Enterprise SaaS · Communications Platform

Multi-language voice support, shipped on infrastructure that scales.

The challenge

ZVA was live with paying enterprise customers, and a global multi-language rollout across eight or more locales was imminent. The engineering team had no automated end-to-end voice testing infrastructure. Manual testing couldn't cover the scenario breadth, couldn't reproduce customer-reported bugs, and offered no regression coverage between releases.

Leadership needed objective performance data for both internal prioritization and external positioning against competing voice agent stacks. Building that internally would have taken months. The team needed it in weeks.

No automated end-to-end testing.

Manual testing couldn't cover multi-language deployments or reproduce customer-reported bugs at scale.

No regression coverage.

Engineers had no traceable path from QA result to root cause when issues surfaced in production.

No objective benchmarking data.

Leadership needed third-party performance numbers against competing platforms.

Multi-language scale problem.

Eight or more locales meant the testing surface multiplied with every new language ZVA shipped.

The impact
22,171 Simulations run in 174 days
261,652 Metric executions over the same period
70 Distinct test sets built
<8 weeks From first demo to signed contract
Before and after
Metric Before Coval With Coval
Test infrastructure Manual, scenario-limited 22,171 simulations in 174 days
Multi-language coverage Inconsistent across 8+ locales Test sets per language with automated regression
Bug detection Reactive, surfaced in production Proactive, caught before customer exposure
Benchmark data Months to build internally Objective benchmark vs. competing stacks in weeks
Vendor relationship Engineering tool evaluation Exclusive ISV partnership for ZVA deployments
How Coval helped

Zoom's engineering team went from demo to signed contract in under eight weeks because the alternative — building this infrastructure internally — would have taken months and missed the multi-language launch window entirely.

01

Objective benchmark data leadership could act on.

Coval ran ZVA against competing voice agent stacks on identical test sets, producing benchmark data Zoom's leadership used directly for internal prioritization and external positioning. The kind of output that would have taken months to build in-house arrived in weeks.

02

Reactive bug-hunting became proactive detection.

Within thirty days of contract start, the team moved from finding bugs through customer reports to catching them in simulation. Production-blocking issues were surfaced before customer exposure, including localization regressions that would have surfaced across multiple locales at once.

03

From engineering contract to strategic partnership.

The engineering contract was signed in November. By December, Zoom had formalized Coval as the designated testing infrastructure for ZVA deployments. Zoom is now actively co-selling Coval into its own enterprise customer base.

The result

What started as an engineering contract became a strategic partnership in four months. Zoom's ZVA team ships across eight or more languages with a regression suite that grows with every new test set, and a benchmark methodology that holds up under leadership scrutiny.

"Coval gave us the testing infrastructure that would have taken us months to build, in weeks. We moved from reactive to proactive detection inside thirty days."

Coval is now the designated testing infrastructure for ZVA deployments and the foundation of a co-sell motion bringing Coval into Zoom's own enterprise customer base.

See what Coval can verify on your voice and chat agent.

Book a Call