Work/
Decision Engine

LENDERKART DECISION ENGINE

Category

Product Development

Client

My Debt Plan

Infrastructure

Microsoft Azure

Role

Product Lead, Algorithm Design

LenderKart — a proprietary lead routing engine built for fintech marketing websites. Every incoming lead is instantly evaluated, scored, and routed to the right lender or CRM system, maximising ROI on marketing campaigns.

Project Overview

Fintech marketing websites generate high volumes of leads, but not all leads are equal. Routing a lead to the wrong lender wastes the lead and the lender's time. Manual routing is slow, inconsistent, and doesn't scale. LenderKart automates the entire decisioning pipeline — from the moment a form is submitted to the moment an offer reaches the customer — in real time.

The Challenge

  • 01 Lead Quality — not every lead meets every lender's eligibility criteria
  • 02 Real-time Routing — matching and routing must happen before the user drops off
  • 03 Multi-lender Integration — coordinating outreach to multiple lender APIs simultaneously
  • 04 CRM Sync — every lead interaction needs to land in the CRM regardless of outcome
  • 05 Scalability — high-volume campaigns demand consistent performance at load

How It Works

From form submission to CRM update, every step is automated, logged, and auditable.

01

Form Submitted

Lead enters the system

02

Data Validated

Profile scored & enriched

03

Offer Mapping

Matched to lender criteria

04

Partner Outreach

Requests sent to lenders

05

Offers Received

Responses collected

06

Offer Displayed

Best match shown to user

07

CRM Updated

Data submitted to CRM

Logic & Algorithm

The engine uses a tiered filtering and scoring approach:

  1. 1 Hard Filters — immediate exclusion if a lead doesn't meet a lender's non-negotiable eligibility rules
  2. 2 Weighted Scoring — remaining lenders scored against positive and negative indicators for this lead profile
  3. 3 Parallel Outreach — requests sent simultaneously to top-ranked lenders via their APIs
  4. 4 Best Offer Selection — offers ranked and the optimal one surfaced to the customer
  5. 5 CRM Write — full lead record, scores, and outcome written to CRM regardless of result

Technical Implementation

Infrastructure

Hosted on Microsoft Azure — cloud infrastructure chosen for reliability at production scale.

Integrations

Real-time lender API connections, CRM write-back, and webhook support for async outcomes.

Key Outcomes

Leads Processed

100k+

Production Status

Live

Lessons Learned

It's knowledge engineering, not just code: The most critical phase was time spent with domain experts translating their intuition into scoring weights. No amount of engineering replaces that understanding.

Plan for lender API failures: Lender APIs go down, return unexpected shapes, or respond slowly. The engine needs graceful degradation — a failed lender call should never block the rest of the pipeline.

Always write to CRM: Whether a lead converts or not, the full record and outcome must land in the CRM. That data is how you improve the scoring model over time.