Good morning. Good afternoon. Good evening. Good whatever time it is.

Wherever you are on this amazing planet of ours, I am Malcolm Hawker. I’m the host of the CDO Matters podcast.

I’m also the CDO of Profisee. If you’re thinking about MDM, Profisee is an awesome choice. We’re gonna actually talk about that a little bit today. I don’t advertise much on the podcast.

Well, actually, I never advertise on the the podcast. That’s not our goal. Our goal is to is to educate and enlightened chief data officers. But the story we’re gonna talk about today is is is so good.

I just I just couldn’t ignore it as a as a podcast as a podcast subject. So today, we’re gonna we’re gonna talk about MDM, master data management, my favorite topic.

And I’m absolutely thrilled to be joined by Raju Mudhunore, who’s a director of data with Lexmark. Raju, hello.

Good morning, Malcolm. And, you know, firstly, I’ll start with thank you. It’s a privilege to be here. Thank you for the wonderful opportunity and looking forward for this conversation.

Yeah. Me too. We Raju and I shared a stage at at Gartner in Orlando when we were talking about the the the great story that Raju has to tell about their experience with with MGM at at Lexmarkk. You know, I I in all of the content that I post, whether that’s on LinkedIn or whether that’s white papers, you know, often a lot of the things that I talk about are things we need to improve.

Right? And because there’s a lot of frankly, folks, there’s a lot of things we need to improve. And and I get often feedback around, hey. We need to be celebrating successes more. We need to be screaming from the rooftops. Hey. We did something cool.

And that’s really what today is all is is all about. So I’ll I’ll I won’t bury the lead. The the lead is is that is that Raju and his team were able to drive two million dollars in revenue. Right, in in in revenue for for Lexmarkk.

It’s revenue. Right, Raju? It’s not just cost savings. It’s it’s revenue. That is the key story.

Right?

Correct.

Alright. Okay. So that’s that’s hey. Bing. Bing. Bing. Bing. That’s a success story.

There there’s so many things to like about it. We’re gonna dive into the details on on how you did that. We’re gonna dive into your architecture. We’re gonna drive dive into your operating model, your team.

Yes. We’ll talk about technology. But that but that’s the lead here, guys, is is that when you do things the right way, and Raji is gonna help us understand what that actually means and how to do it. But when you do things the right way in data and analytics, yeah, the stars can align and we can drive meaningful value to our organization.

So so two million is is is a great number and it’s certainly something to to celebrate. So let’s let’s start with the top.

What what first drove you to to to to to MDM, Raju? What what was the the, hey. We think we need MDM in order to solve the problem. What what was broken and and and what what led you to to MDM as the solution?

Yeah.

So to set the context here, you know, Lexmarkk, as you might have heard many of you, is a is a global leader in cloud enabled imaging and IoT solutions from several decades. Alright. But we have been having lot of legacy systems and it was always a challenge on how do we optimize our device usage or even supply fulfillment across one seventy countries and numerous third party vendors in a era where simply selling more devices was is is no longer a viable growth strategy.

Right? Accurately, timely supply fulfillment is essential for our customer satisfaction and revenue. But, you know, we had fragmented inconsistent data across our ERPs, which is SAP, our CRM, Microsoft Dynamics three sixty five, and lot of it is still in the legacy systems like Oracle, you know, Siebel.

It made it really difficult to kind of bring in that flexibility and support to have realized the value of end to end data consistencies.

And that’s obviously the number one reason as we know we have to work on our foundation, work on our basic needs of data, and evolve around that. That’s when we were trying to, you know, put in a path like, you know, we need a dedicated MDM system in house to to host and to get there to realize the value.

So you got data in in three sixty five. You got customer data and I assume equipment data sitting in SAP.

Yeah? Correct. Yeah. And and what it and and something was I gotta backtrack a little bit.

Something was really interesting in what you said, which was we’re already penetrated in a hundred and seventy countries. And and, Lexmarkk, I mean, you’re talking about, like, large imaging pieces of large imaging equipment. Right? Like, large photocopiers, large printers, like the the some of the biggest of the big machines that you can get out there.

Yep.

Already in a hundred and seventy countries and you said, hey. In order to capture more revenue, we’re gonna find we need to find ways to run more efficiently. That’s pretty much what I heard you say. Correct?

Correct. That is right. Okay.

So what you said was, alright. We got this problem and that we’ve got data in three sixty five dynamic data in SAP. You probably got data in a multiple others other systems, but the CRM ERP divide is a classic one in in MDM.

So so you recognized, hey. That to get more efficient, you know, we we we need to bridge these systems together. Now was the was the goal always kind of how engaged was was the business in all of this? Or is this something that you were that you were driving and that you identified? What was the collaboration like with the business in on the early stages of of MDM?

Yeah. So I I feel like it’s it’s it’s a bigger initiative. Right? Like, SMARC underwent and this whole idea of having a digital transformation.

Because as you know, as I mentioned again, is, you know, we had a decades long legacy in just the printing and imaging. Right? But now we’re trying to kind of evolve out of that hardware centric leadership into a technology company.

And we wish to leverage our own IoT, our own cloud and AI, innovative solutions and services across our customers.

And this digital transformation is is kind of defining our new business model, our customer engagement, to kind of create the global impact. Right? And we have started commercializing our IoT and edge computing expertise, launching solutions for industrial IoT, remote monitoring, you know, healthcare and many more often in partnership with other technology leaders out there.

And and we’ve had a great success. Right? So all of this evolution started about three and a half, four years.

And that that was the time when we internally brainstormed with our business.

Right? For example, a big line of vertical businesses, annuities business, And realized that, you know, if we really want to go and take advantage of this new technology and evolve into the business leader as a technology company, we will have to do more.

And their biggest one of the, you know, obviously, the challenges that I have just mentioned was more like, you know, how do we how do we find new channels of revenue? Like, for example, our business model, traditional business model was built around generating revenue through, you know, services and supplies over the life cycle of our printers. So rather than just relying solely on hardware sales, we we have to make sure that we are engaged with that customer throughout the life cycle of the printer and generate value through providing different services and supplies.

Right? However, with a growing number of customers, you know, growing number of customers have actually begun sourcing supplies such as cartridges from, you know, unauthorized or black market channels bypassing Lexmarkk.

And that’s that’s that’s one just example that annuities business came to us saying, hey. How do we solve this?

And that’s when we kind of launched our annuities digital thread initiative, which basically leverages the advanced data analytics to kind of identify those non loyal customers and target them with the tailored reengagement strategies. Right? This effort has enabled us proactively renegotiate and recover lost accounts to focus those win back strategies and campaigns.

Alright. So so this was part of this focus on MDM was part of a broader digital transformation. That’s awesome.

Right?

That’s certainly something that I saw a lot when I was an analyst and still something that I’m seeing to this day. There are there are still a nontrivial number of chief data officers who are also chief digital officers, which which I absolutely love.

So was the core problem from a customer data perspective in that, you know, you would be shipping out these machines, SAP would be aware of where the machines are being shipped to, right, where they’re actually landing, but the customers potentially buying them from a CRM perspective or whoever you’re selling to may not necessarily be the same places where these things are actually getting delivered. Is was that one of the core breaks from a data perspective?

It’s it’s it’s a you know, even to just track devices and consumables throughout the entire life cycle of that one hardware that we produce, right, from manufacturing to end of life at a serial number level as as as a nightmare. But just think about it. Right? We manufacture the original equipments, and then we we like I said, we operate in, you know, across one seventy countries. So there’s different geographies, different distributors, and different our our channel partners. We have about six thousand plus channel partners. Right?

And then each one of these partners would have their own vendors before the device actually gets deployed into deployed at, you know, at a at a at a at a customer site. Excuse me for this.

No worries. Happens to us all.

So yeah.

So so even to, you know, by to figure it out, like, the whole path to market and across different geographies was was clearly a a big challenge.

And we couldn’t have solved that without integrating our data from enterprise systems and connected devices, right, through these IoT platforms. Obviously, you mentioned our ERP, which is on SAP S4HANA.

We had to integrate our order system or, you know, SOL tools with our CRM, which is our d three sixty five is our core CRM systems. And then our manufacturing units, which are, you know, locally here in US and in whereas Mexico.

We we would have to tie all of this to get that deep insight into the usage behavior of the customer, the third party supply threats, and overall market integrity.

So that was that was kind of the mixture where we were heading into to solve this digital threat.

This nightmare from a data perspective. When you start talking about six thousand kind of distributors slash resellers who ultimately kind of own the customer relationship, the direct customer relationship, but you you also have a customer relationship because you’re provisioning the the cartridges, you’re provisioning the service to to those customers. So you’ve got a a a multilevel. You’ve got a hierarchical relationship with your customers, but you’ve also got a direct relationship with your customers.

You’ve got two different two different platforms, SAP versus Dynamics that are managing customer data and on and on. This is this is by the way, folks, if you if you’re not doing MDM, but this sounds familiar, chances are pretty good you need MDM. This is this what Raju is describing here is a classic classic MDM problem because, typically, the way the customers are defined in a CRM may not necessarily fully align to how they’re they’re defined in an ERP. One is very much a sell to, the other much is very much a bill to or delivered to.

So this this just kind of screams of of of MDM.

Tell us like what tell us, you know, what was the deployment like? I mean, were you were you, you know what what was that lift like? What was the experience like? And what what would you share with others in terms of what you might do differently from a deployment perspective?

Yeah. So I think to your point, Gabe, you know, as you mentioned, obviously we have a lot of customers through our primary vendors, our partner vendors of six thousand but we also have direct customers, Right? Which are huge in numbers. Right? Like for example, Walmart, you go across the world, every single Walmart you’ll find like smart printers. Right? Walgreens, Apple.

You name it, you get it. Oh, no. There’s several of the, you know, top five hundred companies globally recognized have are are like smart customers. Several banks, several federal institutions.

So it becomes more and more challenging, especially when you go into the sensitive domains of like federal. There’s more, you know, there’s more security concerns, different compliance that you have to manage. And and the data, you know, saving that data, you know, securing that data is is becomes more and more challenging. But anyways, you know, when it came to the fundamentals again, as as part of that digital transformation journey, we made one thing clear was like our legacy MDM solutions, you know, have to be fixed. Because we had SAP MDM in the past, which was, you know, obviously, you know, at Lexmarkk and very inflexible, you know, unsupported, unable to handle the evolving business needs, especially the complex hierarchies. And like we have mentioned many to many relationships, right, between customers, our devices, our service vendors.

So we started out looking a modern scalable MDM solution to to to which could unify and govern data, enabling all of the advanced analytics and the machine learning for demand forecasting and so on other use cases. Right?

And that’s when we started to go out in the market, scout for a cloud native MDM. And we did a thorough assessment internally with the partnership with the business and business had a big say in it. We kind of analyzed some of the leading MDM tools back then.

If I remember it right, it was one of them was Informatica and and SAP’s own modified version which is called MDG.

And and Profisee, pitched against. We looked into a lot of the details in terms of, you know, really how how does how good is these tools on the functional features like, you know, modeling capability to integrate with other tools inside the ecosystem of Lexmarkk.

How good are they at some of the basic things like, you know, data cleansing, validation, matching, and survivorship.

Obviously, lot to do with the data stewards, which which our business people are. And and overall, we wanted to see something that would natively fit and would help our dedicated data governance tools to operate even better.

At the same time, we we, you know, because we are, you know, SOC compliant and other compliances and are needed to be audited, we were looking for a solution that is fully thoroughly transparent and visible for auditing.

At the same time, obviously, the complexity of the business. Right? You’ve seen the hierarchy management.

It’s many to many multi hierarchy, multi domain systems here at Lexmarkk, and we wanted something that could handle all of these complexities and be able to simplify a workflow with more of a UI. By the way, that’s a big hit with our business community because as they started getting into this tool, you know, obviously, the UI is the center of the traction because it’s let them create simple workflows, let them define the data quality rules, just up there’s tons of them automatically out of the box, but also through this graphical user interface, the drag and drop features and just be able to define some fundamental things without the need of the engineering folks of IT is always a huge win situation for our for our, you know, business analysts or the data stewards.

So that’s that’s how we started out there. And then, you know, we analyzed, we pitched all these three.

Ultimately, business chose Profisee and we we went ahead. We had the full support of the business. We, you know, all the funding, everything was approved. And then we started kind of putting it out on the ground to to do a couple of quick POCs and then eventually roll it out to on a larger scale.

And then today, we have our our full customer and vendor managed completely on on Profisee globally for us, for all all of our customers management. We have primarily, I would tell, you know, three big channels of inputs of the master data, especially on customer, for example.

Our salespeople obviously, you know, through go through the CRM process on where we do the sales. Any of the lead new customer that comes in, a potential customer, they will have a new input record, so to speak, created in the CRM. Right? And that’s what is fed to Profisee. Likewise, our orders coming through our SAP would also feed into Profisee. And then a big chunk of our ecommerce platform as people, as customers come in and place orders or or purchase or do any transaction, we capture that customer information as well.

All of this is fed into Profisee and then Profisee is internally, we have workflows defined through our business and our business has an oversight on it to kind of cleanse this, apply all the DQ rules, kind of merge them into and form this whole three sixty degree customer record, so called golden record, which is then now distributed back into all of our system of records, including into our analytics platform, which is our, you know, built on Azure Databricks on top of our adoption to Microsoft into our digital, you know, new age lake house.

So that’s that’s how the whole deployment is, you know, sitting at a very high level, if that makes sense.

Yeah. So we’re gonna we’re gonna dive a little deeper into Azure and Databricks in a second. But I do wanna double back.

If you can recall, how long did your first implementation take? I mean, you’re this is this is classic operational MDM. Right? Where you’re talking Yeah. Bidirectional flows of of of information between source systems and a hub, which is which which is a a bit a big lift. Do you recall how long that took from beginning to end?

Yeah. I think we were quite lucky in that terms because we have a very, very strong business unit, in house business unit.

‘ve been doing MDM is not new to them. Right? Who’ve been doing MDM with legacy tools, but this was a new modern way of approaching it. Right? So the good news was we had lot of, you know, in house knowledge of what our data stands for, what it really means, how what are the different sources, how to integrate them.

And with as we started the integration with Profisee, I would say, you know, the initial six months was the time when we did a lot of the groundwork and we did engage the professional services of MD Profisee team.

But the full fledged implementation was about a year where the first use case was flying through, actually deployed in production.

Yeah. So that’s that is actually pretty good across what we saw when we were at Gartner across all use cases, both analytical and operational. So analytical is the, hey. We’re up and running in, eight weeks.

And that’s actually that’s actually possible, and we can do that. I’ve had multiple conversations on the podcast about how to do that. Actually talked about that in a previous podcast talking about potential tariff impacts and how to create three sixty degree views of something. That’s an analytical MDM.

Across those forms of MDM and all the operational forms, which is what Lexmarkk has done, the the the average that we that we saw was around nine months. Now keep in mind, that’s with the overwhelming majority of first phase MDMs being those analytical styles. So the fact you’re up and running in a year with multidirectional bidirectional integrations to multiple systems is actually really, really good. I know that doesn’t maybe not sound great to a lot, like, wait a year.

I have to wait a year. But when you’re talking about a two million dollar return on investment, that’s actually really, really good.

And you had deep business engagement, which is also really good. You touched previously on integration to governance tools. Are you using and you mentioned Azure. Are you using Purview?

Yes. So Okay. So that was another critical critical factor for it because we are strategically invested more in Microsoft.

And we’re trying to find, you know, at the time when we were choosing this MDM platform, we also one of the big criteria for was that any tool that we bring in or any platform needs to be able to seamlessly integrate with our Microsoft ecosystem. Right? Including all the Azure services and Microsoft Purview because we also making it a natural fit. Right?

Profisee actually was, you know, fitting really nice into that criteria. Yeah. And then there is, you know, we always wanted to make sure that the synergy between our MDM system and our data governance tools always needs to be there because it it would think about it. Right?

Any MDM system that you’re trying to build, the biggest challenge is is how do you build that master model.

Right? And if imagine now if your MDM system is able to seamlessly reach out and it’s a two way communication between your, you know, MDM versus your data governance tool dedicated catalog.

Your MDM, in this case, Profisee, can actually now peak the and and, you know, kind of browse through the full model of the system of record data, the entire landscape via the Purview catalog. Right? It can it can discover what are the systems you have, what are the schemas, what are the data elements.

That information is so critical in that it it’s it’s it let you create that master model inside your MDM. Right? And once your master model is fully cooked here, the same can then be passed on back to the catalog to add this new model into the catalog. That kind of unifies your data governance. Right? Your end to end policy enforcements and and seamless data cataloging.

That to me is a key key, you know, way to maximize your business value.

And it was for us, it was also doubling down on the Azure investment.

So I would say, you know, I I think it was very, very critical for us, a strategic partnership with whatever the MDM tool, the Profisee in this case with Microsoft was was is proven a really, really charm for us.

Yeah. It’s there’s there’s a there’s a few things to like there.

One is the integration between the system that is helping you document and define your policies, which is Profisee. Right? And the system which is actually helping you enforce them, which in this case is Profisee.

So if you are a CDO out there thinking about a data catalog, don’t have one today. Data catalog, largely synonymous with the idea of a data governance tool. These are largely synonyms. The differences are are are trivial between the two.

They are largely the same thing. If you’re thinking about a catalog, then you need to be asking yourself, okay. Great. I can document my schemas.

I can understand my relationships. I can document my my data definitions, my business class, my business terms, and do all that stuff. It’s great. But then once I have a policy defined, which is how do I define a customer?

Who’s the key data steward? Who’s the data owner of this element? How do you actually enforce that? That’s a question you need to ask yourself if you’re looking at the ATLANs and Calibris of the world.

Those are great solutions. Don’t get me wrong.

But where the world is heading is a deep integration between catalogs and MDM, and that’s one of the the value props that Raju was just talking about. So that’s that’s certainly one thing to like. The other thing that you you had you had talked about was, you know, integration to Azure, which is an integration into your analytical stack. Right?

You talked about IoT being a kind of a key to the business case here. Right? Meaning IoT is you got these sensors and all of these machines across the world, and those those those sensors are capturing data about, hey. I think I might be breaking breaking or I might actually be broken or I’m just simply run out of paper or or ink.

Right?

But you get all this all this telemetry data coming back and you need to associate that to a specific customer.

Is that all that IoT data is sitting in in Azure? Yes?

That is correct. We we heavily leverage Azure IoT hub, Microsoft IoT hub.

That is then, you know, in the back end hooked into our devices to collect all these telemetry. Primarily, I would say the telemetry that we have is is the usage data from the printer and then also the health check data of the printer. To your example, that that’s that’s what’s that’s helping us doing lot of predictive maintenance use cases.

Yeah. And and that is, folks, the the you know, if I didn’t touch on this early when I when I kinda led the story with the two million dollars, We’re we’re getting to the heart of it here, which is the ability to associate that telemetry data and that equipment data and all that that IoT data with the right piece of equipment with the right customer.

Right? And with potentially even a service agreement that I assume that you’ve gotten a service agreement associated to a customer and and on and on and on. Like, that is that’s the magic that we’re talking about here, that’s where the money came from, which is now you’re able to capture more accurate insights about what’s happening and where it’s happening and for what customer it’s happening to. Correct, Raju?

That is that is absolutely right.

Perfect. Perfect. Well, let’s talk about Databricks.

I hear a lot these days of people saying, okay. You know, hey. Databricks, they got a catalog, Unity.

And, you know, they’re enabling some more advanced data management functions. I can just do MDM in in Databricks. What do I need an MDM for?

Is is as a part of this evaluation, did you consider, you know, do building some sort of native capability in in in Databricks? And if you did, what led you against it?

Yeah. So I think fundamentally, Databricks, if you look at the background of the whole company, right, they’re more of a Spark based shop. Right? And then they ventured into building an end to end data analytics, a unified platform that could kind of support both your data science community as well as your traditional business intelligence community of generating reports and dashboards, etcetera.

We never thought about them as an MDM. Right? Because MDM is a totally different ball game. So they as they sit today for us, they are our unified analytical platform and they’re downstream to our MDM.

So Profisee, once it’s, you know, like the customer and vendor that golden records for these are created as part of the distribution, like I mentioned earlier, back to not only the system of record, but it also sends a copy of that master data into our analytics into our analytics end to end unified platform, which is Databricks.

And then that’s one of the other things is, you know, like I said, integration. The capabilities to in integrate seamlessly with these, you know, cloud based tools across our our landscape was another plus with the Profisee is that, you know, it would have either direct connector or would have at least exposed API so that our inbuilt framework can actually build an integration channel to to our data in Databricks Lakehouse.

So honestly, you know, to circle back on your question, we never thought or have applied any master data capabilities to our to a Databricks environment. It’s it’s more it’s built for something more of advanced analytics. You can do tons of data science, AI, ML activities.

The beauty of it is it brings all of that data into a single analytical surface so that both all your downstream consumers, whether your internal business stakeholders or your external business stakeholders, are all looking and peeking into the same set of data. So it’s more of an integration and doing heavy transformations, but it’s not really the tool that’s, you know, help you master your customer data or your vendor data.

So you do have a team of data scientists at at Lexmarkk that are that are doing the cool ML predictive stuff. I mean, they’re you’ve you’ve got a team that’s doing that. Correct?

Correct. That is that is that is right.

Yes. So if if they want to use a customer master record that is provided by Profisee as as a part of any sort of ML based, AI based workflow in Databricks, they can use that because it’s exposed there. Or if they wanna do something completely on their own and go straight to source, I assume that they could do that too. Is that is that correct?

Yeah. So for for our scientists, we have strategically, you know, our data science team is about forty plus team members now. Okay. This is all fully ingrown. Thanks to our, you know, capabilities to go out there and, you know, take advantage in partnership with North Carolina State and getting the fund enough to build our own AI academy in partnership with the schools, state schools there with NC State.

We have been able to grow and scale up our own data science teams internally, which is now about forty plus team members. Right?

They they take tremendous advantage because think about this. Right? Data science teams, they need to be focused more on the the business needs and how to solve them leveraging the advanced AI and machine learning algorithms. For example, one of the big machine learning algorithms commonly acronymed as MLDDA, which is basically a, you know, stands for machine learning days to depletion algorithm. Meaning, it’s gonna predict a given printer in our case, which is a connected device, is how many more days it could basically deplete out of the cartridge or the ink in the cartridge. Right?

So that’s just an example, ML. Now for our data science teams to focus on that, they don’t want to deal with the data wrangling. They don’t want to, you know, deal with the nitty gritties of how the data is arriving there, when is it arriving, focus on all the day to day data pipelines and operations around them. They want to focus on higher value analytics.

And this, you know, I would definitely say that Profisee has reduced that need for the master data for customer and vendor so that our internal team resources can now focus on that higher objective and not be cluttered about the quality of the data. Right? Remember, the whole data science is only as good as your basic foundational, your internal data is. If if you feed the crap or the junk to these data science models, they’re gonna just predict and hallucinate all their lives.

Right? So it’s it’s absolutely critical for us to be successful in our new age data science evolvement. The data is provided crystal clear of any error free, any null value free. Right?

And it has to be properly integrated and provided facilitate our acceleration in our data science teams.

So using Databricks parlance, what you’re creating in MDM is is a gold medallion record. Right? Correct. You know, MDM would be the silver layer, the processing layer, and what you what you create is this gold medallion curated, trusted, governed data product.

My friend Helen Grinster from Gartner would call a master data record the quintessential data product. That’s what you’re exposing. Your data scientists don’t need to go and remake the wheel. They don’t need to go do the wrangling, the data quality, all the transformations to do it.

It’s there. It’s curated. It’s governed. It’s signed off. It’s auditable.

You’re ready to go. It’s interesting. When when you share this story, your story is is similar to a few other other kind of bigger companies that I’ve talked to, particularly Nestle Purina is another one that I’m very that I’m very familiar with, where they’re doing the exact same thing.

Where it seems to me where they’re using, you know, a master record, which is the record that you should be using as a data scientist to to do any AI, to do any ML. And it seems to me like there’s a maturity curve here. Meaning, if you’re less mature from a data science perspective, there’s still this belief with a lot of companies that everything is net new, and that I need to start from scratch. And then if I’m gonna build a model that that predicts customer buying behavior, I need to start net new and go straight to source and only get data from source and do my own entity resolution, do my own wrangling, do my own transformations, and that’s the only way that I could solve that problem.

Yet, there are some of the biggest companies in the world that are saying, hey, why do all that?

When you’ve got this curated trusted gold master record, use that. You don’t have to nobody likes the wrangling. Nobody likes the transformations. Nobody likes all the quality stuff.

Here it is. It’s got a bow on it. It’s ready to go. It can be consumed.

You’re doing it. Purina’s doing it. Other companies are doing it. And and it seems to just make logical sense.

If you absolutely positively need to start NetNew for whatever reason, perhaps there’s business cases there, but I I I love what you described.

The last thing I would touch on, and and this is something else I don’t think a lot of companies necessarily consider when they think that Databricks can be or or Snowflake can be a viable MDM platform, is something that you touched on earlier, which is that integration to the operational systems, particularly your CRM. Right? Which which is if you cut if you’re a customer going through a CRM or if you’re a customer going through a Lexmarkk portal and you’re putting in data or or accessing data, your experience is gonna be the same there as it is to an internal user.

And are you going to be able to connect that externally facing CRM or that externally facing distributor or wholesaler website to to Databricks to access that data.

I’m not sure you’d wanna do that for a lot of different for a lot of different reasons. So, I mean, I just see the operational nature here as something that a lot of people overlook, but something that you’re doing that you’re doing very, very well. So is there a search before create capability within within Dynamics to make sure you’re not creating duplicate records based off of what’s already in MDM?

Right. So like I said, you know, yes, the the customer record, for example, is is does get created in the first place in the dynamic to three sixty five. Right? But then it gets reconciled with Profisee because there’s a real time integration into Profisee, and then Profisee picks up the golden record and feeds it back to the three sixty five.

If there’s a new record created. Right? If it’s an existing customer, it’s just gonna give you the existing record. So, yeah, there’s there’s always that back and forth.

Back to your point. Right? Can somebody literally create an MDM kind of a process on these modern data platforms like, you know, Snowflake or data’s Databricks?

The long answer is yes. Yes. If if you could if you really want to do that, you can you have the tools of your choice because these platforms are well integrated out of the box with your CRM and SAP. It’s not that they don’t have data. Right? But think about it.

If you were to create, it’s gonna be a such a tedious process of creation. Anybody who’s been in the world of MDM from last three decades, they would understand clearly what I’m trying to convince here is if you that’s like building an MDM system from the scratch on Databricks or on Snowflake. It’s just gonna take maybe five, six, in ten years, I won’t be surprised.

And time is of essence in this digital world right now. I don’t think that’s a viable solution by any means. Right? So it’s it’s far more better off to master the data using a dedicated tool. Every tool has a, you know, has a purpose.

And that’s where we put in our brain and our brainstorming thought was like, you know, let’s just have a native MDM platform that can just focus on the MDM and not solve the whole world. And let’s do the, you know, bigger things like AI, ML, and traditional analytics on top of Databricks.

There’s there’s two more things, and then I’ll stop picking on Databricks.

One of them was you talked about the importance of business engagement through through easy to use interfaces and data stewardship. Right? So what would it take to build kind of customer facing, easy to use, easy on the eyes interfaces that are configurable by domain. Meaning, somebody in sales and marketing sees one thing for a stewardship experience versus somebody in finance sees something else.

Could you do that in in in Databricks Potentially. But what would it take to do that and how what would it take to maintain that? The second thing that I would touch on is I experienced something like this the hard way. The first time I ever did MDM is I buried a whole bunch of business rules in extremely complex ETL processes.

This is so this is twenty years ago. Right? Like, we were Yeah. We we were do we were doing all sorts of transformations.

We buried them in ETL processes.

And I kind of view that that that what folks are doing in in Databricks in a very similar light, particularly when it comes to matching. Can you run a Python script to do entity resolution to try to do matching? Yes. You can.

Can you explain the behavior of that? And is the behavior of that easily configurable? Is it configured? Like, I can turn a dial, or or is it code? And I need and I need to like, I have to go through an SDLC process to make a change to it.

Two other things I would consider to your point, Raju, about can you? Yeah. You you might be able to. You you might be able to do this.

Should you? I think it’s it’s the question that you and I are both really kind of hammering on here. So really, really appreciate the insights on that. That’s interesting.

Sure. Yeah. Yeah. Yeah. No. You know, you hit on all the right points there. Right?

As soon as the complexity comes in, the multi domain, multi hierarchy, it’s gonna be a nightmare to even think of systems like Databricks or Snowflake and manage that. There’s no way. And and not much has changed. Even now, if you really want to put the data quality rules, you would have to do what you did two decades ago is build an ETL with those rules embedded in it.

Right? And the moment you need to change any of that, you have to go through the full SDLC cycle. The business doesn’t get a chance to come by themselves and tweak if they need to tweak a rule. Right?

So, yeah, all the good points. Absolutely. This you know, this platform is not built for MDM clearly.

And if you have following Databricks or Snowflake, actually, the Databricks is having the summit annual summit right now in this Right now as we record.

That’s correct. Yep. Yeah.

And and and and they they they have never advertised or they have never come out as an MDM platform.

You know? Can you build to back to your basic fundamentals? Probably yes. But should you is is the biggest question that everybody should ask.

Yeah. Alright. On that, that’s a great spot to end. Hey. Two million. Right? This guy’s I I I mean, there’s there’s a lot more to look forward to here.

I know you guys are working on additional stuff. I know you’re looking at expanding some of the domains here. I’m I’m just, like, beyond thrilled we could actually tell a story that MDM is driving two million in revenue for a company as awesome as Lexmarkk. So congratulations on your success, Raju.

Thank you again so much for for joining me on the stage at Gartner. That was a lot of fun. And thanks for coming out and spending time with us today to share your wisdom with the SetoMetas community.

Thank you. Privilege is mine. Thank you for having me here. I look forward to many more talks like this.

Awesome. Alright. If you’ve made it this long, thank you for listening. Thank you for watching.

Please take a moment to consider subscribing. If you like the content that we’re doing, I would love to have you join the broader community. We do this every two weeks. We also do a monthly live event, the third Friday of every month on LinkedIn.

I jump live, you can ask me questions about anything. MDM, data quality, data strategy. There’s nothing off limits. We’d love to have you join the community.

Thanks again for joining today, and we will see you on another episode of CDO Matters sometime very soon. Bye for now.