Even there, there are certain companies which have more data and they need to do more, okay, with that data to compete in this world. Today, we will look into the operational backbone. One is the size of the company, right?
That brings in complexity. Then, multi location companies bring in more complexity. Multi country companies bring more complexity.
The point is, uh, Business like companies is bringing even more complexity. So you have different, uh, layers of complexity in the company. One interesting tension of an, uh, whether to focus on standardization or whether to focus on mix.
So in, in your opinion, like So some amount of standardization has to be brought in and how your report should be and you know, also it helps if you have a standardization, then integration of data across different systems and all becomes easier. Hello and welcome back to AI Connect, the show where we discuss data analytics and artificial intelligence. Now, as we know, data has increasingly become the lifeblood of enterprise decision making and innovation.
Keeping that in the perspective, we recently launched a three part series focusing on how a certain organization is investing on dedicated data centers of excellence. On the first episode, we looked at how organizations are structuring their data organizations and also the evolving role of Chief Data Officers. Today, we will look into the operational backbone of data centers of excellence, including the key building blocks of a data organization, such as Um, the, the talent, uh, team, um, the governance framework, and also the measurement impact, which essentially transforming a traditional data organization, which was more of a cost center to transforming it to a truly, um, uh, a profit center, basically, which is driving innovation and growth.
To talk about that and more, I have Anirban Majumdar, who is the co founder of Rations. Anirban will share his insights in terms of how enterprises are structuring their data team for the maximum business impact. Now, Anirban, glad to have you on the show.
Thanks, Jaydeep. So, um, let's dive in and talk about, um, the first aspect. Now, many organizations are moving away from the legacy data team and investing on dedicated data centers of excellence.
Right. Now, um, what is the fundamental difference between the old model and the new model? Can you share some light?
Yeah, sure. So earlier, uh, I would say even now in many, uh, enterprises, there is a standard data team, which in many cases, it's under the CIO's organization, which is only related to ensuring that all the data is brought in, stored somewhere, and then archived and all those things. But the data team may not have the, you know, bandwidth or the capability to figure out how do we ensure that this data is used to provide higher benefits to the company beyond the regular, you know, storage and archival requirements.
And so for the IT team, the data team in IT. Uh, the data fields are not being stored, they may not know how that makes sense to a finance team or the sales team or the, you know, logistics team. So, which is why it is very essential that there is a central data team which will have access and have the view of overall data, uh, the sphere in this company, and then to figure out, okay, how do we ensure that we get more value across different functions across different geographies of this, uh, data.
So that's where companies are slowly moving into. The centralized, uh, data organizations and there we have now so many companies including most of the Fortune 500 ones having chief data officer, right, which is now, uh, no longer, uh, uh, you know, small part of the CIO organization, but in many cases it may be reporting to CIO or someone else, but it is a separate distinct organization. And that's where the, uh, data, uh, Uh, COE or data automation comes in today.
So it's a good segue to, uh, probably the second point that I wanted to ask you, is, um, as organizations set up dedicated data centers of excellence, often there is a debate in terms of where does that organization sit? Inside the larger org, and what would be the reporting chain and so on. And, um, looking ahead, how will that, uh, how would the data center of excellence cater to, um, inter departmental requests?
Right. What's your viewpoint on that? What do you think really works in your experience?
Uh, so we have seen different models out there, uh, but I would like to think that it also depends on, uh, the kind of organization you are, right? So I would like to create, uh, say, like, there are probably the three large buckets you can put these organizations in terms of The, their data generation, data consumption habits, right? So at one end you have all this, you know, uh, e commerce or fintechs and those kind of companies, the real technology based companies, which are more or less running on data.
Without data, these companies are dead. So there is one set of companies like this. Then you have the more traditional enterprises, even there, there are certain companies which have more data and they need to do more, okay, with that data to compete in this world, like say your retail, your banks, your financial services, right?
So these are the companies or even pharma. So they have a lot of data, they generate a lot of data on a daily basis. And they have to figure out, you know, what all they can do so that they are always in the, uh, race and winning the race.
The third order, more traditional, say manufacturing and other companies will be there, or governmental organizations where you are not generating that much data. So, but you also still need the data to compete in this world. Uh, but again.
That, that amount of data being generated on a regular basis, daily basis is not that high for them to, you know, uh, keep doing something with that on a regular basis. So if you look at this, then it makes sense for the third type of companies, it could be that the data organization can still be under the IT. Or, you know, uh, maybe the CFO or someone, it doesn't matter, but it can still work with that, right?
But for the other two, it is very essential that these data organizations have a view of the full company, all the functions, you know, horizontal, vertical, whatever way you want to cut and, uh, slice, dice this, uh, company. So they have to have that. So it makes sense to have it, uh, under the COO, CEO, CEO and have provided access to all the different departments.
And that way they can, you know, uh, provide value to all this, all the parts of the company. Does that change in, uh, or has any dependency on company size or irrespective to SMEs? Yes, of course.
So there are, uh, There are two, three different, uh, angles from which you have to look. One is the size of the company, right? That brings in complexity.
Then, multi location companies bring in more complexity. Multi country companies bring more complexity. Then multi, uh, business land companies bring in even more complexity.
So you have different, uh, layers of complexity in the company. And the way it is done, and especially we have seen that, More than anything, if a company is spread across multiple countries, that brings in the most amount of complexity because you have so many different regulatory compliances and other things. So, of course, so all these things, uh, make it even more important that you have a centralized data organization so that they can take care of all these different aspects.
And ensure that, you know, the business user has the right data and the right insights in front of her. Got it. And I think, uh, the other aspect of it is that, um, looking at the specialization, right?
Functional specialization, such as supply chain, um, such as logistics. Um, such as marketing, sales, I think across these different functions, um, I mean, we can have maybe a dedicated episodes on each of these in terms of how data centers of excellence can, um, and, uh, are impacting, um, basically decision making, um, and innovation, right? Um, maybe that's something that we can probably take it up, but I'm very curious to actually also maybe in one episode cover the functional aspect of it, considering that I think, uh, Prashant's, um, has, uh, you know, been working with different types of functions in the data organization system.
Right. That would be an interesting take on that. Correct.
So, uh, coming back to, again, um, on the data centers of excellence. Um, one interesting tension of an, um, in data center of excellence is whether to focus on standardization or whether to focus on innovation, right? And so in, in your opinion, like leading organizations, how do they balance between these two, right?
Because both of them have. Uh, the only importance, um, and standardization gives you basically scale, replicability, um, discipline, you know, um, quality. Right.
Regulation, of course, brings in UI aspects and your way of sort of, um, adding and arranging management. And then applying data. So I think that both are required standardization and innovation in this organization.
So standardization is required because what happens is when data was, the data teams are spread across different. functions or different, uh, geographies and all. So they went their own way, for example, in terms of the tools and technologies being used.
So we have seen organizations, pretty large organizations, for example, for visualization, they're using Power BI, Tableau, and all different types of things. Different teams are using different tools. So in such cases, Bringing in standardization helps because ultimately these keep rolling up to a certain level, right?
And so then the same, you know, senior executive, he can't be seeing one report in a Power BI dashboard and another in a Looker or something like that. So some amount of standardization has to be brought in and how your reports should be and you know, also it helps if you have a standardization, then integration of data across different systems and all becomes easier. So from a tool and technology perspective, yeah, it also makes sense to have some standardization.
It also makes sense to have some standardization in terms of how do you do data governance, how do you measure and then ensure data quality, right? You cannot have different standards for different people for that. So you have to have a well defined data governance plan.
You have to identify who are the data stewards, who are the data owners, and ensure that they are doing the right set of things or getting the right set of things done to ensure the right quality data is getting into the company's data warehouses and data lakes. Right? So those kind of things.
have to be, you know, of a certain standard, uh, you know, uh, so that everyone gains across the company. But then, as you mentioned, there are different things happening across different functions, right? So, you cannot standardize that part.
You should not try to standardize, but see what is the latest things happening, what are the different, for example, You have this agentic AI coming in in a big way. Now there are certain processes where this can fit in very quickly. But there are certain other processes where it may be a very difficult fit.
So as you see and experiment and all, you figure out, okay, let me take up these two or three use cases and this could be in finance, this could be in sales, customer support, and you try it out and see how you go ahead. So that kind of quick experimentation and then coming to a conclusion whether This is worth going further or stopping. So that also has to be a part of the data seal.
That makes sense. Um, switching gear from this to, uh, you know, I'm also curious to know, um, what is the talent composition in traditional data organization, data team? versus, let's say, dedicated data centers of excellence.
Is there a, do you see that there are any specialized roles that are evolving, um, or are critical or, um, are in, uh, sort of already in place, um, from our client interactions? Yeah. So, absolutely, it's happening.
The, the change of pace is very rapid. So, in the earlier traditional data teams, you would have primarily data engineers, And some of the folks who are more into BI and visualization kind of thing. But now, we are seeing multiple different roles in the central COE, right?
You still have the data engineers, but you also have a small set of people who are looking at the whole data governance, data quality, uh, thing along with them. Then you have the people who are, I would say, business analysts. Come inside generators, right?
So they may be using some amount of dashboarding tools But their primary job is to analyze data and provide insights to the different business Executives in the company. So that's the second part. Then the third part comes in terms of The whole data science part of it.
So, whether it's Gen AI or traditional AI or any kind of work that you do, these are the people who are talking to business, understanding the requirements, building models, trying it out, and then finally coming up with some amount of models, uh, which, you know, can be taken to production. And finally, you have those analytics engineers or ML engineers who are taking it to production and also keeping the model refreshed on a regular basis. So, the whole scope of data COE has moved from, right from getting the data in from different sources to providing, you know, this model based predictions, recommendations, and generating, uh, different, you know, documents, summaries through LLMs, or it could be some traditional AI, you build, you know, customer, uh, insights and all.
So everything can come under this single data COE. Got it. Got it.
So are organizations, uh, organically sort of Training people internally to, um, go and sort of perform on these different roles, um, as we sort of see that space is evolving very quickly. Or they, increasingly depending on a larger team or larger ecosystem to support, um, when they have a rapid. sort of, um, scale in their business to support that need, right?
Yeah, so, uh, growing organically, uh, this whole data share becomes very difficult, especially when you have newer roles coming in, you have newer tools and technologies coming in. So, there's a lot of specialization which is happening in this whole space. So, this is where what we have seen, uh, you know, companies are really leveraging.
The power of data and the power of AIML. They are building their own core team. They are also working with partners to do various things, right?
So it's a very nice way they are combining the strengths of their own team and the partner teams. And getting things done. And that is helping them to move much faster than they would have possibly done by trying to hire or train everyone on their own.
And without having to operation over it. Absolutely. Awesome.
I think this is all I wanted to cover in this episode. So, so far we have covered in the first episode. How, um, leading organizations are structuring their data organizations and the evolving role of, uh, the chief data officer.
Today we discussed more of an operational backbone of how the data organizations are run, um, and what are typically the composition of the skill sets. Um, what are we looking for, for the next, I mean, I think one we can probably look at the functional aspect of it. Anybody else you have in mind that probably will add value to our audience who are, uh, either looking to set up a data center of excellence or probably, um, looking at opportunities within data center of excellence, uh, from maybe even job perspective, you know.
Yeah, so I think, uh, we can, one, look at Uh, towards the key success factors to ensure that data COEs succeed. And then also look at, you know, how different functional aspects and how different other aspects can be taken care of within the same data COE. Uh, which will ensure that the different business functions can get what they expect from the data team.
Got it. Excellent. So I think we have, uh, nice couple of topics to cover in the coming episode.
But for today, we are going to wrap up here. Thanks for joining us. And we will continue to do this Data Center of Excellence series.
We will appreciate your feedback, comments, um, on our YouTube channel. Um, and if there is any suggestion that you have, please share with us. Thank you very much once again for joining AI Connect.
And we'll come back soon with a new episode. Thank you.
