learning-decision-needs-whylisted

# A decision without a why does not exist Every change gets three fields, and an empty field means no decision was made: - **why**: the fact, number or quote that made you change it; - **expected**: a prediction you can check later; - **rejected**: what you consciously did not take, and why. People and companies dropped during lead selection are logged with a reason code at that moment, not thrown away. ## Why A log of consequences ("message sent, lead replied") is history, not knowledge. Every tool has it, and a year later it is worth nothing. Six months on, a log without motive cannot tell a deliberate change from a random one, and the candidates you dropped cannot be reconstructed at all. The valuable part is which reason and wording worked for which kind of sender in which situation, together with what was rejected. That part is not in any public database and cannot be rebuilt over a weekend. ## How to apply 1. Before applying a change, write the three fields. Make the writer refuse an empty field (a required column or a validation in the loader). 2. Good **why**: "a broad search returned 209,135 people, so the segment is not chosen yet". Bad: "improved the text". 3. Good **expected**: "reply rate above 5% on the first 30 sends, otherwise we change the segment, not the copy". Bad: "it will get better". 4. Good **rejected**: "no enterprise 1000+ before first data: long cycle and no case study". Bad: empty. 5. At selection, store each dropped candidate with a reason co