DoorDash has automated quality evaluations for nearly 100% of customer interactions across 19,000 agents in partnership with Observe.AI and Amazon Web Services (AWS). The move replaces manual sampling with full-coverage AI analysis, giving support teams real-time visibility into sentiment, behavioral patterns, and emerging product friction points.
The collaboration marks a shift in DoorDash's customer experience strategy-using conversational intelligence to understand what drives satisfaction, not just whether an interaction passed a checklist. By automating evaluations, the company freed human quality teams from routine scoring and redirected them toward higher-value analysis, such as identifying subjective issues and coaching opportunities.
Moving From Checkbox QA to Customer Insight
Traditional QA processes relied on limited samples and binary scores. As DoorDash scaled globally, that approach could not surface the real reasons behind customer frustration. The new platform, built with Observe.AI and using AWS transcription and AI infrastructure, evaluates interactions at scale and surfaces behavioral signals that manual reviews missed.
"We didn't want automation for its own sake. We wanted to better understand what drives customer sentiment, move beyond binary scoring, and put the customer at the center of our decisions," said Xenia Strunnikova, Head of Customer Experience, Fraud, Trust & Safety S&O at DoorDash. "This partnership enabled us to do that at scale."
The shift reflects a growing use of AI for Customer Support to move beyond compliance and toward diagnostic insight-identifying the behavioral drivers behind pain points rather than simply marking interactions as pass or fail.
From Days to Near Real-Time Hotspot Detection
Observe.AI's signals-including sentiment, comprehension metrics, and behavioral indicators-let DoorDash infer customer satisfaction even when feedback isn't explicit. These signals have compressed issue detection timelines dramatically. Product friction points that once took days or weeks to surface now appear in near real time.
"With nearly 100% coverage, we can proactively identify hotspots we might never have seen before. What used to take days now happens almost immediately," said Joaquin Dufeu, Director of Strategy & Operations for Customer Experience & Integrity at DoorDash. "That changes how we operate-and how quickly we can improve the experience."
Strengthening Responsible AI Adoption
DoorDash designed the system to augment human judgment, not replace it. Automation provides objective signals and full interaction visibility, which supports more consistent coaching, clearer alignment with BPO partners, and stronger accountability. The approach reinforces responsible AI principles by keeping people in the loop for subjective decisions and safety-critical assessments.
A Platform Built Through Iterative Partnership
The companies describe the effort as co-creation-building and refining signals together in cycles to match DoorDash's operational realities. "This was never about deploying a tool," said Deepak Kumar, Observe.AI Chief Customer Officer. "It was about building a platform alongside DoorDash that enables automation, diagnostic insight, and real-time intelligence at enterprise scale."
Why this matters for customer support professionals
DoorDash's move shows what becomes possible when QA shifts from sampling to full-coverage analysis. Support leaders can use similar AI-driven signals to spot emerging issues before they escalate, coach agents on specific behaviors rather than abstract scores, and demonstrate the concrete impact of quality programs on customer outcomes. The key takeaway: automation isn't about eliminating human judgment-it's about giving teams the data to apply that judgment more precisely and more quickly.
Your membership also unlocks:
