Structure: fields, types, keys, relations
Metrics: formulas, filters, time
Questions: synonyms and follow-ups
Governance: access, versions, audit
The test object is larger than a table
Data tests verify constraints on rows and fields. Semantic tests also verify that definitions such as “gross margin” or “active customer” are selected correctly in a business question. Both are required.
Give each metric a positive case, a boundary case, and a confusing negative case.
Put equivalent questions in regression
Include formal terms, team shorthand, conversational phrasing, spelling mistakes, and follow-ups. Record expected metric, scope, time, and forbidden fields, then compare both query and result.
After a metric change, rerun the set and confirm that only expected dimensions move.
Test permissions with versions
Ask the same question as different roles and projects, then check visible data and exports. Switch semantic versions and verify that a historical result can still be replayed. BuildTable’s exact testing, versioning, and permission support requires POC confirmation.
Manage failures as defects
For each failure record question, data, version, error class, owner, and status. Classify structure, semantics, access, and interaction separately instead of patching a single prompt.
Public references
Build an AI-ready data foundation
Download BuildTable or talk with us about your data modeling scenario.
Download BuildTable