Why Enterprises Should Treat Data Like a Living Organism

[Maciej Kłodaś: MKLO]

Hi everyone, my name is Maciej, I’m the leader of analytics experience competency group at C&F and this is C&F Talks, a place where experts meet to discuss trends and challenges from the perspective of an IT partner. Please remember to subscribe and like our channel because your feedback is very valuable to us and keeps us going. So my guests today, and this is a special episode because we have two guests in our studio, Simon Winnicki, Maciej Klinewski.

Guys, can you share who you are, what you do, what you specialize?

[Szymon (Simon) Winnicki: SW]

My name is Simon, I’m at C&F for around seven years, I’m a senior delivery leader at our pharma commercial business unit and we specialize in delivering different types of data solutions, data products and data applications, generally all things data for pharma commercial companies and together I have Maciej.

[Maciej Kliniewski: MKLI]

Hi everyone, so I’m Maciej, I’m together with Simon, the team in the business unit of pharma commercial, working as delivery lead in C&F since four years. So yeah, working with our clients to identify their needs and find a way to deliver what they need.

[MKLO]

Okay guys, so it’s not a surprise we’ll be talking about data today and I’ve heard from you a very nice comparison, data as a living organism and it’s very catchy. So tell me why such comparison, how you perceive data and the all, you know, environment of data at our clients?

[SW]

So I think this kind of ties back to what I said before. So all things data, because when me and Maciej were thinking what we would like to talk about and share our knowledge of, we couldn’t really pin down one specific area on topic because I think that we specialize in the general topic of data management and then projects touch upon different areas. So I think that we are not that specialized in one specific area, but more the value that we can bring is based on the fact that we know about all of those different small bits and pieces that connect together.

And then I think that to actually say that the company is data fluent, you need to combine all of those different pieces together and they all need to fit together so that this living organism of data can start acting, reacting and interacting.

[MKLO]

So we have dependencies. I will later ask you what is the liver in this organism, what are lungs and everything else, right? So tell me from your perspective, what is the as-is state right now, how you perceive, you know, the overall market state in terms of data?

[MKLI]

So in terms of market data as-is state, so until now we observed like everything was the, is business requirement driven. So there is a need and we are finding a way to actually satisfy the requirement without like broad picture, without thinking about how it, what consequences it can have in long term. So yeah, as in the living organism, as in how we perceive like our health, let’s say, so fixing or like if you imagine you have a headache, right, you’re taking painkillers, it’s a fix, it works, you feel better, right?

But thinking about health, you need to think long term, you need to change your habits, you need to change the mindset and that’s, yeah, long term we need to build healthy data, healthy data systems. So it’s right now even more important considering the market is driving towards like AI readiness, more semantics, it’s even more important to actually find healthy data across all the entire organization.

[MKLO]

Okay, so as in health and nutrition trends change over the years, we can call you the data doctors right now. So tell me, how do you diagnose this ill organism being the data environment at our clients? What are the challenges you see, how to spot them and then how to, you know, prescribe different drugs to make it better or treatments?

[SW]

So I think first, as we said before, like our bread and butter is pharma commercial. So we also need to maybe define the fact that when we look at pharma organizations, they are mostly not, no, I don’t want to say they are not data driven, but that their revenue and their core of business doesn’t come from data. And before we had all of the modern data systems that we have now, those companies were still able to operate.

And because of that, I think that often in the pharma organizations or in general life science organizations, there, it lacks data offices, for example, because it’s more distributed towards the domains and then domains have a certain level of freedom on how they want to approach data. And this obviously brings some benefits because then they can have, in a lot of cases, they can implement things faster and implement things. Flexibility, but then flexibility also takes away the standardization and the fact that if you don’t have a data office at the company, it’s sometimes hard to persuade the stakeholders that data is actually your underlying goal.

And I think one of the things that tends to happen, or maybe right now it’s changing, it’s happening a little bit less, is that the data was a by-product of a dashboard or a by-product of an application. So when we had a project or when we had an initiative for a program or even a company-wide program, in most cases, data was not what they wanted to implement or do. No one actually, maybe not cared, but data wasn’t the most important factor, but the most important factor was the outcome of it.

So a dashboard where someone can see exactly. And obviously at the end of the day, that’s what matters because that’s what business will see. But unless you think about your data systems and invest and think long-term, then you’re going to just do point solutions.

And you did this dashboard, you do one solution, another dashboard, another solution, another dashboard, another solution. But if you look at all of those initiatives holistically and you start identifying, for example, the points that are repeating itself, then maybe it’s better to invest a little bit more, but once, and I don’t know, set up your master data systems for doctors, so then all of those dashboards can benefit because they will use the underlying data system, which obviously means that first you need to invest more, but in the long term, you are going to invest less and less and less. Plus you’re going to get the, let’s say, hidden benefit of data quality.

[MKLO]

Okay. Let me, sorry, let me take a step back here. So you said the data office, can you shed some light on the concept of data office then?

[SW]

Yeah. So, I mean, like, if you look at companies, for example, like Netflix, they would have a chief data officer or maybe right now, then I would be slightly different chief data excellence officer. But nevertheless, there is someone on the business side and there is someone with high level of leadership who, whose primary job is to care about the data, not care about the applications, not care about the analytics that come with the data, but care about the data.

They can set the standards that can drive the technology. They can set company-wide standards. Okay.

Right now we are going to use Snowflake. We are going to use Data Mesh or we are going to use Data Vault or another approach. And thanks to setting those standards, you can also foster the data literacy, data understanding.

Those are the initiatives that are in general quite expensive because you need to train a lot of people. You need to think about which areas you want to train the people in. And it’s not a one-time thing.

It’s a process. You need to have like a product approach. You need to think long-term, but if you have a project-based approach, so when you don’t have a data office, it’s much harder because the budgets on the digital side are smaller and they are project-based because they are paid by the business.

So you cannot think, okay, what am I going to do in five years? Because you only know, I’m going to have the budget. Exactly.

I have the budget for 2023. Maybe someone there is a company-wide initiative. So you already know, okay, I’m going to have the budget for two or three years so we can a little bit plan ahead.

And you have a lot of overlaps from project to project. Exactly. You have a lot of overlaps, but then without this like long-term strategic thinking, you cannot say, okay, so right now I’m going to go into three-year data governance program because your governance program would be, okay, I have this year, I need to implement one, two, three, four.

So you just tick the box, tick the box, tick the box. Okay. We implemented Colibra.

But do you have the process behind it? Do you have the people? Do people know what they are supposed to do?

Did you set up any new positions whose roles are data governance, or did you simply just throw into someone’s already existing role and the batch of the responsibilities? Okay, you are now the data steward for this and that area. And then obviously people are not going to be happy.

In general, probably if you ask among most of the organizations and people who data governance, they’re going to be like, I don’t want to do that. And I think that’s because we start with technology and then we go to the process, not the other way around. But at the end of the day, does the technology matter that much?

Obviously different technologies have different capabilities, but if you have a process, you can have technology with less capabilities and still have much better result than if you use the most advanced technology, but you don’t have the process and you just tick the boxes with implementing it.

[MKLO]

And now in the era of AI, the data governance and data quality is, yeah, it’s very important, right?

[MKLI]

So maybe to add on to what you said, Simon, it’s also a challenge from the perspective, how complex it is, because we are talking here about global organizations. So the business is run locally. It runs differently in Japan, in UK, in Germany, but we think of from like global perspective.

So like managing everything from like data office from global perspective is a very big challenge for organizations to care about it, especially from AI perspective, building data solutions that are truly global. That’s something, yeah, actually making this much more effort to build like a healthy, healthy, healthy process and healthy situation.

[MKLO]

The difference between local markets and global markets vary, right? From semantics to, you know, cultural differences to understanding data differently. And we know that we work on different projects and they’re always the same challenge.

Do you have any, you know, use case from your field when you’ve seen such challenges when approaching such projects? Yeah.

[MKLI]

So actually last year we ran a project with our client regarding like, yeah, ROI analysis. So yeah, everyone cares about like return of investment, right? So what actions you take, how they actually evolve into real like benefit for the organization.

So we, as was state was like the ROI analysis was run locally, as I mentioned. Yeah. Every market had own roles or like business perspective on how they actually, like how activity of the pharma business corresponds to real sales.

And it was like driven by some local insight group of people. They built reports locally with, yeah, excels, like drag drop solutions. Yeah.

Everything was suited for their local purpose. And we observed they’re actually like doing the same thing across different markets, but it’s all based local. So the idea was to have a standard centralized global solution, which with the same like data structure, with the same business role set up, just managed by a configuration of some data, just to enable some data, just to implant such configuration for a business role.

And thanks to that, we were able to scale it across more markets. And even like within teams, they were able to run the analysis from market to market. So someone working on Germany could actually involve and validate and like compare results with France and with Japan or with other market.

And now it’s evolving. More markets are being enabled thanks to the fact that at the beginning, it was, there was investment to build global general standard approach. And now it’s being followed.

Yeah. So people change across organization who was involved last year, changed the role and the knowledge is not left with the person, but is with the system.

[SW]

Yeah.

[MKLI]

So now it can be, it can be enhanced, developed and used and it plays the purpose. Okay.

[MKLO]

So again, coming back to the challenges, how do you introduce such global solution to local markets? How do you secure adoption of such processes? Do you see some, you know, challenges and friction when big companies are trying to cascade such global solutions to markets that were used to using their own small ecosystems?

[MKLI]

Sure. So for sure it’s a challenge. Yeah.

Because we’re working on commercial projects. Yeah. So there’s always limited budget and tight timelines, right.

But we need to like deliver solution with good quality and thinking about multiple aspects of that. I think like, yeah, I’m not sure Simon, if you agree, but the first thing is to assess what’s actually we need to do, what has to be done, what they need and how they actually currently.

[SW]

Um, I would call it differently. I would say the first step is to partner. Cause if you want to assess, you need someone to assess, assess that with, you need to have a source of this knowledge.

So I would say the first thing is to partner is to find some, like probably the first thing that you would do before even implementing such, such initiative is that you need to find sponsorship and leadership on the business side. And you need to find someone who’s a little bit higher up the organization so that they do have the budget to even implement such an… And their mental buy-in to do it with you.

Exactly. Exactly. But like, also like, you know, simple things you need to have this amount of money.

Cause as we said at the beginning, his investment is higher, but then like it, it then generally like it decreases, decreases, decreases over time. So let’s say if you have 10 markets that you wanted to implement this solution, if you do it 10 times locally, then definitely you’re going to pay more than if you do it once, but then you need to have someone that would either pay for that or collect money from all of those markets that would do it locally before, because the first market will probably cost as much as for other, for other ones run locally. But then the second will be the twice as much. The third will be the same as the locally, but then the fourth will be half, the sixth will

[MKLO]

So this is again, the challenge, or you need to find a sponsor who will be open-minded enough to pick up such initiative.

[SW]

And then you need to find someone from the local market that you can partner with again. Cause what you said about adoption, like if you just come to a local market and say, all right, we’ve got a new tool for you, use it. Here is the cookbook.

We don’t like it. I know no one’s going to like it. Everyone’s going to be like, what they are like, they are taking our jobs.

Like is, you know, am I going to, am I going to be left out of the organization? Cause maybe right now they are going to do it at the global level. And I think the way to do it is to partner, come to them, say guys, like, you know it best.

We are not coming to you to take, take away your job or to say that we know how to do better your local analysis. We know how to standardize. We know how to implement things in the higher level.

And we just want to give you tools. We just want to empower you to do your analysis better, do your analysis in a standard way. Plus don’t forget that by doing this, you’re also going to benefit global stakeholders.

Cause then if you run 10 analyzes locally and try to compare it, 80% of cases, it’s going to be nothing because the systems won’t connect. You know, someone’s going to use local names of the product. Someone’s going to use this idea of a doctor or something else.

And then if you do it from the global level, you also bring consistency. And if you bring consistency, this allows someone at the global level to look at those 10 markets and maybe see some trends which are regional or just in general have an overview of what’s being done. And I think that’s another interest, like hidden, not hidden, but maybe a little bit hidden benefit of doing global solutions is that you bring consistent reporting to a higher level.

So maybe you can also even make your higher management aware of some things that are being done because before they were being done at the local level, exactly. So they were too fragmented for someone at such a high level to, to be even able to look at it.

[MKLO]

Okay. So you need to secure this, um, I would say lack of resistance from the local markets to say that them to them, that this will go, you’re going to benefit from that. We won’t take away.

[SW]

You’re going to benefit. And also don’t, don’t enforce too many things. Of course they are like, they are some things that you need to enforce.

You need to enforce using the master systems. You need to enforce not using Excel’s to map something because you need to have something which is enterprise grade, which is not going to break every two weeks. Cause then, then it’s probably every, if something breaks and breaks once, okay, maybe people won’t lose that much trust.

But if it breaks three times, everyone’s going to go, all right, I’m just going to go to my local consulting company. They will do my analysis. I will have my results as I always wanted.

[MKLO]

Okay. But let’s start from the beginning. Imagine that I’m the, you know, your client and I have this, I have this open mind to do the change within my company.

So how, how would you start this, uh, this exercise with me? Let’s say I’m the sponsor. Okay.

You already found the sponsor.

[MKLI]

Yeah. So yeah. As Simon mentioned, so let’s, let’s imagine it’s like this, the sponsor we are already working with.

Yeah. So let’s say that the partnership we, we already built in our previous projects. Yeah.

So yeah. First we need to, yeah. Um, find what’s the current state.

Yeah. What’s the current state of situation. Let’s imagine we can, we can, um, assess like there are like five markets, uh, doing the reporting on their side.

We need to understand who is responsible for, for, for these markets, uh, who acts there, who knows the answers. Uh, if we ask like, where do you gather the data? If it’s structured, if it’s supported, if it’s a regularly delivered, yeah.

Or it’s like on ad hoc basis, uh, we need to find these answers and we need to first have a, have someone, uh, to work with, yeah. To cooperate. Um, and then based on that, we can actually find and decide on the priority prioritization here because some market can be like very straightforward.

The data is ready. The data is clean. The data is supported.

The data is coming live on a daily basis. This is a state where we can already start work and build a POC or like first release, uh, of our like data product. So, but first we need to understand what’s the, what’s the current situation.

And based on that, we can plan our, our delivery because also like knowing at the beginning what we are working with, we are making decisions. Yeah. On how to structure the product, what business roles, what configuration we need to, we need to implement, and it needs to like support future implementations as well.

So the knowledge at the beginning is, is super important to, to find where to place the entire solution.

[MKLO]

Okay. Uh, when we’ve been discussing prior to recording this session, you’ve been mentioning a, um, a kind of a rule book, uh, called DM BOC. It’s kind of a similar to PM BOC where you have all these rules and elements of, of project management.

And this is the same thing with data, right? Can you, you know, share some, some details around that?

[SW]

Yeah. So when we were talking about that, we mentioned before and like trying to, let’s say, identify how do you want to structure the conversation and what to go with. We were talking like we knew that the question is going to come.

Okay. So how to approach this? And I don’t think there is like a one definite cookbook.

You don’t have a cookbook for that because all of the organizations are so different. It’s really dependent on the history, on the fact of having a data office or not having a data office on the fact whether IT is global or it’s more decentralized and like to what extent you allow, let’s say your different domains to have their own technologies, their own approaches and so on and so forth. But it’s definitely good to have any template to start, start with.

I don’t think it’s a cookbook, but I think it’s a starting point that you can use, you know, to, to at least know in which areas you can assess your situation. So yeah, so DMbox, this is the data management body of knowledge and it identifies, I think in the second edition, if I remember correctly, there are 11 data pillars that they identified. I’m not going to read them through.

If someone wants to, let’s, you can check that quickly in the internet. But when we looked at that, we saw a few maybe missing pieces that we saw. Maybe it’s also the fact that it’s right now in process because we are also in times which are being changed.

So we thought that first thing, which is kind of missing from there is the AI readiness for the data topic, which right now is probably, you know, at the mind of every data person exactly around the world and definitely, you know, in the goals being put there by the business. And then I think that another layer which was missing, there was the organizational layer. So we kind of imagined, like tried to, let’s say, categorize those pillars and thought about an anecdote that let’s say we’re going to try to share right now.

Let’s see if it resonates. So we have the first group of our data pillars, which are our data foundations. Let’s say this is a skeleton of an organism.

So this is the, this is our underlying technology. This is the approach that we have for data security and compliance. This is the general data architecture and system architecture that we have in the company.

Like these are the foundations that we can build upon. And this, if we are talking about our anecdote, the data as a living organism, that would be our skeleton. Then we can move to the second group of our pillars, which would be the trust and excellence, we call it.

This would take envision things which also are dependent on the technology. But I think that you really need to excel in it to be able to say that you implemented or that you are doing well in that pillar. So things like data governance, as I said before, things like data quality, again, these are things where you can have the best tooling, but unless you have approach, business understanding, and actually like you can tie those things back to the business goals, it’s going to go bad.

It’s either not going to be used by anyone or it’s going to be used by everyone. And then this third, let’s say, group of pillars that we identified and that we think we’re missing from the DM book is the organizational and trust layer, let’s say. And the organizational and trust layer would mean things like data literacy that we mentioned before, the sponsorship, for example, having a data office, in general, like the approach to the organization, to the data match, like the domain driven ownership.

And, oh, I missed one point. So our trust and excellence layer is our brain. And then if we have our skeleton and our brain, our nervous system starts to work, and we can use our brain to move our body.

And right now the third layer, the trust and excellence. Where’s the heart? That’s our heart.

And then when you add the heart to the mix, your blood starts flowing, your circulatory system works, and the whole organism can interreact. So that’s where I think we can go back to this like data as a living organism that we have, let’s say, those three pillars. Obviously, our body is much more complex, the same as data, but we need to start somewhere.

And only if you really focus on all three of those pillars, your body is actually going to work. Without this, well, maybe you’re going to be able to move your body, but your heart’s not going to be pumping. So you’re not going to go for too long.

And then like the last thought that I want to add here is that we need to also look differently at like the organizational approach versus the project approach. So we said about all of those pillars. And I’m not saying that this is a starting point for every project when we want to do a project that we need to have all of those, you know, all of those pillars implemented, because we also need to be realistic, the same as a doctor, when you go to him, and you have two different illnesses, he also needs to choose, okay, which I’m going to treat first, maybe this one is fatal, and the other one is not. So let’s start from the fatal one, you need to be. So in case of data, you also need to act as a doctor, or we need to act as doctors, as you mentioned before, and say, okay, we can see that there are symptoms here, here and there.

But realistically, having the budget, having the project, having the timeline, what Maciej mentioned, we have commercial projects, they are often fast paced, and you need to deliver something quickly. So let’s at least select some of those pillars and try to implement them. Like, for example, the project that Maciej mentioned, let’s go with the standardization, let’s go with having the standard approach for quality, for analytics, we’ve have, we already have the sponsorship, because we found someone who wants to do a bigger program.

Let’s do that. Let’s show people the value, let people, let’s educate the people. And when we do that, maybe we can go and start implementing data governance for this process.

But if we start from data governance, the patient’s going to die. Exactly.

[MKLI]

And, yeah, it’s, it’s sorry to interrupt you, Simon, but it’s also like with a doctor, right? If you know your doctor, if doctor knows you already, and you come with new symptoms, the doctor will act differently with you than with other patient that’s completely new. That’s smart.

[MKLO]

Okay.

[MKLI]

And also, if you’re like, like 30, 40 years old, the, the action will be different than to the patient who’s like the 60 year old. Yeah. So I’m saying that we need to like, use the pillars, be aware about them, but how we use them for our project at the point in time with our clients, it’s, it’s up to us.

We need to like define what’s the most important part at the time where we are in the project life cycle to actually, yeah, build the long term healthy organism.

[MKLO]

This is what I wanted to ask you. So depending on whether you are starting with a clean slate or a mature organism, you have different challenges and different risks. So what risks do you see?

I assume that working with, starting fresh is a lot easier. You can implement standards as you wish, but working with a company which has established environment, and you need to change that because we, you see those symptoms, it’s more complex, right? So what are the biggest risks you see, whether it’s budget or buy-in or whatever?

[MKLI]

Yeah. Maybe let’s, let’s come a bit back. Yeah.

So what, what we consider as a sick or like ill organism in data. Yeah. So we have, for example, like data in Silas, we have lack of ownership, what, what we already told.

So lack of automation in tests, in quality checks. Yeah. The inconsistency with the semantics or like definitions.

So risk in the projects, in data systems, which we are working with are to not to come to a state where we have a sick, sick organism. Yeah. So prevent, prevent.

Yeah. So thanks for that. We need to prevent, prevent sick organism here.

So the risk is, as you mentioned, magic. So there is, there is limited cost. Yeah.

There is a, there is limited time. And we need to think about that. It will be used.

It will, it will satisfy the need, not like immediately, but it will, it will be used across the, across the organization. So especially with, with the AI readiness, we need to make sure like the meaning is, is the same. Yeah.

So the risk with that, If we deliver something very quickly, we don’t see the broader picture of what actually the type means, what is the description, yeah, if the description of data is something different, the meaning is different across different areas, different markets, different data systems, then it will not be AI ready, yeah, there will be a risk that AI will learn from the data. To different patterns.

Different patterns, yeah, so it’s important to think about at the beginning and to address that with our actions, to invest the time to understand and to have the same meaning, so yeah, that could be an example.

[SW]

Okay, if I may add, I think like a lot of the points that Maciej mentioned come from the fact that we always look short term, we always want to fix something short term and obviously often we do need to fix something short term because something what’s important broke, all right, let’s fix it, but the problem is if you fix it and leave it as is, and if you do it once, that’s fine, as with the organism, if I don’t know, you have a headache and you take a painkiller, that’s fine, you took it once, it worked, you feel well, headache gone, perfect, but if you do it second time, third time, fourth time, fifth time, sixth time, and if you continue doing that over time, finally, either the painkiller is going to have the bad effect on you, or finally, you’re going to build up the tolerance against the painkiller, so we’re going to have to take more and more.

[MKLO]

Or the root cause is much more dangerous. Exactly, exactly.

[SW]

The surgery or like transplantology, right? That’s what Maciej said, everyone wants to have right now AI-ready data products, and that’s great, but if we just say, okay, I need to have AI-ready data products, you look at the data products and say, oh, I want to have comments, I want to have tags, I want to have semantic models built either in Snowflake or in Salesforce or in other tool, if you don’t have the foundations, and in a lot of cases, you don’t, for example, you don’t have business terms defined, you don’t have company-wide semantics defined, then it’s going to be much harder to implement it, and there is a very high risk that you’re going to implement it as a point solution, you’re going to get a consulting agency, or you’re going to get even our company, you’re going to say, I need these five tables to be AI-ready because I want to build an AI model on that, and someone’s going to spend a lot of time doing that, but they are not going to address everything that was before, they are not going to address all of the data layers that were before, even though the work that they did could also address this, and if you want an AI-ready data product here, at the, let’s say, fourth or fifth layer of the data product, you should start at the first layer, the source layer and go up, you shouldn’t start here, because you can’t really go down, it’s much harder to go down than to go up.

[MKLO]

It’s like you will be an office worker and you go to the doctors or trainers saying, lad, within two weeks, I need to run a marathon, and you need to prepare me for that, and you might do it, or you might die at the end or halfway, so an apple a day keeps the doctor away, right? So how to keep, or what are the best practices to keep this data organism healthy? Not using in-point solutions, but looking holistic, what are the actions, what are the processes to implement to keep it healthy and running?

A very difficult question.

[MKLI]

For example, we need to define acceptance criteria, for example. You need to have quality checks against these acceptance criteria, and it has to go automatically, because we don’t know how the data is changing in this foundation raw data layer, but we need to make sure this displays the purpose we build upon. So for that, we need automated tests.

So this is an example. What else is that developing the system, we need to still retain the foundation on what it was built. So providing the same definitions, the same semantics, the same structures, not going like shortcuts, but building with the foundation.

[SW]

And I think that kind of ties back to the coffee chat we had before about reusability with one of our board members, and I think you need to ensure that people can reuse what you already built, because that, again, goes to those point solutions. So in the world of data, for example, way too many times I’ve seen that there are three different projects across four years, and they, for example, integrated the same underlying data source. So you need to have a standard process.

I’m not saying that you need to define each and every step on how someone integrates the data to the company, but it’s good if at least you define, for example, that if something goes to an enterprise level data, like let’s say the foundational, like the most trusted layer, that there is any type of body or governance that takes a look at what’s going to be done, how is it going to be done, and that this body also defines any types of standards that other people can implement.

That kind of ties back to the cookbook. So I think it’s hard to define a cookbook for every company, but then when you look at the company, when you understand the company, the operating model, the goal, like the current goals, let’s say, because the goals also change, you can get… Spot patterns.

And then you can also define cookbooks which work for this specific organization. And it doesn’t mean that I’m going to implement all of the data quality for that domain, but it means that I will empower this domain with tech stack and with people that can explain to them why data quality is important and how they can do it. They are not going to do it for them, but they will show them this is how it can be done.

This is how it should be done according to us. Of course, we are not going to force you to do it this way. You can do it your own way, because unless you have a data office and very high level of control, you can’t really enforce people to do something because if they have their budget, they will be able to do it, especially if they are on the business side of things.

But if you show them the benefits of doing this the right way, then they are going to, and if you like, take them with you on that journey, they will join. Like the same with the doctor, you know, you have people who are kind of sceptic to going to the doctor, but if the doctor is able, you know, to find a way and, you know, find a cookbook that works for this person, okay, maybe, you know, maybe this person, you know, is sceptic around this type of, you know, medication. So let’s find some, you know, of course, I’m not a doctor, so I don’t want to overstep my knowledge.

But you know what I mean? That maybe sometimes you can also, you know, choose a little bit different treatment so that it’s going to work with that one specific patient because it has its own intricacies and you need to find those intricacies and you need to try to standardize, but not enforce, I would say.

[MKLI]

Okay.

[SW]

Yeah.

[MKLI]

And like the process is important here, especially on a living organism, yeah, that are already live, maintained, used on a daily basis. Yeah. We need to make sure like the change management is addressed.

[SW]

Yeah.

[MKLI]

If there is any, let’s imagine these quality checks we had. Yeah. We see that there is a change in source.

We need to adopt the change. We need to make sure it is adopted well across the meaning is retained and the change is communicated well to our users so they can adapt as well because maybe their like reporting will be affected. Yeah.

So they will see different numbers, but we need to make sure like they know why and if it’s okay or not, then it can be assessed and decided.

[MKLO]

Yeah. Okay. Guys, do you see trends on the market?

Do you see how AI is impacting it and this data environment? What is the future of this data organism? What do you think will change in the upcoming years?

[SW]

I think that the companies are going to invest more and more into interoperability between all of the different systems that they have, because I think this is often a challenge right now, especially in the data world is that you have those different tools.

[MKLO]

Yeah.

[SW]

Like you have the fragmentation of tools and in general, like, you know, I would say it has like the benefit of everything. It has its benefits, but it also has its limitations because if, for example, for those, let’s say pillars that we, that we introduced earlier, if you choose one tool for data quality, another tool for data governance, and then another tool for, you know, your data engineering and data architecture, that’s great. Cause probably all of those tools are the best in like their own world.

But then the hidden cost of that is that then you need to ensure that all of those systems are working together and interacting. So if you have, let’s say snowflake in the center and you have metadata in snowflake, but then you also have a data governance system where the metadata is initially defined, or maybe where like the business terms or the, let’s say the semantics are being defined, then you need to integrate it to snowflake and use it in snowflake to tag and like create the metadata for all of the data that you have there.

But in the best world, you would also get it back to the data governance system. And then you have a data quality system where again, it needs to get the data, it needs to have the results, but then the best, if you also have the results in your, in the heart, like in the center in snowflake and best, if you can also see them in Colibra. And then you probably going to, you might have yet another tool that you’re going to, or some custom program that you’re going to use for access, for access control.

And then something else that you’re going to use for the technical documentation. And that’s great. But unless all of those systems are integrated, it brings you much higher cost because you need to keep them up to date.

And what I often see is that those systems are not kept up to date. And this just creates like loopholes. This creates inefficiencies because this even like discourages people to use those tools because they do something in the data governance tool, and then they have to do it again in another tool and then again in another tool.

And they are keep being asked to do the same thing, which is not great, but then keeping all of those tools together, especially that they are changing and changing and changing is hard. Like one thing that I can see is that, for example, snowflake that I already mentioned, I think might also be trying to take on those other functions, so that it can become like a one-stop shop for data. Because we can see, first of all, that they are introducing much more data quality and data monitoring tooling within their platform.

So perhaps I don’t think it’s yet at the state, how are the data, like how the data quality best in class tools are looking, but if it continues to evolve that way, then maybe we can stop using the data quality tool. They will be good enough. Exactly.

Because what we will have in snowflake will be good enough and we’ll be able to make up so much money for not buying another tool and keeping them together that we can add a head force, add some workforce that’s going to keep it well together. And that’s going to even invest a little bit more, even if the functionalities are less. The same with data governance.

Also, snowflake introduced the internal marketplace. You can set up the context. Again, it’s not at the level as the data governance tools, the best in class ones,

[MKLO]

For most companies, it’s evolving good enough.

[SW]

I don’t think it’s, maybe it’s yet there because there are still some things that other tools can do better, like access control, but it’s evolving, it’s evolving, it’s evolving. So I wouldn’t be surprised in a few years, snowflake will be your one-stop shop, like also scheduling, for example, and automation. They are also investing in that area.

And if that happens, I think that just ties all of this, this whole work together because you don’t need to integrate all of this tooling.

[MKLO]

Jump from one tool to another.

[SW]

Exactly. Because it’s integrated by default, like it’s integrated by nature. And I’m excited for that because I think that’s, that’s going to, that’s going to make, make our work more, more effective.

And that’s going to allow us and other data practitioners to focus less on the tooling and more on the process.

[MKLI]

Yeah. Like, yeah. Like with, with snowflake, it’s always like that.

Like we, we are knowing that, like reading the documentation and by the way, it’s, it’s quite good for, for snowflake, but going there like month by month, it’s changing on a regular basis. Yeah. So the limitation that we observed a month ago is no longer there.

So it, I’m saying like, it’s, it’s, it’s, it’s growing rapidly. Yeah. And it’s recently also, we, we saw the integration with DBT.

So, so another solution that we, that we use across our projects. Yeah. And like it, it builds like trust also in our clients.

Yeah. The, the solution, the, the, the solution they chose is, is right. It’s right for them and, and plays the role and could be scaled and, and could be leveraged across the, across the organization.

Yeah. Maybe because we are talking about market trends and, and what we see. I would add also like the, the, the more focus to, to measure the investment companies do.

Yeah. So the, the ROI analysis, so, so where they put their efforts and what they get back from that. Yeah.

The, the data is there, but we need to like, try to use it for, for the purpose to actually target our initiatives, target like clients initiatives towards the, the best possible result. Yeah. Then after that we would like to measure the efforts and calculate the attainment of the efforts and so on.

But, but we see multiple initiatives going around. Yeah. So we will have tools, we’ll have AI ready.

Now it’s, it’s a challenge to, to make work, work out.

[MKLO]

Excellent. Okay guys, that’s it. Thank you very much for being here.

It was a great, great talk. And well, please subscribe, like, leave a comment. What do you think about the topic?

Right. What would you like to hear next? Because we are preparing new episodes, so it would be great to hear your feedback and see you next in, in the next episode of C&F Talks.

Thank you.