A Data-driven approach to CX and call center optimizations

ROLE

UXR Design Manager

DURATION

6 Month

SERVICES

Team management, Stakeholder Interviews, Competitive Analysis, Journey Mapping, Wireframes, UX, Mockups, UI, Prototypes, Design Ops, User Research, Data Analysis, A/B Testing

CHALLENGE

Google Store's contact center costs were climbing. The content and tooling meant to stop that weren't doing their job.

STRATEGY

I led a cross-functional research team in a mixed-methods investigation — merging agent transcript analysis, user journey mapping, and hypothesis-driven design — to find where the experience was breaking down.

IMPACT

Identified $7.1M in process savings, reduced call volume and contact center costs by 50%, and delivered a self-service WiFi troubleshooting workflow that attracted media coverage.

CHALLENGE

High expectations, stagnant escalation rates

Millions of Google Store customers encounter support moments every year. A device that won't connect. A setup that stalls. A question the manual doesn't answer. The infrastructure built to handle those moments included help center content, self-service tools, and escalation paths. On paper, it was comprehensive. In practice, too many customers were bypassing it entirely and reaching for the phone.

Something wasn't working. The challenge was that the problem didn't announce itself cleanly in the data. It had to be reconstructed from the signals customers were leaving behind.

STRATEGY

Building the right research operation

I managed a multi-disciplinary team of eight — UX researchers, an SEO specialist, and a data analyst — embedded in a broader EPAM engagement spanning 19 Google stakeholders across 44 concurrent projects. The challenge wasn't just finding answers — it was designing a research approach that could surface meaningful signal in an environment with significant access and privacy constraints.

A visual of our program level approach to working with more than 15 different Google stakeholders at once

Transcript analysis as behavioral archaeology

Direct user access wasn't on the table. Privacy constraints limited what data we could touch. So I directed the team toward what was available: agent transcripts, journey data, and behavioral signals that customers had already left behind without knowing it.

Transcript analysis told us what customers were saying and where they were giving up. Journey mapping told us what the experience looked like from their side of it. Together, they produced a picture of the support system that no single dataset could have generated alone.

A structured approach to a multi-signal problem

The core investigation combined agent transcript analysis with user journey mapping. Transcript analysis surfaced recurring language patterns and failure moments — the specific points where customers stopped trusting the experience and reached for the phone. Journey mapping gave those moments structural context, revealing how individual friction points connected across the full support arc.

For validation, I implemented unmoderated user testing through Maze, working within PII and access constraints that prevented direct user contact. Every methodological decision was made in service of maintaining evidentiary standards under real-world conditions.

Hypothesis development and prioritization

Findings were translated into a discrete set of hypotheses, each mapped to a performance metric and scored on impact and effort. The matrix identified quick wins — those became designed mockups, ready for product teams to act on in the near term. The remainder formed a structured backlog, sequenced by priority and tied to measurable outcomes.

In addition to call transcripts, my team analyzed current user journeyes  (like contact us pages) to identify areas for optimization.

Operational resilience under shifting conditions

The broader account required more than research leadership. As Google reorganized internally, I worked at the partner engagement level to recalibrate resource allocation, protect commitments, and expand evaluation scope where possible — delivering increased velocity at a moment when the client environment was least stable.

On the account level, I helped kickstart the conceptualization of an operational model to grow the account from project to practice level headcount

On the team level, my goals were to align the UXR team on design requirements, breakdown the work, track/support workstreams, and manage team bandwidth

Facilitating regular team ceremonies like ‘Bi-weekly 1:1s’ and ‘Monthly team retros’ gave valuable insights into opportunities to maximize team efficiency

Foster team collaboration in a cross functional and distributed team (US + LATAM)

IMPACT

$8M+ total savings

Testing plan alignment and near term rollout of a/b testing was a key result from the initial phase

Call Volume Down

Reduced call volume and center costs by 50%, driving $400,000 in savings and customers are now empowered to troubleshoot their WiFi devices on their own with a workflow that attracted media attention!

Testimonial

“Eugene has been a great partner in identifying actionable UX opportunities to improve the Nest and Pixel customer support experience. He was able to provide valuable recommendations and quantify the impacts of proposed changes that will reduce contact volume and improve overall customer satisfaction.”