Case Study – Retail

Executive Summary

DoorDash’s multi-sided marketplace faced challenges in efficiently serving Consumers due to an outdated Interactive Voice Response (IVR) system with touch-tone menus. This led to frustrated callers resorting to pressing “0” to reach a live agent, causing misrouted calls. DoorDash aimed to improve call routing accuracy for customers by over 35% for 15 unique call reasons in just eight weeks, while reducing contact center costs, enhancing customer experience, decreasing hold times, and promoting self-service.

DoorDash partnered with CloudCookies, experts in conversational design, to transform their customer support operations. CloudCookies’ phased approach included improved intent recognition and optimized conversational design, leading to a successful transformation of DoorDash’s customer support operations. By modernizing their Amazon Connect IVR and leveraging Amazon Lex’s Natural Language Understanding (NLU) capabilities, DoorDash achieved significant cost savings and improved customer service. The partnership enabled CloudCookies to accurately identifiy the top 18 most common customer issues with more than 98% accuracy, delivering an additional annual savings exceeding $3 million.

:)

“CloudCookies took us from a proof‑of‑concept to a production‑ready, nationwide solution in eight weeks—and beat the target accuracy by 27 %. Our customers get answers faster, and our agents focus on the conversations that matter most.”

-Chaitanya Hari, DoorDash Product Manager

:)

:)

Customer Challenge

DoorDash’s U.S. support line handles millions of calls on Amazon Connect, but self‑service intent recognition stalled at roughly 50 % across 15 high‑volume call drivers. The misroutes fueled avoidable agent transfers, dampened first‑contact resolution, and drove up telephony spend. AWS tapped CloudCookies to uncover the root‑cause issues and engineer a scalable conversational‑AI blueprint—one that would push accuracy past 65 % and clear the way for a full nationwide deployment.

:)

:)

:)

Partner Solution

In the initial phase of the engagement, CloudCookies dedicated their efforts to enhancing the accuracy of customer intent recognition and routing. Their goal was to achieve a minimum accuracy rate of 75% for routing customer intents and an overarching engagement goal of exceeding 90% accuracy when comparing expected outcomes to actual results across 18 specific intents. These included tasks like updating account information, checking order status, refunds, unauthorized charges, general customer service requests, and resolving issues with orders.

To optimize efficiency, CloudCookies committed to an agile approach, targeting a rapid turnaround time for each phase. Their objective was to complete every phase within just two weeks, ensuring a swift deployment of improvements and solutions to enhance DoorDash’s customer support operations through:

• Rapid Voice‑of‑Customer Analysis
  • Ingested 4 MM historical call transcripts and disposition codes.
  • Re‑clustered utterances to validate intent taxonomy and uncover “hidden” sub‑intents.

• Conversational Design Overhaul
  • -Re‑authored Lex prompts, slot‑filling logic, and confidence thresholds.
  • Introduced dynamic re-prompts and intent disambiguation to minimize fallbacks.

• Training + Continuous Testing
  • -Generated synthetic data to balance low‑volume intents.
  • -Staged weekly A/B accuracy tests in production shadow mode until >90 % sustained.

• Routing Logic Refinement
  • -Updated Connect contact flows to honor new intent confidence bands.
  • -Implemented analytics pipeline that surfaces real‑time misroutes for quick tuning.

:

:)

The implementation process aimed for operational excellence, efficient tuning, and the promotion of self-sufficiency, with CloudCookies providing break/fix support and staff augmentation for support.

:)

:)

Why CloudCookies

Results

CloudCookies’ solution for DoorDash delivered outstanding results, showcasing the transformation of their customer support operations:

• 92 % intent‑matching accuracy (↑ 42 percentage points)
• 49 % fewer agent transfers — callers reach the right queue almost twice as often
• 12.8 % boost in first‑contact resolution — more issues solved on the first call
• $3 M+ annual operational savings by cutting handle time and transfer costs
• 15 % reduction in Amazon Connect & ancillary AWS costs
• 600 % ROI in the first six months
• 100 % U.S. customer coverage after nationwide roll‑out
• 27 % above the original accuracy target (goal 65 %; achieved 92 %)
• Foundation for future expansion into additional intents, self‑service flows, and channels

Building on this remarkable success, DoorDash has strategic plans to expand its self-service capabilities further, introducing additional languages, expanding the range of self-service use cases, and implementing new machine learning-based predictive intent and routing capabilities on a global scale. These results underline the substantial impact of CloudCookies’ solution in revolutionizing DoorDash’s customer support operations and fostering greater efficiency and customer satisfaction.

:)

About CloudCookies

CloudCookies is a distinguished Advanced Tier AWS Technology and Consulting Partner, specializing in business process automation, conversational design, and self-service solutions. We exclusively harness the power of AWS services to provide innovative solutions for our clients. Our services cover consulting, migration, integration, custom development, training, and support, ensuring our clients receive top-tier expertise. What sets CloudCookies apart is our extensive experience and deep knowledge, making us the preferred choice for those seeking transformative AWS solutions that break the mold.

:)