Scorecards

A Scorecard converts several business factors into one explainable numeric result. Each factor contributes a partial score, and the platform adds those contributions to the initial score to produce FinalScore.

Common uses include credit and fraud risk, lead prioritization, customer health, application quality, and supplier assessment.

When to use a Scorecard

Use a Scorecard when stakeholders think in points or weighted bands: “age 25–40 adds 10 points” or “three overdue payments subtract 30 points.” Use a Decision Table when combinations directly produce categories, or a Ruleset when the main outcome is a series of mutations.

A common pattern is:

Scorecard calculates FinalScore
        ↓
Decision Table converts the score into Low / Medium / High
        ↓
Workflow routes each band to the next action

Configure the initial score

The Initial Score is the starting value before partial scores are applied. Use 0 for additive scoring. Use another value only when the business model intentionally starts from a baseline, such as 100 points with penalties subtracted.

Document whether higher means better or worse. A model that mixes positive “quality” points and positive “risk” points is difficult to interpret.

Add partial scores

Each partial-score section selects a model attribute and includes a description. Add cases for exact values, ranges, or the all-other-values fallback. Each case has a label and numeric score.

Labels are important evidence. Stable income: 5+ years communicates more than Case 3, particularly when a reviewer sees the debug result.

Worked example: application risk

Assume higher scores mean higher risk and the initial score is 0.

Age

Value/range	Label	Score
Under 21	Limited credit history	20
21–65	Standard age range	0
Over 65	Additional affordability review	10
All other values	Age unavailable	25

Debt-to-income ratio

Value/range	Label	Score
0–0.20	Low debt	0
Over 0.20–0.40	Moderate debt	15
Over 0.40–0.60	High debt	35
Over 0.60	Very high debt	60
All other values	Ratio unavailable	40

Payment history

Value	Label	Score
Clear	No adverse history	0
Minor	Minor late payments	15
Adverse	Significant adverse history	50
All other values	Unknown history	30

Request:

{
  "age": 34,
  "debtToIncomeRatio": 0.47,
  "paymentHistory": "Minor"
}

Calculation:

Initial score                     0
Age 21–65                         0
Debt ratio over 0.40–0.60        35
Minor payment history            15
FinalScore                        50

Response:

{
  "FinalScore": 50
}

Ranges and fallbacks

Avoid gaps and accidental overlaps. Decide which case owns an exact boundary such as 0.40 and verify it in Testing Zone. Add an all-other-values case when missing, unknown, or new categories must have a deliberate score.

Bulk input can accelerate entry of long bands or category lists, but review the generated cases. Sorting visually does not correct overlapping definitions by itself.

Convert a score into a decision

A Scorecard produces a number, not an entire business process. In a Workflow, follow it with a Condition or Rule node:

FinalScore < 30       → Low risk → automatic approval
FinalScore 30–59      → Medium risk → manual review
FinalScore >= 60      → High risk → decline

Keep the banding logic separate when policy owners need to change thresholds without changing how points are calculated.

Test and explain

Save tests for:

• exact lower and upper range boundaries;
• a value in every band;
• missing and unknown values;
• the lowest and highest possible score;
• combinations that land exactly on a downstream decision threshold.

Debug output identifies each partial-score contribution. Use the labels and descriptions to explain why the final score was produced.

Common problems

Symptom	Check
Score is higher/lower than expected	Initial score, overlapping ranges, and sign of each contribution
A factor contributes nothing	Field mapping, missing input, or no matching/fallback case
Exact boundary enters wrong band	Inclusive/exclusive boundary configuration
Workflow route is wrong	Map FinalScore correctly and inspect downstream thresholds
Reviewers cannot explain the result	Add meaningful partial-score descriptions and case labels