Data Mesh Isn’t for Everyone: It Requires the Right Mindset and Organizational Structure, says Petr Žáček, Ataccama

Ivana Karhanová: Who are data products for? I mean, which companies?

Petr Žáček: Data products can be used by any company, but adopting a data mesh might not be easy for many. Data products encourage organizations and teams to treat data not just as a resource but as a product they’re creating for others. It’s about the data as well as the information on its use, access, restrictions, and consents. Think of it like a physical product, complete with packaging and a manual, which together constitute the whole package delivered to the customer. So, the mindset of approaching data as a product is something I believe every company can adopt. It’s a healthy way to look at data, ensuring ownership is integral to data delivery.

Ivana Karhanová: You mentioned an important aspect: data ownership. Who owns the data in most companies now?

Petr Žáček: It varies, but unfortunately, in some cases, IT teams own the data for the business, which is an anti-pattern. Nowadays, it’s more in the hands of the business, with the emergence of dedicated data teams. However, full ownership by the business isn’t necessarily common due to mindset challenges.

Ivana Karhanová: If there’s a central data team, doesn’t that centralize decision-making and processes again?

Petr Žáček: Establishing a data team doesn’t solve everything on its own. It becomes interesting when data teams exist across different business functions, leading to collaboration and decentralization. However, companies still want control, leading to a balance between centralization for policy enforcement and decentralization for efficiency and agility. Data products, with established guidelines, allow for this balance by enabling decentralized management while maintaining some level of control.

Ivana Karhanová: Building and using data products requires additional skills, right?

Petr Žáček: Maybe, maybe not. It depends, as data products and data mesh are great concepts, but not everyone is adopting them fully. It’s about how organizations need to change, possibly reorganize, and what tools they need. Some organizations may already have the right mindset and just need to formalize it.

Ivana Karhanová: Can you mention some industries that might benefit from this approach?

Petr Žáček: Banking and financial services are ahead due to regulatory reasons. They have multiple functions that need to fit together, which align well with the concept of data products. Industries not traditionally focused on data, like retail and manufacturing, might find it more challenging.

Ivana Karhanová: Construction companies might be similar, dealing with physical products?

Petr Žáček: Exactly. Companies focused on the tangible world might struggle with adopting a data product mindset. However, there’s a way to establish this mindset even in these industries, starting with simple or well-defined use cases.

Ivana Karhanová: We should mention that Ataccama delivers an AI-powered master data management platform. Do your clients understand what a data product is about?

Petr Žáček: There are many assumptions and simplifications. It’s not necessarily wrong if they have their interpretation, as long as they’re thinking about data products. Moving towards creating something resembling a data product is a step forward.

Ivana Karhanová: If a company decides to work with master data management, what should they be aware of?

Petr Žáček: Besides master data management, our platform includes capabilities focusing on data quality, data cataloging, and business glossary lineage analysis, among others. Understanding data and establishing ownership are key starting points. Modern tools help with data discovery and analysis, making it easier to establish ownership and understand data usage.

Ivana Karhanová: So, all these capabilities should be part of the data product, right?

Petr Žáček: Exactly. Understanding the data, its ownership, and identifying gaps are crucial. Tools available today can significantly aid in this process, allowing for a more straightforward establishment of data products.

Ivana Karhanová: And when creating data products, do you focus on the report itself or the underlying dataset?

Petr Žáček: Initially, focusing on the report as the data product is a good starting point. It naturally leads to examining the data used in the report, who owns it, and if it’s another data product delivered by a different team. This approach helps map out the data landscape and establish ownership and trust.

Ivana Karhanová: Understanding where data comes from and ensuring its quality are essential for trust, right?

Petr Žáček: Exactly. AI might eventually simplify obtaining trust by summarizing the necessary information, but the foundation is understanding the data’s lineage, quality, and ownership.

Ivana Karhanová: How long does it take to build data products, especially starting in the Czech Republic?

Petr Žáček: It varies. Establishing the first data product might take months due to organizational challenges. However, once the process is established, creating additional data products could become much quicker, potentially just a matter of clicks with the right tools and processes in place.

Ivana Karhanová: And what’s the reality compared to the ideal scenario of quick data product creation?

Petr Žáček: The reality depends on the tooling and how well it’s integrated and managed within the company. Scattering of information across different tools can be a challenge, affecting the time required to create data products.

Ivana Karhanová: Do you believe data mesh is a sustainable concept in the long term?

Petr Žáček: It’s uncertain. While some analysts consider data mesh almost dead, the principles of treating data as a product and aiming for decentralization with policy enforcement are sustainable. Data mesh isn’t for everyone, and it adds complexity. Not achieving data mesh isn’t necessarily a bad thing if there’s no need for it.

Ivana Karhanová: Petr, thank you for sharing your insights.