Question 260
At fictional Stonecroft Analytics Group, a Snowflake data and AI engineering team is conducting a source-system modernization and AI search review before any automated output can change customer or financial records. The separately composed technical incident is: A sales assistant generates different SQL aggregations for the same gross-margin question because the semantic description leaves date treatment and returns ambiguous. Assume only these documented facts and use safe, authorized production-change procedures; exact current Snowflake feature syntax and region or account availability require current review. Which repeatable assurance or control is most relevant to preventing this situation?
Explanation
Writer-proposed answer: D (UNVERIFIED)
Preliminary individual option explanations — NOT expert validated
Correct Answer
Review verified examples after semantic-model changes and add reconciliation tests for critical finance and revenue questions. Business semantics and verified examples constrain text-to-SQL variation; plausible SQL is not necessarily the correct business answer.
Incorrect Answers
This response concerns Summarization loses contractual exceptions, which is different from the stated failure. A sales assistant generates different SQL aggregations for the same gross-margin question because the semantic description leaves date treatment and returns ambiguous.
Token budget exceeded by prompt assembly is insufficient to demonstrate this result: Regression queries across paraphrases use the agreed measure and reconcile results to independently approved financial calculations.
Structured completion schema validation is not the cause of the reported failure: A sales assistant generates different SQL aggregations for the same gross-margin question because the semantic description leaves date treatment and returns ambiguous.