From Data Mess to Data Mesh 

Print Friendly, PDF & Email

In order to generate value, it is becoming increasingly important for companies to utilize data as a strategic asset to support and automate decision-making across the value chain. However, data can only be helpful if it can be easily accessed. To ensure profitability and business success, both the producers and consumers of data need to cooperate seamlessly and transparently.

The data mesh movement has been underway in recent years with the goal of reducing interdependencies and enabling self-service business intelligence. Yet, organizations underestimate the cultural shift that is needed.

What is a data mesh?

Developed by Thoughtworks’ Zhamak Dehghani in 2019, data mesh is a platform architecture that leverages a domain-driven and self-serve design. As opposed to traditional monolithic data infrastructures that utilize a single central data lake for consumption, storage, transformation, and output, a data mesh supports distributed, domain-specific data consumers and looks at data as a product, with each domain managing its own data pipeline.

Dehghani states there are four principles of a modern distributed architecture:

  • Domain Oriented Ownership – Data ownership is separated into several functional domains that own the end-to-end solution.
  • Data is a Product Mindset – Data is atomized into individual products that are built and maintained.
  • Data Infrastructure as a Platform – Infrastructure components are abstracted into a self-serve platform to simplify entry and standardization.
  • Federated Computational Governance – Meta-data about data products is collected centrally to facilitate discovery and governance.

However, moving to a distributed data ownership takes organization, planning and buy-in from the right stakeholders, and we cannot underestimate the cultural change needed to accomplish this. Over the last 4 years I’ve been supporting organizations in this journey and here are my top 4 learnings: 

1. Data is king: Make everyone understand the business value of data.

Firstly, data can be a company’s most precious resource, so it is imperative to have access to and valuable insights into it. To unlock business value, everyone needs to be able to understand the data. Establish a solid Data Education Program amongst your workforce and make sure it’s  engaging, has a passive and an active strategy, and gamifies the experience… Data is fun!

2. Map the status quo: Understand your starting point.

The new data fads are often embraced by organizations before they have a clear understanding of their starting point. I know this sounds absurd: but you must create data in order to eliminate (unnecessary) data. To effectively execute a transformation, it is crucial to fully understand the pain points first. Jumping blindly into a transformation without having an understanding of the current data state within the organization will result in gaps in the overall delivery. As a first step, conduct a Maturity Assessment covering people, processes, technology, and data, not only to understand your starting point, but also to define your target state.

3. Speak the same language: Create clear definitions.

How often do we find ourselves arguing about metrics when in fact we might be using the same nomenclature, but with very different meanings? In order to ensure full transparency and achieve a successful data transformation, it is crucial to align metrics and methodologies.

4. Prioritize: Get the buy-in from all stakeholders.

    Organization’s domain leaders must understand what it means to own their own data and embrace the Data Mesh Principles in order to be successful. To ensure the proper support is given, you must obtain internal buy-in from the management team and across all domains. As part of your efforts, you will need to educate and incentivize stakeholders by demonstrating that data can drive business value. This can be done by collecting evidence from maturity assessments and highlighting the company’s return on investment.

    Finally, I would like to emphasize that the journey is not technical in nature, but rather cultural. Decentralizing data ownership will allow data to become a first-class citizen for both data producers and consumers. This makes it extremely important to ensure that the right stakeholders are brought along the journey in order to create autonomy, reduce data complexity and advance the data mesh journey. 

    About the Author

    David Castro-Gavino is the Global Vice President of Data at HelloFresh where he oversees the Data Alliance in an effort to drive data maturity within the company. David joined HelloFresh in 2020 to lead the company’s data journey by providing frameworks, processes, technology and tools that govern collaboration between data producers and consumers. 

    David has been in the data  industry for almost three decades – prior to joining HelloFresh, he was the Head of Data & Customer Analytics at Booking.com. He also spent nine years in a variety of senior technical and commercial roles at dunnhumby, a global customer data science company, from implementing a Data CoE in India to partnering with retailers and brands across the globe to leverage data, science and analytics into the center of customer-focused decision making.

    David graduated in Computer Science from Kingston University and holds a Masters of Business Administration from Hult International Business School. He currently resides in Berlin, Germany.

    Sign up for the free insideAI News newsletter.

    Join us on Twitter: https://twitter.com/InsideBigData1

    Join us on LinkedIn: https://www.linkedin.com/company/insidebigdata/

    Join us on Facebook: https://www.facebook.com/insideAI NewsNOW

    Speak Your Mind

    *