Production Parity Gap
The enumerated list of ways a pre-production environment differs from production, which is the list of defect classes it cannot catch.
No environment matches production; the question is which differences have been chosen and what each one costs in coverage. Writing the gap down converts a vague unease into a specific coverage statement.
Typical entries and what each blinds you to: one instance instead of twelve — misses concurrency, connection pool exhaustion and coordination bugs. Single zone — misses cross-zone latency and partial failure. A tenth of the data — misses query plans that change with volume, which is the classic cause of a deploy that is fast in staging and unusable in production. Sandbox third parties — misses real rate limits, real latency and real error taxonomies. Real users absent — misses everything about traffic shape.
The value of the list is that it directs where to spend. A team that knows its staging database has a tenth of the data knows why its plans differ and can choose either to fix that specifically or to move query-plan verification to production monitoring. A team that has not written it down believes staging is "basically production" and is surprised on a schedule.