Data management continues to change as companies try to find the best ways to manage and store endless amounts of data. Data has become a differentiator for many businesses as they aim to know, analyze and build trust with their consumers.
For years, many companies have been utilizing data models that are fairly inefficient. Yes, the data is there, but it is disorganized, centralized, and limited in scalability — a monolithic structure with all data managed by IT.
The chaotic paradigm creates organizational bottlenecks, with departments sometimes unable to utilize much-needed data and IT professionals unaware of the domain-specific concerns of various business teams.
A data mesh strategy contrasts with the traditional approach and provides a better way to share the spread of data.
A data mesh decentralizes data management with “data products” — the pinnacle of data integration — using a combination of data, code, data integration tools, and infrastructure. Data products are managed by data product owners who belong to domain-focused teams.
Benefits of Data Mesh for Customers and IT
A data mesh is beneficial in many ways. It allows everyone upstream to access data, test out different scenarios, run models and make quick decisions. Data products have owners who are responsible for the care and livelihood of their data.
Data from one or more data products are consumed by internal applications and external customer-facing digital products. When done correctly, IT no longer has to carry all the weight as data products are easily discoverable and can be managed by the many people within the business.
A data mesh architecture also supports the idea of real-time updates, which are critical to ML models making quicker decisions. Those outputs are used to deliver a better customer experience (e.g., recommendation engines).
How Data Mesh Can Be Used at Every Stage
Now that you know the basics of data mesh and why it is important, you may be wondering, “What does data mesh look like in real-time?” Here’s how data mesh can be a valuable asset to use throughout the entire digital product development life cycle.
1. Before the Launch
Start the process by establishing expectations through conversations with business partners. Identify and prioritize data products based on common use cases and determine the best way to meet those expectations.
Data products fit into two camps — source-aligned data products, which are operationally focused and typically used to facilitate integrations, and consumer-aligned data products, which are meant to satisfy business-oriented needs.
A highly effective way to identify and prioritize your data products is to perform a business capability assessment.
Intuit undertook this discovery with a 245-user survey intended to reveal its data-centric needs and challenges. The results of that mission were a new strategy that empowered data workers to create the best data products they could, enabling increased productivity throughout the company.
2. During the Launch
For digital products and internal applications, embed data product owners into feedback processes and all other ceremonies. Stay immersed in feedback to drive each aspect of the launch.
For example, update a data product with another attribute or buy third-party data. Like other product owners, data product owners need a standard intake process for requests and a way to manage them. What gets prioritized should depend on the investment cycle around a given data product consumer (e.g., internal applications, customer digital products, etc.).
Engaging both IT and business stakeholders is critical to data product owner success, and it requires a balance between technology and soft skills.
3. After the Launch
Don’t become complacent. Think about the next big problem to solve and innovative ways to do it. Keep tuning the model and figure out what works for your company because one size doesn’t fit all.
All digital products follow an investment life cycle. Prepare for the next one and revisit your data products to ensure they are delivering the data necessary to provide value both inside and outside (i.e., your customers) the enterprise. Do a retro and apply the lessons learned back into your processes.
Make data an asset to the organization by demonstrating the value it brings to all of your products.
Digital products are things that evolve over time; have road maps; and require time, energy, and problem-solving ability. By allowing data product owners to work directly with digital product teams, the launch process will be even more fine-tuned to meet consumer needs.
Featured Image Credit: Photo by Anete Lusina; Pexels; Thank you!
