Unified Credit Decisioning
A single scoring workflow that processes all channels through bureau integration, scorecard selection, and risk classification — returning channel-specific responses in real time.
Workflow Pipeline
Every scoring request follows a seven-stage pipeline, ensuring consistent decisioning across all channels.
Stage 1
Inbound scoring request received via REST API. The request carries applicant identifiers, channel code, and any pre-populated data from the origination system.
Channels: AO (Affordability Online), CR (Credico), WA (WhatsApp/InfoBip).
Stage 2
Identity verification, duplicate detection, and input sanitisation. Checks for existing active accounts, blacklisted IDs, and mandatory field completeness before proceeding.
Early exits for invalid or duplicate applications reduce unnecessary bureau costs.
Stage 3
Automated credit bureau enquiry routed to the correct regional Experian endpoint based on the applicant's country of origin. Full credit profile retrieved and parsed.
Four bureaus: Experian SA, Lesotho, Namibia, Eswatini.
Stage 4
The engine selects the appropriate scorecard based on the applicant's credit profile: Paying Segment for customers with payment history, Non-Paying for thin-file applicants, or Thin-File for those new to the bureau.
Scorecard selection is fully automated based on bureau response data.
Stage 5
The selected scorecard computes a numeric score, credit limit segment, maximum term, risk grade, and strikethrough indicator (approve/decline). All variables logged for audit.
Deterministic scoring with full variable-level traceability.
Stage 6
Additional business rules applied after scoring: showroom risk indicators for physical stores, rating scale mapping, and lead source classification for downstream reporting.
Nine lead sources tracked for conversion analytics.
Stage 7
Channel-specific response formatting. AO receives a simplified score with flags. CR and WA receive a full response including lead wording for agent or conversational consumption.
Average response time under 2 seconds end-to-end.
Scorecard Types
Scorecard selection is automated based on the applicant's credit bureau profile.
For customers with established payment history on the bureau. Uses historical payment behaviour, existing facility performance, and account tenure as primary scoring inputs.
For applicants with a bureau record but no meaningful payment history. Relies on demographic data, enquiry patterns, and address stability for risk assessment.
For applicants new to the credit bureau with minimal or no records. Conservative scoring with lower initial limits, relying on identity verification and basic demographic signals.
Score Output
Every scored application produces a structured response with these key fields.
| Field | Description |
|---|---|
| Score | Numeric credit score computed by the selected scorecard |
| StrikethroughIndicator | Approve or decline flag based on score thresholds and policy rules |
| MaxTerm | Maximum repayment term in months allowed for this risk profile |
| CL_Segment | Credit limit segment determining the maximum facility size |
| ShowroomRiskIndicator | H/M/L risk classification for physical store transactions |
| RiskGrade | Overall risk grade for reporting and portfolio segmentation |
| RatingScale | Mapped rating scale for regulatory and internal reporting purposes |
Channel Routing
The same scoring pipeline delivers tailored responses based on the originating channel.
Simplified response — score and flags only. The self-service front end consumes a minimal payload to drive the customer-facing approve/decline experience.
Full response with lead wording. Agents receive the complete score output plus pre-formatted text for customer communication and next-step guidance.
Full response with lead wording. The conversational bot formats the scoring result into WhatsApp message templates for immediate customer delivery.
The scoring engine is one part of a complete credit lifecycle — from origination through collections.