Abstract
The expanded guide follows synthetic invoice records through landing, bronze, silver, gold and semantic consumption; builds foundational and derived products; and carries the result through service commitments, late corrections, publication and reuse.
A guide to defining data products that people and systems can use, with clear meaning, ownership and operating commitments. Publishing a table does not establish meaning, supported use or reliability; a product is a promise to its consumers.
It covers product boundaries and ownership, a primary ownership taxonomy kept separate from analytical and operational use, freshness and finality, master and reference data choices, the product contract, quality and reconciliation, lifecycle and certification, and the measures that show whether a product is used and trusted.
It is the companion to the Data Product Certification Standard: this paper decides what to build and own, the standard decides when a version can be relied on.
Expanded guide · 29 October 2026
How source records become a supported data product.
Follow a synthetic invoice through controlled intake, bronze source records, silver validation and a gold publication. Then combine supported invoice, payment and reference products into a dated receivables balance.
Processing stages describe preparation. The product contract defines the consumer promise: meaning, rights, quality, freshness, ownership and support. A gold table is not automatically certified.
Foundational and derived are product relationships.
Invoice Lines and Payment Allocations publish supported source-aligned capabilities. A derived Receivables Exposure product owns the combined definition, input compatibility and reconciliation. A separate current-invoice API supports requests that need live state.
One example through publication and change.
At cutoff T0, synthetic invoice INV-100 has 1,000 invoiced, 100 in approved credits and 400 in applied payments: 500 outstanding under the example definition. A later payment of 200 changes the current result to 300. Historical versions and closed-period restatements need explicit handling.
The expanded PDF also compares analytical, operational and knowledge products; explains service indicators, objectives and agreements; and follows a source correction through testing, approval, publication and consumer notification.
All records and values are illustrative. This is reference guidance, not a client result or a statement of Studio capabilities.
Decision it supports
Whether a dataset deserves to become a product, and what it must promise.
What you will produce
A product contract with consumer, grain, freshness clock, owner and change policy.
In the paper
Three passages, as published
Separate identity from meaning
An enterprise party identifier can connect records. It doesn’t establish that every domain defines customer, exposure or active status the same way.
Measure consumer impact
A failed test matters because it threatens a use.
The test of success
A consumer can interpret and use the product, the team can operate it, and an executive can explain its value and exposure.
Next step
Discuss this paper against your own program.
A free sixty-minute working session with Richard on your top gaps. You leave with a one-page summary within three business days, and three next questions.