Driving Conversions in Lending Onboarding with RevRag.ai

In digital lending, one of the biggest challenges lies not in acquiring users but in converting intent into action. A large segment of customers view loan offers but drop off before completing the journey. From our data, we noticed:

• Offer acceptance rates were below 30% in certain segments.
• Users abandoned the journey at critical points, particularly when confused about interest rates, tenures, or fees.
• Repeat users who were eligible for an offer were also dropping off at a noticeable rate without clear feedback.

The Problems We Identified

1. Fragmented Experience – Customers had cost/tenor questions, but no instant support on the offer page.
2. High Abandonment Rates – Fresh offer-page drop-offs were ~50%.
3. Repeat Users Leaving Mid-Journey – Despite being eligible, users weren’t converting on repeat visits.
4. Limited Behavioral Data – Our analytics (Mixpanel) tracked offer-level data (ticket size, tenure, IRR) but missed granular user interactions and sentiment.

To address this, we explored AI-assisted onboarding journeys with RevRag.ai — embedding a conversational agent within the app.

We designed a phased rollout with focus on coverage, speed, and learning:

Cohort Selection: Instead of attempting a wide rollout, we narrowed down to Fresh + Non-AA users:

• Why Non-Account Aggregator (AA)? Covers ~80% of the base, allowing quicker iteration and avoids extra training/dev cycles.
• Why Fresh? Even with 50% coverage, the impact potential is high, and training a single conversational model limits scope creep.
• AA users will be revisited later once real usage learnings are available.
• Avoid scope creep by training a single conversational model.

Hypothesis Framework: We designed three clear hypotheses to test during the initial rollout:

• Offer Page Engagement (Fresh) – Hypothesis: Adding the AI voicebot widget on the Fresh offer page will increase “View Details” clicks by ≥30%. Impact: Low, but important for top-of-funnel engagement.
• Offer Abandonment Reduction (Fresh) – Hypothesis: By clarifying EMI, cost, or tenure questions, the bot will reduce abandonment by 10%. Impact: Medium, addresses a critical friction point.
• Repeat Drop-off Reduction – Hypothesis: A proactive bot nudge for repeat users will reduce drop-offs by 8% and also capture reasons for non-conversion. Impact: High, since repeat users have strong intent but need clarity to proceed.

Execution Plan

1. Tech Integration with RevRag.ai

• Added necessary permissions in AndroidManifest.xml : WAKE_LOCK, RECORD_AUDIO, MODIFY_AUDIO_SETTINGS, MICROPHONE.
• Integrated RevRag SDK (React Native) for bot deployment.
• Set conditional logic for bot activation: If a user stays idle for ≥10 minutes after viewing an offer, the AI Assistant reappears to re-engage them.
• This balances helpful nudges without overwhelming the user.

2. Analytics & Monitoring Setup

• Analytics Sync: Two-way integration with RevRag to track user journeys.
• Call Recordings Review: Daily sampling mapped to user IDs; CSV reports → dashboard access → S3 pipeline.
• Sentiment Analysis: Flag negative intent/emotion for escalation and service recovery.
• Quality Audit Parameters: Defined audit scorecards for bot conversations.

• AI Evals: Evaluation parameters to capture toxicity/bias in the conversations.

3. Offer Segmentation Filters

We created filters for targeting users with low acceptance but meaningful volume based on time series analysis:

• Exclude AA Eligible users.
• Prioritize Interest Rate ≥ x%, Tenure = yM tenure offers (expandable to y'M) correlation with acceptance.
• Kept high-performing offer segments

4. Customer Experience Hygiene

• Ensured Google Play Store compliance with permissions for microphone & audio features.
• Set up evaluation frameworks (conversion, abandonment, engagement metrics).

How This Solves Conversion

1. On-page Clarification → Lower Abandonment
2. Repeat Engagement → Higher Conversions
3. Sentiment + Reason Capture → Actionable Insights for UW
4. Analytics-First Approach → Continuous Learning

Early Impact Metrics to Track

• % increase in “View Details” clicks.
• Reduction in fresh offer abandonment.
• Repeat offer drop-off reduction.
• CSAT from bot interactions.
• Conversion lift across focus cohorts.

RevRag.ai is helping us reimagine onboarding — from static offer pages to AI-assisted journeys — enabling higher conversions, better user insights, and faster product learning cycles.

✨ This case highlights how AI-driven interventions, if scoped well, can directly improve lending conversions while creating a continuous learning loop for the business.