How Myntra Scaled Customer Support 3x Without Adding Headcount

As Myntra's customer base grew, so did the complexity of keeping the support experience fast, consistent, and human. Inbound query volumes surged during festive sales, exposing the limits of manual operations and legacy automation that could route calls but not resolve them. Key challenges included:

• Holiday surges requiring 3× staffing, driving 15,000+ unanswered calls per day and $1.8M in avoidable operational cost annually.
• Every minute of hold time adding measurable friction - a 1.2% drop in repeat purchase intent per additional wait-time increment.
• Fragmented workflows across WISMO, returns, and refunds leading to delayed resolutions and 1.2–1.8% of annual revenue lost to churn.
• Existing automation unable to validate identity, fetch live order context, or adapt mid-conversation - routing without resolving.
• Inconsistent bilingual coverage across Hindi and English, creating uneven experiences for a diverse customer base.

‍

Myntra partnered with Nurix to deploy Meira - a Voice AI Agent on NuPlay - purpose-built for high-volume inbound CX. Meira was designed not just to handle calls, but to understand intent, validate context, and resolve queries end-to-end across Hindi and English, at any volume, without adding headcount.

Nurix's Forward Deployed Engineers worked closely with Myntra's team to hand-stitch integrations into Myntra's custom CRM and Exotel telephony, and to rethink support SOPs for an AI-led interaction model - because designing for machines isn't the same as designing for humans. The result was production-ready in 6 weeks.

Key workflows included:

Real-time inbound query resolution: Meira validates customer identity, fetches live order data, and resolves WISMO, return, and refund queries end-to-end - integrated directly into Myntra's CRM with zero infrastructure changes or downtime. Customers get answers without waiting for a human, and without repeating themselves.

Bilingual by design: Hindi and English handled natively and fluidly, matching how Myntra's customers actually speak - no forced formality, no switching friction, no degraded experience for vernacular callers. Meira moves between languages naturally within a single conversation.

Multi-agent architecture: Specialised sub-agents for refunds, order tracking, and cancellations are orchestrated in real time, boosting containment rates and reducing conversation latency by 40% compared to single-agent flows. Each sub-agent is optimised for its domain - no generic handling.

Human-in-the-loop escalation: Complex cases escalate seamlessly with context-rich warm transfers - full conversation history, intent signals, and CRM context passed to human agents so every handoff continues the conversation rather than restarting it.

Peak-season resilience: Stress-tested for Myntra's festive sale volumes. Meira scales up and down with demand - no surge hiring, no service degradation. 99.8% uptime maintained throughout peak periods with consistent quality across every call.

Continuous learning: A live CX Insights Dashboard with predictive alerts and anomaly detection enables ongoing improvement post-launch. Each interaction trains the model - Meira gets sharper with every call.

Together, these workflows created an always-on resolution layer that ensured no inbound query went unanswered, no customer had to repeat themselves, and no human counsellor spent time on cases the AI could fully resolve.

‍

With NuPlay live across all workflows, Myntra saw meaningful improvements across efficiency, resolution quality, and cost:

• 50% reduction in Average Handle Time
• 75%+ routine queries resolved end-to-end by Meira
• 80% peak containment during festive sales
• 15,000+ average calls handled per day

‍

‍