QualityHQ – Rocket Mortgage

ROCKET MORTGAGE

Rocket Mortgage: QualityHQ

Executive Principal Product Manager · 2024–2026 · Enterprise · AI · Experience Design

Before QualityHQ, Rocket Mortgage's quality team could review roughly one percent of all customer calls. The rest were invisible. QualityHQ brought AI-powered scoring to the full call operation, shifting analysts from manual review to confirming and acting on work the system had already done. I led product strategy and experience design as a separate workstream from AgentIQ, with a dedicated product owner and engineering team, building on the same platform and data foundation.

CONTEXT

From one percent coverage to twenty-four. At scale.

QualityHQ is built on the standardized call records AgentIQ generates automatically after every interaction. Those records gave the quality team something they had never had: a complete, structured, analyzable picture of the full call operation. QualityHQ is the product that turns that data into action. The two products share a platform but serve entirely different users with entirely different goals.

THE COVERAGE PROBLEM

Analysts listened to and scored calls by hand, one at a time. At that pace, coverage was approximately one percent. Quality problems could persist across the other ninety-nine without anyone knowing.

AI scoring changed the math. At initial rollout, QualityHQ was automatically scoring twenty-four percent of all calls. Analysts were no longer starting from zero. They were reviewing, confirming, and where needed elaborating on scoring the system had already completed.

FROM SCORING TO CONFIRMING

An analyst who previously listened to a call in full and built a scorecard from scratch now arrives at a review that is already structured. The AI has scored the call, flagged relevant moments, and surfaced areas that need attention.

The design challenge was making that feel like a gain. Analysts needed to feel their judgment still mattered, because it does. The interface gave them clear visibility into how the score was reached, the ability to confirm or override any element, and room to add qualitative context the automated scoring cannot capture. Their expertise was redirected, not replaced.

ROLE-BASED EXPERIENCES

QualityHQ serves three user groups with different responsibilities and different relationships to the data. The product is role-based throughout.

Individual analysts worked from a dashboard of assigned agents and pending reviews: what needs attention, how complete each review is, and what quality issues exist for each assigned agent.

Managers needed a broader view: aggregate performance across the analysts and agents they oversaw, with the ability to identify patterns and see where coaching was needed.

Leadership had a different relationship with the data entirely. Rather than task-oriented dashboards, their experience was built around a conversational AI interface. They could ask natural language questions about quality trends, agent performance, and operational patterns across the full system, and get answers grounded in the underlying call data.