Skip to content
Two people practicing yoga in high lunge pose with arms raised.

Increase in Hiring Quality

Time saved by recruiters

Two people practicing yoga in high lunge pose with arms raised.

Our client is a fast-growing direct-to-consumer nutrition and wellness brand. This company ships thousands of orders monthly across 200+ cities in India. Known for its subscription-led model and a loyal urban customer base, the brand has built its growth on fast fulfillment and high repeat purchase rates.

As order volumes scaled, so did the complexity of last-mile delivery — particularly in Tier 2 and Tier 3 markets where address data is inconsistent and delivery partner SLAs vary significantly. NDR volumes grew in proportion, creating a silent but compounding pressure on margins, ops bandwidth, and customer satisfaction scores.

The Problem Statement

For any D2C brand shipping at scale, NDR (Non-Delivery Report) management is a silent revenue drain. Every undelivered order is a potential return, a refund, a lost customer. And the traditional fix — calling customers manually to reschedule delivery — is slow, expensive, and inconsistent.

This brand was sitting on thousands of NDRs every month. Their ops team was stretched. Reattempt confirmation rates were low. Return-to-origin costs were climbing.

They needed a better way to recover orders, not more agents.

What we Did

Rootle.ai deployed a fully automated outbound Voice AI agent to handle NDR follow-ups end-to-end — calling customers, identifying the reason for non-delivery, confirming or correcting delivery details, and capturing the preference for reattempt. No scripts handed to agents. No queue. No delays.

The PoC ran across 5 campaigns covering 3,075 orders, totalling over ₹40 lakh in order value.

The KPIs We Agreed On Before Writing a Single Script

Three numbers. Locked upfront. Non-negotiable.

Target
45%+
Contact Rate
Hover to see result
Achieved
49.3%
Contact Rate
1,517 customers reached out of 3,075 dialled across 7,808 call attempts.
Target
30%+
Reattempt Confirmation
Hover to see result
Achieved
35.1%
Reattempt Confirmation
533 customers confirmed redelivery. ₹6.9L in orders back on track.
Target
100%
Structured Outcomes
Hover to see result
Achieved
100%
Structured Outcomes
Every conversation tagged. Reattempt, cancel, callback, wrong number. Zero unresolved calls.

Target set before Campaign 1  ·  Measured after Campaign 5

What Made It Work

Not just calls. Structured outcomes at every step.

End-to-End Automation
Every call placed, tracked, and resolved by Voice AI — no human queue, no delay, no inconsistency across campaigns.
Real-Time Data Capture
Addresses, pincodes, landmarks, reattempt confirmations — all structured and captured mid-conversation, ready for direct CRM or logistics sync.
Intelligent Outcome Triage
Every conversation classified — reattempt, cancel, callback, wrong number, fraud flag. No untagged outcomes. No manual review required.
Escalation Handling
Cancellations, refund requests, angry customers, and fraud cases detected and escalated in real time — before a second attempt was wasted.
Retry Logic Built In
Avg 2.54 call attempts per contact. Voicemail detection, callback scheduling, and alternate number collection — all handled without agent involvement.
24×7, Zero Downtime
Campaigns ran continuously across time windows with no shift changes, no breaks, and no drop in conversation quality from first call to last.

How We Tuned the Rootle Engine

Five adjustments across five campaigns. Each one measured.

Retry
Cadence
AVG 2.54 ATTEMPTS
Adjustment 01
Retry Cadence Calibration
Campaign 1 used flat retry logic. By Campaign 2, peak answer windows were identified from the data. Retries restructured around those windows. Contact rate jumped from 50% to 58%.
Voicemail
Detection
86 DETECTED
Adjustment 02
Voicemail Detection Tuning
Early campaigns had voicemail bleed. The engine was tuned for Indian telecom carrier voicemail patterns. 86 voicemails correctly identified and logged for follow-up.
390
Wrong
Numbers
FLAGGED & LOGGED
Adjustment 03
Wrong Number Handling
Campaign 4 surfaced a 31% wrong number rate. A mid-campaign flag was raised to the ops team. Each wrong number captured and logged in real time, producing a clean CRM correction list.
Convo
Flow
UNDER 3 MIN/CALL
Adjustment 04
Conversation Flow Refinement
Transcripts were reviewed between campaigns to find drop-off points. Opening line shortened, address confirmation prompt moved earlier. Average call duration stayed under 3 minutes while resolution rate improved.
246
Escalations
Routed
REAL-TIME TO CRM
Adjustment 05
Escalation Routing
Cancellation and refund requests routed to a priority queue in the client's CRM in real time from Campaign 5 onward, ensuring the ops team could act before a second delivery attempt was dispatched.

The Outcome

More recovered orders. Cleaner delivery data. Fewer returns. Lower ops cost. And a logistics database that came out of the process significantly more accurate than it went in.

All fully automated. Zero human agents. 24×7.