---
title: "Highlights: Implementing a modern Data Governance strategy"
url: "https://atlan.com/implementing-a-modern-data-governance-strategy/"
description: "SMG, Elastic, Autodesk and Fox on implementing data governance: starting points, ownership models, and what led each team to a modern data catalog. A Re:Govern 2023 panel."
---

> Atlan is hosting Context Conference, bringing together the leaders and builders at the frontier of giving AI the context it needs to understand their business. It runs online on October 28, 2026, from 11:00 AM to 2:00 PM ET. Atlan co-founder Prukalpa Sankar opens and closes the day. Leaders from AstraZeneca, BNY and Verizon share why they invest in context and what they get from it. Registrants get early access to The AI Context Gap, a new study from MIT Technology Review Insights. Register: https://atlan.com/context-conference/

# Implementing a modern data governance strategy

Videos / On demand / Recorded October 20, 2023 / 40:53 / 7 chapters

Watch: https://atlan.com/implementing-a-modern-data-governance-strategy/

A Re:Govern 2023 panel: SMG, Elastic, Autodesk and Fox on where their governance programs started, how they define owners, and what pushed each team to a data catalog.

## What you'll see

### 01 / Four starting points

Watch 0:00-13:44: https://atlan.com/implementing-a-modern-data-governance-strategy/?t=0

Amy on a merger that split SMG into business units with different maturity, Takashi on building trust in Elastic's core metrics, Mark on Autodesk's platform team serving 60 data teams, and Mihir on Fox starting with its top 10 assets.

### 02 / Who owns what

Watch 13:44-26:29: https://atlan.com/implementing-a-modern-data-governance-strategy/?t=824

Data product owner roles defined before anyone held the title, SMG's four kinds of owner and a gamified rollout, Fox's README template for every KPI, and Takashi's rule that the owner is whoever can change the process behind the data.

### 03 / Why a catalog, and why Atlan

Watch 26:29-40:53: https://atlan.com/implementing-a-modern-data-governance-strategy/?t=1589

Fox on lineage and stale dashboards, Autodesk on an open source catalog it could not extend, SMG on dbt docs business users could not read, and Elastic on positioning Atlan differently for each audience.

## Speakers

- Austin Kronz, Director of Data Strategy, Atlan (host)
- Amy Raygada, Senior Data Product Owner, SMG Swiss Marketplace Group
- Takashi Ueki, Director of Enterprise Data & Analytics, Elastic
- Mark Kidwell, Chief Data Architect, Autodesk
- Mihir Modi, Data Governance Manager, Fox

## Chapters

1. [0:00 Where SMG, Elastic and Autodesk started](https://atlan.com/implementing-a-modern-data-governance-strategy/?t=0)
2. [10:25 Fox's incremental rollout of the catalog](https://atlan.com/implementing-a-modern-data-governance-strategy/?t=625)
3. [13:44 Defining roles and ownership at Autodesk and SMG](https://atlan.com/implementing-a-modern-data-governance-strategy/?t=824)
4. [21:26 Domain owners and KPI templates at Fox](https://atlan.com/implementing-a-modern-data-governance-strategy/?t=1286)
5. [23:58 Who owns the data? Whoever can change the process](https://atlan.com/implementing-a-modern-data-governance-strategy/?t=1438)
6. [26:29 When a team needs a data catalog](https://atlan.com/implementing-a-modern-data-governance-strategy/?t=1589)
7. [31:12 Choosing Atlan: dbt, UX and one platform](https://atlan.com/implementing-a-modern-data-governance-strategy/?t=1872)

## Related videos

- [How data governance works at SMG Swiss Marketplace Group](https://atlan.com/how-data-governance-works-at-smg/) (3 min)
- [How data governance works at FOX](https://atlan.com/how-data-governance-works-at-fox/) (3 min)
- [Context Agents demo](https://atlan.com/context-agents-demo/) (13 min)
- [What is a context layer for AI systems?](https://atlan.com/know/enterprise-context-layer/videos/what-is-a-context-layer/) (10 min)

## Transcript

### 0:00 Where SMG, Elastic and Autodesk started

**Austin Kronz**

[0:06] Please welcome Amy Raygada from Swiss Marketplace Group, Mihir Modi from Fox, Takashi Ueki from Elastic, and Mark Kidwell from Autodesk. Thank you all for joining. Amy, we'll start with you. I want you to tell me about you and your team, but also where is your organization right now when it comes to data governance? You know, did you just start last month? Have you been there for years but you're in a new team? Or, you know, what has been this genesis of data governance and, and how has your organization gone down this path?

**Amy Raygada**

[0:37] Hey, Austin. Thanks a lot for having me today. Yeah, so let's go to that. It's been actually quite interesting. Well, first, I'm a senior data product owner from the enablement team, data platform enablement team, but before that, actually, when I arrived to the company two years ago, we were just a central team with different business units, and we were basically chair resources for each of the business units. There was no governance, data quality was not a priority. We were just doing whatever business was asking for. It was a little bit of a mess.

[1:11] Then we went through a merger because the company absorbed one the other company, and basically I joined when we were dividing these into these separate business units, each ones with their own separate data team, and us, the central team that, in theory, you know, will be, like, giving all the best practices, right? Like, all the important things that the, the... In general the company should, you know, follow. However, this was more complicated than expected because the maturity level of each of the business unit was very different. And also, each business unit, as I mentioned, they have their own data director, so they have their own vision, their own ways of working.

[1:53] And of course the central team trying to make them work in the same data platforms, for example, or same structure. It was not an option, you know. So we got a lot of resistance out of it. We started thinking data governance more, more with the data mesh concept at the very beginning. And we started to implement this last September, actually, from one year ago approximately. We hire an architect. In May last year we started to think about what kind of use cases we can use, you know, in order to be able to implement all this with our architect.

[2:29] We were able to have quickly, you know, all the data platform, all the access control side monitoring and, you know, these kind of topics that you actually need. And of course implementing, you know, all, all these other tools that probably we'll speak further. So now I think it's much better one year and a half later, but it's still a lot of gaps. Right now we are implementing data owners.

[2:54] We want actually the people to own it, to be able to understand that it's not only responsibility of the data team to take care of the data, of the quality, and how you dispose out of it as well, you know, regarding data privacy and all these topics. But also, you know, it's a hard topic because not everyone wants to be responsible from this.

[3:14] So I will say it's getting better, and now we got support from the C-level, which is something very important to start your data governance journey, I will say, because we started, you know, bottom up instead of top bottom, just to show something, a business case that we actually get supported by the top level. And I think now things are gonna move a bit quicker, but we move from nothing to have some owners, to have Atlan as a data catalog, to have a modern data stack and data platform, and eventually we would like to unify all the data platforms that we have because, as I mentioned, they are different.

[3:48] But yeah, it's a tough journey, but we have done quite of a big progress by now.

**Austin Kronz**

[3:54] Yeah. That's awesome. Thank you for that. I, I think the data mesh paradigm has brought about a lot of what you talked about, where it's almost centralizing in preparation to decentralize, right? We kind of had to get best practices together in order for these decentralized teams to function in, you know, a scalable way. So thank you for that. Takashi, I'll go to you next. Same kind of question. Where is Yeah, thanks. Let's do that.

**Takashi Ueki**

[4:19] Yeah. So, yeah, my name is Takashi, yeah. So I head up our enterprise data team and I'll kind of answer this kind of jointly. So when I joined a little over two years ago, this kind of concept of, like, an enterprise data team didn't exist.

[4:31] And, you know, I was kind of brought in to kind of help stand this up with a lot of the problems that kinda Amy was referring to, like siloed analytic teams in different functions, you know, no trust in kind of a lot of our core, or lack of trust in a lot of our core data assets, not a single source of truth for a lot of our data, and kind of teams just kind of, you know, building their own code, building their own, defining their own metrics and reports.

[4:53] And then, you know, you have C-suite level folks saying like, "Hey, I'm getting an answer over here from finance," and sales is saying something, you know, completely different, right? So our goal as a team is, you know, kind of three key pillars in terms of making sure that we're aligned in terms of how we are defining and the policies around our metrics, our hierarchies, and our, the attributes associated with our data. You know, driving our data governance program, really ensuring that we have the right ownership and accountability and stewardship within the business and on, on our technology side. And then enabling and delivering on our enterprise reporting and analytics.

[5:27] So, you know, kind of as I mentioned, you know, when we first started, a lot of, you know, lack of trust in our, in what we were reporting. So it's like what we've done over the last year and a half is build a lot of the foundational components that are needed from a governance perspective, and started to kind of pull some of those key reporting and data assets back into kind of a central enterprise level, so that, and kind of building a framework of saying, "If we are publishing it from our team and considering it to be certified, here's what that means," that you can have a certain level of trust in it.

[5:58] And then working with our functional analytic teams to say like, "Hey, if you're gonna be doing anything in this space, at least point to the same source of truth," right? So we can at least know where you're starting from at a bare minimum. They're gonna spin off and do their own, you know, reporting based off of that, but at least it kind of, you know, fixes one part of the problem of we're pointing to the same place. And that has kind of helped improve a lot of the trust in our data.

[6:21] And I think, you know, I mean, where we're going from here is, okay, well, we still have a lot of, you know, acute data quality problems. And, you know, as Amy mentioned, it's like, okay, well now our focus needs to be on how do we drive the right ownership and accountability within the business, and making sure that they understand that they are ultimately the owners of the data, and that they are responsible for it. And that means driving the right changes from a process and policy and system configuration perspective to enable the quality of data, you know, downstream.

[6:50] So that's kind of, you know, what our focus has been in making sure that we have those right agreements with our business partners.

**Austin Kronz**

[6:56] Awesome. Yeah, I think we'll touch on what you're referring to, which is it's an inevitability that people are gonna use data. Can we at least agree on what you're using, where it came from - Yeah... Who could we talk to when you use it? So I see this a lot, is it's more of a two-way street when it comes to trust. You know, it's not just the end user that wanna trust the data team, that wants to trust. Mark, we'll go to you next. What's the latest at Autodesk? Where have you been the last year on your data governance journey?

**Mark Kidwell**

[7:24] So I'm the architect for Autodesk Analytics Data Platform, which is our, our large scale platform serving enterprise and product usage, and a lot of other different kinds of use cases for, for Autodesk data sets. In terms of scale, we've got a probably 60 or more data teams, hundreds of use cases, thousands of users using our platform now.

[7:43] We've always approached our problems from the platform side, and I, and I think even though we've had a lot of the similar journey as, as I've heard from, from folks already that they've told about their journey, you know, we, we started off with a, a mandate to be both a platform team and an enterprise data team, and do a lot of all of these things. And as we've evolved, we've become more of a focused platform team providing platform level capabilities, providing services, providing tools, and standards for all of our users. And, you know, our team is...

[8:18] As we've evolved from that, it's, it coincided with a movement from us doing everything to people taking on some of the ingestion processes, and publishing processes, and self-service capabilities, you know, that, that we've provided or enabled them with. They're responsible for that. And that theme of trust is a, is a big one. We, we didn't wanna let go in some cases, and we weren't really sure what was gonna happen. So, you know, the, a lot of the last couple years has been, well, if, if we're gonna trust the data sets that other teams are bringing into the platform and publishing on their own, what do we have to do to, to put...

[8:54] What do we have to put in place to make sure that they're successful and that these data remain trustworthy? Who do we have to partner with? So, you know, like, like everyone, it's a combination of different processes, getting the right people, getting the right technology in place. But I think that's, you know, the last two years since we started our official kickoff of data mesh, even though we, we were probably hybrid before that, I, I think we've made a, a huge transformation in, in the way we approach data governance and, you know, providing all the tooling and, and making sure that, you know, people understand their roles, responsibilities. Like everybody else, though, it's a journey.

[9:31] We're never done with the journey. I think right now we've got all the greatest tools we can have. We've got a modern data stack. We've going through a huge modernization effort, but now we've just got the next challenges ahead of us. You know, how do we scale up to even more people, more self-service better insights, and of course, more data? There's, there's a lot of data sets that we still don't bring into the platform and, and we wanna bring those in next.

**Austin Kronz**

[9:56] Awesome. Yeah. Mark, you and I had a chance to go to the, the Snowflake Summit and talk a little bit about data mesh. And I remember one of the things you were telling me was your team is almost going from how does our data engineering group be the best data engineering group, to how do we be the best data engineers to enable other data engineers, right?

**Mark Kidwell**

[10:15] Yeah.

**Austin Kronz**

[10:15] So it's a little bit more of an outward-looking vision that you're approaching now. So correct me if I'm wrong, but I, I think that's sort of how you phrased it, but.

**Mark Kidwell**

[10:23] That's right. Yeah.

### 10:25 Fox's incremental rollout of the catalog

**Austin Kronz**

[10:25] Yeah. And we'll double-click on people a little bit, but Mihir, let's go to you, and same intro question here.

**Mihir Modi**

[10:30] I've been in Fox a little less than two years, but, you know, spent into data industry and media industry over more than 10 years. Started my journey as a data engineer, kind of, you know, gradually progressing towards a data modeler, data architect, and then kind of, you know, landed into Fox, wherein I was kind of leading the data quality, you know in charge of that, basically. So I'm part of the business intelligence and the analytics strategy team.

[10:53] We started a journey, like little around, like 10 months before, we thought like, "Hey," you know, as other panel kind of, you know, mentioned, that we have data silo problem, nobody trusts your data, different data teams across Fox, you know, playing around with their own data sets, and difficult to find the data. So that's where we kind of, you know, thought to lay down a data strategy, and governance being the core component of it, basically. Right?

[11:15] So what we did was basically we started dividing our data into different logical domains, and then we thought of, you know, building a semantic layer on top of it, aligning with the data mesh principle, which we all kind of, you know, are embarked upon, basically. Right? So that was kind of, you know, a key start, and I would say where we are at right now is basically, so that kind of, you know, prompted us to come up with the modern data catalog. That's where we started our journey with Atlan, searching around, hey, which is the best tool for us, which will work both as a governance platform and, you know, primarily as a catalog.

[11:48] We started our journey, we started evaluating and, you know, we were looking for something different, right? At Fox, I would say there's always a culture wherein you are asked to challenge yourself and do something different, right? So it's the, the governance journey and strategy which we have laid out is not a typical journey. Like, you have a data council, you have, you know, a board of members, executive stakeholders from different business teams, then kind of, you know, implement. No, that wasn't the idea. The idea was to be nimble. The idea was to take an incremental approach and to go, you know slowly across these domains, right?

[12:25] Where we are at, like before six months, we launched, started launching our data catalog to different, you know, you know, our data team. That was our first kind of getting a feedback, "Hey, here are the top 10, you know, kind of data assets into the catalog, which are most curated assets, and which will answer your top 10 to 20 business questions," right? That's how we started kind of, you know, slow.

[12:46] We exposed the catalog to the data teams, and now we are on a journey wherein we are, you know, having sessions, having office hour sessions with different smaller business teams, targeting each individual business teams, explaining them that, hey, what's the importance of governance, how this catalog can play a role, how can a producer and consumer, you know, personas kind of align with each other, right? That's how we have taken an incremental approach and going targeting small teams, and then the plan is to kind of eventually launch catalog across Fox. I would say so far the response has been fantastic.

[13:19] People are now kind of, you know, interested that, "Hey, here is the place where I can kind of look for, I can search, I can kind of discover, I, I can, you know, go to my BI dashboard." So they're kind of excited, and yeah, that's the, you know, and we have to kind of, as, you know, Mark and other panelists kind of mentioned, it's a journey, it's a learning basically, so we're kind of gradually kind of always evolving basically and working around that modern data stack.

### 13:44 Defining roles and ownership at Autodesk and SMG

**Austin Kronz**

[13:44] Yeah, it's great. I think, you know, I heard someone say it best. It was another customer that sort of said, "Data governance is almost this high-level umbrella term that means so many different things." And with something like Atlan, I think their quote was, "Atlan sort of gives a face to data governance." And it's very easy to say, "This is our BI tool. This is where we build charts and graphs. This is our data warehouse. This is where we store and analyze data." Right? But I think data governance has just always been, you know, it was a lot of process, right? We're gonna talk a little bit about process. But I wanna start with people.

[14:13] There's already a ton of questions in the chat asking about ownership. So Mark, maybe I'll start with you on this one. But what are those roles when you think of data governance and, you know, maybe roles isn't even the right word. Maybe it's just functions, 'cause I'm sure people are wearing multiple hats all the time, but, you know, what are those data governance roles, and then how does your team interact with engineers and analysts and business users? Like when it comes to ownership, how does that relate?

**Mark Kidwell**

[14:43] Yeah, that's a, that's a really good question because the, the kind of roles that we see or want on paper are often very different from the kind that our data teams that we work with might have or their engi - their actual business titles or things like that. And, and even internally, we've, we've evolved our thinking on this. You know, we, we came from a background where a lot of us were just engineers. You know, we were building this data platform. We were engineering this platform for other engineering teams to, to use.

[15:08] And the thinking of, well, if we're gonna enable other people to adopt the platform to start building on it in the way that we want to, then we had to, to more fully flesh out the roles that seem commonplace now or, and widespread use, but the ones around data mesh. You know, things like data product owner or data product manager. We had to more formally define those. And you're right when you say that people wear many hats. We, we didn't expect to be able to start working with data teams and find a person who had a business title or role called data product owner.

[15:38] We, we had to first explain to them what that meant, explain to them what their responsibilities were. But in stepping back to even before that, we had to agree internally on what that meant and what we wanted people to do, then go to them and work with them, sometimes iteratively. And, and, you know, it's something that's evolved over time. I think we've successfully rolled out our definitions of these different roles, what we expect from teams in terms of a, a data team that we work with and their staffing, and there's, there has to be data engineers or somebody has to be playing the architect role, the product manager, or, or data product owner role.

[16:14] And then what their responsibilities are in terms of setting up contracts with their consumers, the, in different personas. We're, we're really focused on the teams producing data, but really we're just doing this all so consumers can find it and use it and build insights off of it. What are the roles they have? How do they work with the, these teams? Again, it's been a journey. But, but I think we have a, a really good pattern for that, and now it, it is just continuing to repeat that, that evangelism and make sure other folks that we work with understand that.

**Austin Kronz**

[16:46] Awesome. Yeah, I think I was hoping someone would bring up the, I'm sure there's a lot of conversations that end with, "Oh yeah, I guess I am kind of that. You know, I guess I am kind of a data product owner. Maybe I am a data steward after all. I didn't know that's what that was." So I, I would imagine many of those conversations exist as data teams are kind of expanding and, and building out this journey. But Amy, same kind of question. You know, how are you interacting with different groups?

[17:09] Is everyone under one roof, so to speak, one team, or is it, you know, you're kind of a hub and there's engineers here, analysts all through the business? What, what does that look like, and how do you manage ownership?

**Amy Raygada**

[17:22] Since we started with the data mesh topic on the data governance journey, we started with a data platform/enablement team. So we have data engineers, BI developers, data product owners. And then the idea was one of the, well, in my case, I was the person in charge to train other data product owners and other teams to, for them to understand how to create domains from the different, you know, business units that we have. However, as Mark said, like, we have to agree on what will be the responsibilities, how they're gonna work, how they're gonna, you know, get these new responsibilities on board.

[17:56] And in our case, it was a bit challenging, for example, when we told product managers that they will become data product owners. It was frightening for them. They thought they will need to be, like, expert in Tableau or, I don't know, learn how to query, how to create their own data platform. It was insane, so we got a lot of resistance out of it. And still, little by little, you know, they were, like, put through these domains in a way, and they are learning because they couldn't understand how to work with a data team in the sense, "What can I ask to my engineers," or, "How can I collaborate with them?"

[18:28] Stay with the owners because we have, you know, a stream line or consumer line domains, and we have to explain to people from the backend that they will be, you know, technical owners, and they also need to own, you know, these datasets, well, these APIs that will become a curated dataset at some point. And then a data engineer will have ownership over that dataset as well. So when you go downstream, you know, to, in, in order to be able to implement all the data contracts, so we needed these kind of technical owners as well.

[18:58] And then we need the business owner, who is the person who's gonna own this data product per se when it's finished, and they will have to provide us, for example, with test cases to be able to do different kind of anomaly detection or something more related to business or how to create the right metrics for them at some point, depending on the domain, of course, they are working. So after seeing the maturity level of the different business units, this is something that we are currently actually implementing because it was a little bit disturbing.

[19:27] We, we have to start, stop calling it like data mesh, you know, and name it just data governance framework because otherwise people will freaked out a little bit, but pretty much it's the same with a different name at this point. So we decided just to simplify a little bit the owners from our perspective. So we have a data governance council now. They were able to approve the framework that I created about data owners, and we have a base with a source owner for the backend, technical owner for the pipeline owner, then domain owner, I mean, depending on the domain that they are being part of, being this, the data product owner in this case.

[20:04] And then we have the metric definition owner or the, or the, the person who owns the business definitions and, and all the business side of, of this topic. For some of the business units, they decided just to keep two owners to keep it simple based on their maturity level, so they have only business data owner and one technical owner. But some others already are even having more than the four that we gave as an option. So they have someone that is like, I don't know, like dashboard or analytics owner and some other owners.

[20:34] And I think the key for data mesh and to be decentralized, you know, and, and give all this is to give this freedom for them to play around with the base rules that we give. So, and also the training, right? Because, for example, for Atlan, I develop this game. So as like a, a gamification so they can have champions and the owners feel like they own it.

[20:57] So this is something that we're also gonna start rolling out in the next months with different datasets specified for each of the business units, and they can actually, you know, relate to their own data and understand, and then they will have, you know, like goodies, for example, like who's the top 10 owner, who's the one who updates the most their datasets. And this how we're trying to foster this ownership and these roles without telling them exactly like, "Oh, now look, this is your hat now. You have to wear it," and that's all. So yeah.

### 21:26 Domain owners and KPI templates at Fox

**Austin Kronz**

[21:26] It's amazing what a, a leaderboard, a little bit of free swags - Mm-hmm... And some motivation will do for a team. Yeah. I'll come to you next on this. I think the topic of ownership is, is, is always interesting. It seems like Amy has four types of owners. We see this a lot in Atlan, business, technical, maybe a steward, like varying tiers of ownership. So what, what about you? How do you interact with different stakeholders?

**Mihir Modi**

[21:49] Yeah. So first of all, starting with like, you know, the core team is like, there's a data engineering team, there's a data domain team, and the BI analytics, and then the data governance, right? So there's inherent, you know, coordination between all of these teams, and governance is like kind of, you know, not kind of, you know, a typical approach that, hey, you have to follow these. You have to kind of, you know, follow these. No, that's not the idea.

[22:09] It's kind of part of the process, and we have laid down our standards, processes such a way that it's kind of, you know, inherent across, right, from building the domains up to the semantic layer on the top, which will answer your business questions, right? So in terms of ownership, what we have done is, as I was explaining earlier, we have divided our data into different logical domains, and each domain has a kind of a data domain owner, right? He or she is primarily responsible for, you know, governing that dataset that, hey, how do I kind of, you know, update the metadata?

[22:41] I need to ensure, you know, all my KPI and terms are kind of linked to these assets, right? And one of the cool thing which we have done is we have standardized a README template across each KPI and, you know, business metric, which will answer your top 10 kind of, you know, business question. What does that KPI mean to the business? We have kind of partnered with business first to kind of understand, hey, what is the common definition? Put it in the catalog so that it's easily accessible for all of them. And that template kind of, you know, defines in a hierarchical way, hey, here is my KPI term.

[23:14] These are the business questions which it answers. This is my calculation logic, and, you know, you are free to reach out for additional questions on this topic, right? So that has kind of helped us out in kind of, you know, having a collaborative approach for data governance, and it's kind of not, you know, feeling as a pain, right? Rather than like, you know, everybody's kind of excited to jump on and then contribute toward those, right? So the domain owners is primarily responsible for curating and having that metadata available in the catalog as well as, you know, we are partnering with them.

[23:44] So me being leading the governance team, ensuring that there are right data quality rules set up, giving them guidance, setting up the processes, and then kind of, you know, let them work on their own way, right? That that's how we have kind of, you know, done.

### 23:58 Who owns the data? Whoever can change the process

**Austin Kronz**

[23:58] Yeah, that's great. I, and I, it's probably a perfect transition to Takashi. I want to get your thoughts on this too because, you know, Mihir's kind of referring to this, hey, let's templatize as much as we possibly can. So coming back to your very first point of we want to let them do what they need to do, but we need to trust that it's in some format that we agree on. So same question to you about ownership. I mean, what what does that look like in Elastic?

**Takashi Ueki**

[24:22] Yeah, I mean, I, I think a lot of what, you know, the rest of the folks said, you know, you know, similar to what we have at Elastic. I kind of wanted to focus on, and there's actually a question in the, the chat too that kind of gets to this point about ownership, because, like, when you talk about - Who the owner of from a technical or technology side or from a systems side, it's pretty straightforward. If it's a data product itself, it's like, okay, well, it's who's building that product, right? But then it's this fuzzy gray area of like, okay, well, who's actually the business owner of that data, right?

[24:50] And it, it's kind of hard to say, like, well, who is it? And I kind of think of it in a way, and this is how we, you know, think through it is, 'cause it ultimately comes back to like, oh, who's the owner of the data. The question always arises when there's issues with the data, right? And they say, "Well, the data, you know, is wrong. The, we have, you know, poor data quality." And what I like to remind people is data in it of itself is not bad. It's just a reflection of a process, a policy, or a system.

[25:14] So when you think about, like, who the owner of that data is, from a business perspective is, you have to figure out, okay, well, who can actually enact change on the policy or the process or the system? So, for example, like, if it's a, on an opportunity within Salesforce, we're getting really poor data quality on one of the fields that our Salesforce or, you know, our field is supposed to be, you know, entering in. Well then, who can actually, you know, change the process, the policy, or the system configuration to ensure that the quality of that field is going to improve? Sales ops. Okay. So who within sales ops can actually influence? That's my owner.

[25:52] 'Cause if it's someone other than that, then it's like, well, I can't even say, like, well, this is the owner of that data, and say, "Well, I have this problem." Well, it's a process problem. I don't own the process. Well, then you're not really the owner of it. So that's kind of how we think through, like, from a business ownership, and I think that's the, the most important kind of role to kind of define and make sure that we have clearly, clearly identified, 'cause it's really those are the individuals that are gonna be able to influence proper governance at source.

**Austin Kronz**

[26:19] I really like that. They have to have skin in the game Yeah... To care, right? You know, it can't just Yeah... You're the you're the owner now. You know, you're, you're in that business unit, you're the owner.

### 26:29 When a team needs a data catalog

**Takashi Ueki**

[26:29] Yeah.

**Austin Kronz**

[26:29] You know, I think it's hard to enforce policy or enforce adoption of policy in that kind of framework, so I really like that. I wanna switch gears a little bit. We're talking a lot about, we'll come back to process, but I think inevitably technology is in, is involved in this. You know, I think you're all Atlan customers, so I'm curious, what is that pain point that arises where you think, "Okay, we, we need a data catalog now"? And, and I would imagine there's some element in pockets of the organization. There, you know, people are tracking their own glossaries or dictionaries. Developers are tracking their own schemas.

[27:03] But, you know, it, it, what is that tipping point where you say, "We need to go look at catalogs"? And then how do you end up at, you know, more modern solutions, or what was your impression, I guess, when you were looking at something like Atlan? Mihir, and I'll quickly start with you.

**Mihir Modi**

[27:18] Sure. I think this is my, one of my favorite question, actually. So, so inherently, we all know why we need a catalog. Like, you know, typical challenges, data access issues, who is the owner, whom do I reach out to kind of, you know, ask more questions, how do I get access to data? Those are, like, kind of basic, you know, challenges which every organization faces. When it comes to a modern data catalog, I think the key differentiator I would say is, you know, if let's say a business user, right, he goes to a Tableau or a Looker dashboard and he doesn't know why my dashboard is not, you know, reflecting the latest numbers, right?

[27:50] That is a signal that, hey, underneath some pipeline has not refreshed my latest data, and I'm not able to convey that message to a business user before he finds out that, hey, there is still data, right? Or let's say, you know, it, it's kind of difficult to understand how my data from raw to kind of target is kind of flowing into, right? So lineage problem. The third thing I would say is most kind of pain point is basically, you know, how do you leverage that metadata, basically, right?

[28:19] We are all, are into a world wherein there are large of data sets growing each day by day, and I would say metadata is the new gold, right? How do we activate that metadata? If we are aren't able to kind of utilize that metadata to, on real-time basis and let our customers, let our business users know that, hey, this was the data quality problem, we are kind of fixing it, these are my automation, you know, pipelines which are running behind the scenes. If those are kind of, you know, coming, keeping on continuously coming in your organization, that's where you definitely need a modern data catalog, I would say.

**Austin Kronz**

[28:53] Makes sense. Mark, I'll kind of position the, the same question. I mean, what are the signals that you need something? And, you know, how does that become the data catalog and, you know, I guess, what did you see in Atlan and think that you thought this could solve that problem right now?

**Mark Kidwell**

[29:13] Yeah. I, I think in, in our case they, they weren't so much signals as alarm bells blaring. We were in the search of a data catalog and had been looking and trying to find a replacement for this homegrown catalog for the better part of, of several years. And it just became more and more acute, where we couldn't find the right fit with what we were looking for, and looking at either other vendors or open source solutions. We actually tried some. We went through an entire deployment of an open source version.

[29:43] Everything, you know, that we did was incrementally better, but the demand for the kinds of things that a data catalog solved for, you know, people were asking questions, really basic things all the time, and to some extent they still do, on our Slack channels, you know, around ownership or how these data work. And it, it wasn't sustainable at the scale where we were already at several years ago. It certainly wasn't sustainable, you know, as we continued to grow and grow. So, so for us, you know, we knew it had to be done, it was just finding that right fit.

[30:15] And even though we invested in building out an open source solution, almost as soon as it, as we completed it, we saw there were a lot of things that we just couldn't keep up with. We couldn't extend it ourselves. There's no way we were gonna keep up with demand for this. So we just started the search over again. We had no problem going back and saying, "You know, this - Didn't, wasn't the, it's a big step forward, but it wasn't the best thing we can do. What's the best thing we can do? And we, we ran a, an evaluation process. We, we looked at the, the leaders in the space.

[30:45] Atlan, it, you know, met all of our criteria and then a lot more. You know, there's a lot of things that we're working with our stakeholders internally or, and trying to find new ways to use Atlan and the catalog to connect our data publishers and our data consumers, different data types, enabling different kinds of workflows, all of these things that we're planning. The catalog is core to that. It's, it's the center of, of bringing people together and finding data on our platform.

### 31:12 Choosing Atlan: dbt, UX and one platform

**Austin Kronz**

[31:12] Yeah, that's great. I think, Mihir, you also touched on this in the beginning of, of the conversation, but there's almost a, a connotation with cataloging. That's why you see active metadata platform a lot around, you know, the Atlan material. And it, and I think it's because collecting metadata is one component, right? You wanna be able to index, to discover, and find your data assets. But I think what a modern data catalog, especially what Atlan strives to deliver, is you shouldn't just be able to collect it and have it sit there.

[31:40] You wanna be able to utilize it, push it into the BI tool, push it into your processes, automate workflows, you know, and all of this, it comes by unlocking, I think, that metadata. So, you know, the first use case might be discovery, but then, Mark, I think what you're alluding to is there's a lot more use cases for metadata than just index and search. And I think a lot of that saves time. But Amy, I'll ask you this same kind of question, right?

[32:04] What are the pain points that you start to find in a data governance, let's say, maybe even pre-program, but and then how do you kind of get to this data catalog solution in Atlan?

**Amy Raygada**

[32:16] It was quite interesting actually, because when I joined this team and the merger, you know, and I ended up in the central team, we were already using dbt, and we were using these dbt docs. But when I spoke with my boss, I told him, "Look, I don't think this is really gonna work because business people cannot really access these dbt, you know, docs, and it's not really easy for business people to come here, here and understand. And also the definitions that are here are very technical, are not really, you know, very business-oriented." So I started to do my POC.

[32:49] I was starting to, you know, get some tools, and then someone told me that someone else in the company was doing the same thing. So they were doing evaluations of other tools as well from another business unit, and then I have to reach out to these people, and we kind of, you know, merge our forces to be able to find the right tool. So we went, for example, to open the data, but it was a little bit too technical as well.

[33:13] We also did Alation, we did Atlan, and at the end we decided to do it with Atlan because it was more easy of, of the integration with dbt first, because we already had all these metadata. Everything was very well-defined on dbt, so this integration will allow us to bring a lot of things automated for us. And also because the, it was very easy to use for all the business people because it's more readable, it's legible.

[33:38] You can have, the owners can edit, for example, the README side of things without actually to fill out all the other stuff manually, because we will bring everything automated using the, you know, custom metadata features or even the metadata that is already, you know, basic for Atlan. So I think in our case, we saw the need ahead because to be honest, the adoption right now is not that big based on all the changes we're doing with ownership and privacy and all these you know, other topics for data governance. But definitely it's something that for some teams that they have a higher adoption rate is helping a lot. So they can discover more things.

[34:19] They can they are checking also on the Looker dashboards from there already, checking dbt, dbt tests and, and we have some scoring now that we're gonna implement as well to understand if this asset is certified or not, so they can... The, for example, the, the turnaround time for a data analyst, let's say, to search on a data asset, it will be three days to make sure it's, you know, curated, clean, with tests and everything. But with these scores, we're gonna be able to save time for this person just to come, explore, see that it's certified, they can take it and use it immediately to develop a dashboard.

[34:55] So I think this is the kind of things that made us go through Atlan and to understand the need of this, because it makes the exploration of data much easier. And if you combine it with your own data platform, because we do a lot of things in GPP like PII data and also some metadata there as well it's powerful when you combine you know, different tools to make it work for the business and, and for the analysts that will need this information.

**Austin Kronz**

[35:22] That's awesome. Yeah, I, I think integrating with your, your stack and making it easy to use, I think one thing you touch on too is multiple personas in one place, right? I think that's also probably a legacy connotation that data catalogs are for this one persona, where, you know, I think in Atlan we try to build this interface where business users can access it just as easily as a data engineer can access it. And we'll try to auto-populate the more technical metadata. We'll allow the business users to enrich the business metadata and the glossary, the terms. That, that's awesome. Really helpful. Thank you.

[35:52] Takashi, I think you, I know at Elastic you were kind of in the same boat of integrating a lot of tools to build a data governance platform. So I guess same kind of question for you. What, what was that pain point, and how did you end up figuring that one out?

**Takashi Ueki**

[36:04] Yeah, I think it's a lot of what, you know, the folks mentioned, and, and a lot of it is like, how do we centralize and increase transparency of, you know, what our definitions are, are and make sure that we're consistent. And I think one of the big things for us when we, you know, looked at Atlan was the UI and the UX experience of like, is this something that people are actually gonna want to go into and actually use, right? Is it easily navigable and, you know, are you able to find what you're looking for?

[36:31] And then I think the integration with a lot of, you know, as you mentioned, that's kind of built into Atlan, where we're able to see actually the full lineage at the column level from - You know, effectively from source, all the way from our source systems, all the way to our end data products, whether it's a dashboard or so on and so forth, right? So I think those are the key things of, like, being able to increase the transparency of, you know, our kind of data process. But I did want to kind of touch on, like, you know, and I see it in the chat, too, of, like, this concept of data cataloging.

[37:00] And when people ask me about Atlan, and I say it's a data catalog, then I kind of pause. I'm like, "Well, it's not just that. It does so much more." And then what we started to shift to is, like, positioning and kind of branding Atlan in different ways for who, which audience we're speaking to, right? So if you think about it like, okay, if I'm a, just a business person, right, and all I care about is, hey, I'm... I heard there's a bunch of Tableau dashboards out, out there that I need to be using. Okay, well, it's your catalog of dashboards, right? This is what you care about.

[37:32] You go here, this is how you find the information. You can see the owner, you can see the definitions and all that information. If you are an analytic engineer, it's like, hey, you can go here and find all the different, you know, tables and views and yada, yada in one place. You can find the definition. You can see, you know, which one is being most heavily used, which ones are verified, and all that information, right? If you're a data engineer we have a pipeline breakage. This is a way for you to kind of do that investigation to see it broke upstream, what happened downstream, what's impacted, right?

[38:01] So it's starting to think about Atlan of, in ways of, like, okay, well, how do we position it to the different users that we're trying to engage with so that it makes sense for their use case specifically?

**Austin Kronz**

[38:12] Yeah, I think that, that is great. There's a, there's an inter - I, I mean, I like that you're just bringing out the fact that there's an internal positioning to get adoption - Yeah... Of these things, right? You know, it's not just build it and they will come, right?

**Takashi Ueki**

[38:22] Yeah.

**Austin Kronz**

[38:22] So I, I really like that. Mihir, Yeah.

**Mihir Modi**

[38:25] I I would just like to add on, like, you know, why Atlan and why kind of, you know, we went ahead with this journey. So one of our key core component of strategy was we were thinking of, you know, a, a catalog solution as a one-stop data platform, right? So as Takashi mentioned, right, our goal was to, hey, you come here, you discover your asset, you understand more about it, and now if you want to understand more about business KPI terms, you go to the glossary, search, understand what that domain talks about, what are the high level KPIs, how is my business calculated.

[38:56] Once you understand a term, once you understand a data asset, you would like to know, okay, which are dashboards kind of linked to this? Just giving an example, let's say if I want to know impressions, you know, that are being delivered. I, I go to the Tableau dashboard right from the catalog. There's no need to kind of navigate differently to, you know, log into Looker or Tableau, right? I can embark that journey right from the catalog. The next step is basically one of which kind of, you know, most crucial factor is, like, now we do... Once you understand the data asset, you also want to play around with that data asset, right?

[39:27] So that's insights tab within Atlan which will kind of, you know, allow us to play around with your query. And what we did differently was we started publishing, you know, standard queries which business users can, you know, analysts can kind of refer to. That, hey, here are my top level filters, and here are my top 10 business questions which can get answered for your, from insights, basically, right? So we are not looking for a different SQL tool to go outside of the catalog, but we're branding, you know, our catalog as a one-stop data platform. And this, then the second thing is, you know, partnership, right?

[39:59] How do we kind of have an influence over the industry and shape the future of the data governance and the data catalog? I think those were the two key differentiators where, where we kind of decided to embark our journey with Atlan.

**Austin Kronz**

[40:12] Awesome. Yeah, that's great to hear. I know Atlan is, we're always putting a lot of content out there. We're trying to push the envelope when it comes to, to data data governance. But I mean, I have to give a huge thank you to, to this group because at the end of the day, you're the ones, you know, boots on the ground actually doing these things, and I think truly are the thought leaders when it comes to trying these things out, figuring out how to do the best data governance, adoption programs, policies, templates.

[40:36] And so with that, I just wanted to give a, a thank you to this panel, and I'm sure, if you haven't already, I'm sure everyone is connected with you on LinkedIn, but please reach out. It's a great group to work with. I, I know firsthand prepping for these things.