Scores, quadrants and calibration
Two axes
Each tender receives two scores from 0 to 100:
- Interest: what the deal is worth to the company (sector fit, budget size, funder, country, categories).
- Win probability: our chances of winning (presence, track record, competition, prequalification, client relationship).
Each axis is the weighted sum of its criteria:
axis score = Σ (criterion value × criterion weight) / 100
The weights of an axis always total 100. Criterion values (0-100) are produced by the rule engine from the record's facts, or overridden by hand.
Composite score and quadrant
composite = round((interest + win) / 2)
The quadrant guides the decision:
| Quadrant | Condition |
|---|---|
| Top priority | interest ≥ 75 and win ≥ 70 |
| Partnership | interest ≥ 75 and win < 70: attractive deal, approach in a consortium |
| Opportunistic | interest < 75 and win ≥ 70: easy to win, moderate interest |
| Low priority | everything else |
Criteria weights
Menu Steering > Scoring criteria (commercial management). Sliders change each criterion's weight; Normalise brings the axis total back to 100 while keeping proportions. Apply creates a new dated weighting profile and re-scores the whole portfolio; the profile history remains available. Reset restores factory weights.
Calibration
The model learns from your outcomes. At each submission the predicted win probability is frozen; once the outcome is known, the application compares the average prediction with the actual win rate:
bias = clamp(round(0.5 × (actual rate − average predicted)), −15, +15)
This bias, in points, is added to the win probability axis of every open tender. It is recomputed after each outcome, or on demand with Recalibrate. The screen shows the number of known outcomes, the average prediction, the actual rate and the applied bias.