A dataset without an owner is a dataset whose quality silently degrades. Nobody has an incentive to maintain it, nobody answers when something fails and consumers discover the problem through the wrong decision the data fed. Data as a product reverses that logic with six DATSIS characteristics: Discoverable (consumers find what exists), Addressable (stable access point), Trustworthy (published quality guarantees), Self-describing (documentation the consumer reads without asking the producer), Interoperable (built to compose with other data products) and Secure (explicit access control).
Subcapability 05 of 05 · Data & Analytics
Data Products
DATSIS principles and Data Contracts that transform ownerless datasets into products with SLA, defined consumers and explicit accountability, eliminating the central bottleneck no backlog can absorb.
What is at stake
The data team has a six-week backlog. Every new analysis starts with a ticket, a prioritization and a wait. Meanwhile, the data that already exists without an owner silently degrades because the only team that could maintain it is busy serving the request queue.
What it is, in practice
How we work
Data product with DATSIS as certification criterion
We structure each critical dataset as a data product with the six verifiable DATSIS attributes: discoverable via catalog, with a stable access point, with quality guaranteed by contract, self-documented, interoperable with other products and with explicit access control. A product that does not meet the criterion does not receive certification.
Data Contracts between producer and consumer
We formalize the agreement between data producer and consumer with documented schema, quality and latency SLA, and notification process before any breaking change. The contract makes schema changes predictable rather than a surprise that breaks a consumer in the middle of the night.
Self-serve platform for domains
We provide the infrastructure that domains need to publish and operate their own data products without depending on the central team: storage, orchestration, monitoring and catalog accessible as self-service. Domains maintain autonomy. Standards are shared.
Internal data marketplace with governed discovery
We deploy a data product catalog that makes the company data portfolio discoverable by any team, with quality evidence, identified owner, accessible contract and a governed access process. The consumer finds the right data without depending on tribal knowledge.
Internal NPS as adoption indicator
We measure internal consumer perception of the quality, reliability and ease of access of each data product. A low NPS identifies a product that was published but does not meet the real need. A dataset with an interface and no active consumer is storage cost.
Measurable gains
What changes in the result when this subcapability matures.
Number of certified data products with active consumers
Data products with DATSIS certification and documented active consumers prove the data-as-product model is generating real adoption. A dataset published without a consumer is storage cost with an interface.
Lead time of new data product from request to governed publication
With a self-serve platform and a documented process, a domain publishes a new data product without depending on the central team. Time that used to take weeks in a backlog reduces to days of autonomous domain work.
Internal NPS of data product consumers by domain
NPS measures whether the product is being adopted with satisfaction or merely tolerated as the only available option. A domain with low NPS has a clear signal that the data product needs to evolve before being certified as reliable.
Percentage of data requests served by the catalog without a ticket to the central team
When a consumer finds the data they need in the catalog without opening a ticket to the central team, the data product model is working. This percentage is the indicator that the central team backlog is being transferred to domain autonomy.
Frequently asked questions
What differentiates a data product from a conventional dataset?
A conventional dataset is data stored with some structure. A data product is data treated as a software product: it has an owner with formal accountability, a quality and latency SLA, documentation the consumer reads without asking the producer, a stable access point and a contract that governs what changes and how the consumer is notified. The practical difference is accountability: when the data product fails, there is an identified responsible party with a resolution process.
What is a Data Contract and how does it protect the consumer?
A Data Contract is the formal agreement between producer and consumer that specifies schema, field types, minimum quality by dimension, maximum latency and notification process before any breaking change. When the producer changes the schema without notice, the consumer discovers it through a pipeline error. With the contract, the producer communicates the change with enough lead time for the consumer to adapt without an incident.
What are the DATSIS principles?
DATSIS is an acronym defining the six properties of a quality data product. Discoverable: the consumer finds the data via catalog without depending on tribal knowledge. Addressable: a stable access point that does not change without notice. Trustworthy: verifiable quality with a documented SLA. Self-describing: complete documentation accessible without needing to ask the producer. Interoperable: compatible with other data products. Secure: explicit access control with rules on who can access what.
How to prioritize which datasets to transform into data products first?
Priority follows two simultaneous criteria: the impact of the decision the data feeds and the number of teams that depend on that data. Data that feeds revenue decisions and is consumed by three or more teams has the highest priority. Starting with those cases creates evidence of the data product model's value before expanding to the full portfolio.
Can a company without Data Mesh have data products?
Yes. Data products are a data management model, not a specific architecture. An organization can start treating critical datasets as products, with owner, SLA and documentation, without restructuring the entire data architecture into Data Mesh. The ownership and contract model between producer and consumer works in any architecture that has active governance. Data Mesh scales this model to the entire organization.
Other subcapabilities in this capability
Data Architecture & Governance
Data Management Body of Knowledge and Data Mesh with federated governance that structure data as a formal asset with owner, traceable lineage and domain accountability that scales without a central bottleneck and enables AI in production.
Data Engineering & Pipelines
Apache stack with orchestration, observability and idempotency that eliminates the artisanal pipeline without monitoring and ensures no executive dashboard ever shows a wrong number with the appearance of a correct one.
Analytics & Business Intelligence
Semantic layer, governed self-service and KPIs wired to the result that replace expensive intuition with evidence-based decisions and eliminate the debate about which number is right in every executive meeting.
Data Quality & Master Data
Six data quality dimensions and MDM Hub Architecture that eliminate the three versions of the same customer across systems and transform data from a source of debate into a verifiable base for every executive decision.
Want clarity on where to invest first?
A complete technology capability assessment with an evolution roadmap connected to financial result.