vendor.score

Every vendor decision comes down to a weighted scorecard, and every scorecard hides the same question: would the answer change if our priorities shifted a little? vendor.score ranks vendors and then stress-tests the result, so you learn whether the winner is robust or an artifact of the weights. Standard library only.

Use

from vendor_score import margin, rank, sensitivity

weights = {"cost": 0.5, "quality": 0.5}
scores = {"A": {"cost": 9, "quality": 6}, "B": {"cost": 5, "quality": 9}}

rank(weights, scores)                      # [('A', 7.5), ('B', 7.0)]
margin(weights, scores)                    # 0.5 points between first and second
sensitivity(weights, scores, swing=0.5)    # {'leader': 'A', 'stable': False, 'flips': [('cost', 'down', 'B'), ...]}

Design

• The ranking is the easy part. Scores are 0 to 10, weights are normalized, ties break alphabetically so output is deterministic.
• Sensitivity is the point. Each weight moves up and down by the swing you choose. If any move changes the leader, the result says which criterion, which direction, and who takes over.
• Margin shows how close it is. A leader by half a point on a fragile weighting is a different decision from a leader by three.
• Bad input fails loudly. Missing scores and out-of-range values raise, instead of quietly skewing a seven-figure decision.

Tests

python -m unittest discover -s tests -t . -v

MIT licensed. Built by AndMaverick LLC.

---

If this is your problem. If you are deciding what to build, buy, and fund, I help leadership teams make that call with evidence. Begin a conversation.