Reverse problem: the data arrived in «wide» form — one row per product, with three months' sums spread across three separate columns m1, m2, m3. Analytics prefers a «long» layout where every «product + month» combination is its own row. Postgres has no native «expand columns into rows» operation, but you can build it by combining several queries. Build a wide_sales table with columns product (string up to 50 chars), m1, m2, m3 (numbers), insert three rows for products A, B, and C. The output should produce three rows per product: «product, month number from 1 to 3, sum for that month». Sort by product, then by month number.