better than an ontology to describe what are the elements of the data landscape with all the property that represent them, but also describe in a rich way, the relationship between the streams. And when I say in a rich way, it's not only rich, it's also semantic, which is the way you, human being, we we we exchange, right, and we talk. Right? And, to me, ontology today, would need to be revived and bring in some stage in data all data conversation or data strategy. In the past, when ontology came, it was a challenge. Right? Because you need to put a lot of discipline building them. It's take a lot of data. It can feel like verbose. Right? But we have now tools that can manage this part, this complexity, this discipline, and the structure part with AI, and it's really the opportunity for us to to use the tool, the ontology, to describe our data landscape in an effective way so, AI can read that. And, the good thing is that all LLM are very comfortable reading ontology today. So they can read those ontology and understand what is there, and not only understand what is there, but also understand the constraint. Right? So what we don't want to happen never happen. Right? Because the ontology will have this context and this formation embedded. And, now if I bring that into the topic of strategy in general, I think about strategy with three layers. Right? The first layer is about resources that you develop and unique resources for your organization. And that's connected to the real framework, which is well known in in strategy. It is you need to to develop valuable assets. You need to develop rare asset that are rare and asset that are embedded in your your organization and asset that are difficult to innovate to to imitate. Sorry. So first, that's a very important element to think about, and ontology is good for that because every organization will have a different ontology. And that's the first resource on top of all the resources as part of your strategy. And the second element of strategy is the the the the thinking of normal not really what you want to do, which what you focus on, but what we don't want to do. Right? And be very clear about that and, and vision need to come back into the data conversation. It has been something difficult to to to to, I mean, to to sustain in our in our in our day to day work, But I think we we we it's time now to bring that in because the easy things, AI will be able to do that for, for us to it's time to to have both visions. Right? And this vision need to give us a focus of, what we need to do what not to do. Right? Now talking about what not to do is always difficult, right, because there are always things that come up, right, in our organization that we need to focus on. And I use a small trick for managing those. I assess each of those opportunity in the lens of, can I use the resolution of this problem as another way to reinforce my strategy, to be able to get me closer to the vision? And and that helps really filter what is noise from what is signal. Right? And then the the last thing is, for us, for any strategy, is, the strategic bets. Right? You have to step back and scan your horizon and see what are the strategic bets that, that, will make a difference for your organization in the next five to ten years. And to me, in my work today, I see three things. First, AI. Right? The elephant in the room. Strategic, but as technology. Second is user experience. You some we neglected that, but user experience is absolutely essential for helping our organization and people to think a little bit more data driven. So the way we expose it to them in the data marketplace, the way we describe this data, the way really we manage all the interaction of people with the data has to have a a a user experience thinking, to that. And finally, last one is, obviously, platform based, and, that's where I would like to finish on, platform like DataHub. Right? You need to pick a platform. Right? GCP, I would list all the the the the the competitor maybe or or data hub, but, it's a very important part of the equation. Identify a platform that you pay on and then work with this this partner to deliver something really transformative. Alright. So, yes, you started with Ontology, and I linked that really at the backlog of of the strategy, and I wanted to lay it out really as a free element just to, to spark some conversation for people that we further this this discussion. Yeah. Excellent insights. And we also came full circle with the, with the strategy as well, where we came top down and you were talking about getting your stakeholders. You ended up bottom up in the user experience and how that is also a major component of the success of a of a data strategy. So as usual or as always, it's great to talk to you, Mikhail. You have incredible insights, and it's been great to listen to you. Thank you for joining us. Thank you very much, Stefan. It's my pleasure and I have enjoyed the conversation.