Work out whether your discounts are buying revenue or giving it away

You are a revenue analyst. Below is a deal-level or order-level export with discount amounts, plus my cost and retention notes. Tell me what the discounting is actually doing.

1. WHAT THE DATA SUPPORTS - the columns present, rows dropped and why, and which of my questions this export cannot answer.
2. DISCOUNT BANDS - table: Band (0%, 1-10%, 11-25%, 26%+) | Deals | Revenue | Gross margin | Win rate | Retention or repeat rate | Average deal size. Note where a band has too few rows to trust.
3. WHERE IT PAID FOR ITSELF - bands or segments where the extra volume or retention covers the margin given up, with the arithmetic shown.
4. WHERE IT DID NOT - discounts that went to buyers who looked likely to buy anyway, and the annual amount involved.
5. THREE RULES TO ADOPT - specific caps or approval thresholds, each with the number that justifies it and what to watch after.

Rules: correlation is not causation - say so where selection bias is likely. No invented benchmarks, no advice not grounded in my rows.

EXPORT:
{{paste or attach deal-level data with discount, margin, segment, outcome}}

COSTS AND RETENTION NOTES:
{{unit cost, renewal or repeat rates if you have them}}