The bronze orders table stores an ARRAY<STRING> column named coupon_codes. The attribution notebook needs one row per order_id and coupon pair, but many orders carry no coupons at all — their array is empty or NULL — and those orders must still appear in the output, with NULL in the coupon position, so that order totals reconcile. Which expression should the notebook use?

1. A. explode(coupon_codes) — rows whose array is empty or NULL are preserved, with NULL in the generated column 💬 A
2. B. explode_outer(coupon_codes) — rows whose array is empty or NULL are preserved, with NULL in the generated column ✓ B
3. C. posexplode(coupon_codes) — as the positional variant of explode, it is the one that preserves rows with empty or NULL arrays 💬 C
4. D. No generator is safe here — an empty array aborts the query, so orders without coupons must be filtered out before exploding 💬 D

Reveal answer

Correct answer: explode_outer(coupon_codes) — rows whose array is empty or NULL are preserved, with NULL in the generated column

Reference: https://docs.databricks.com/aws/en/sql/language-manual/sql-ref-functions