Organizations are spending millions building data products that get used once and then are forgotten. It's a cycle that drains budgets, delays AI initiatives, and erodes trust in data.
Experts recently convened for DBTA’s webinar, Stop Rebuilding What You Already Have: Why Data Product Reusability Is the Key to AI-Ready Data, to make the case that reusability isn't just a "nice-to-have," it's the economic engine behind successful data strategy.
Principal Advisor and industry analyst at Radiant Advisors John O'Brien explained that there is a compounding effect with delivering data at scale which data products can solve.
“Products actually represent this generations approach to solving the scaling problem, which is delivering real high quality consistently, prioritizing needs and data features,” O’Brien said.
According to Ryan Crochet, director of product marketing at Quest, a data product consists of the following:
- Reusable: Built once, consumed many times
- Valuable: Delivers measurable business outcomes
- Trustworthy: Backed by governance, quality, and lineage
- Discoverable: Easily found in a catalog or marketplace
- Accessible: usable by reports, AI, apps, etc.
- Composable: Combines with other products for new value
“The most defining piece of it…it’s another dataset directly correlated to a business outcome,” Crochet said. “It has to have a purpose within the business.”
Chief value officer Stephan Liozu at Quest said he hopes with the emergence of AI the data product concept will accelerate exponentially.
When it comes to reusability, O’Brien suggested it’s the test as to what should become a data product.
Reusability requires deliberate design and governance on the technical side and organizational side. For the technical side, it includes:
- Standardized schemas and definitions
- Strong metadata and cataloging
- Stable, versioned interfaces
- Modular architecture
- DataOps automation
The organizational side requires:
- Clear product ownership
- Shared funding models
- Incentives that reward use
- Reuse metrics and measurement
- Overcome the “not invented here” culture
“A lot of the data teams are disconnected from the business,” Liozu said. “They need to ask, ‘is that data product being created solving a business problem that would be good for the business?’ It’s having that business mindset in the process of designing data assets or products so they can be reused fully. The way to make it reusable is thinking about who else can use it, what else can be built on top?”
For the full webinar, featuring a more in-depth discussion, Q&A, and more, you can view an archived version of the webinar here.