Ideally, from day one. The earlier you involve QA in requirements and design discussions, the fewer defects you'll create. That said, it's never too late to improve quality practices.

It depends on your stage and needs. Early-stage startups often benefit from outsourced QA to access expertise without full-time hiring. As you scale, hybrid models (internal QA leads + outsourced execution) work well.

There's no magic number, but a good rule: automate repetitive regression tests, critical user journeys, and API tests. Keep complex exploratory testing manual. Aim for 70-80% automation coverage of regression scenarios.

A basic automation framework with initial test coverage typically takes 4-6 weeks. Building comprehensive coverage is ongoing—you add tests as features are developed.

Yes. Legacy systems are often harder to test (poor documentation, technical debt, limited testability), but that makes testing even more critical. We've tested mainframe applications, decade-old codebases, and everything in between.

QA (Quality Assurance) traditionally focused on testing and defect detection. QE (Quality Engineering) takes a broader view—building quality into processes, automation, CI/CD integration, and production monitoring. We practice QE.

We track multiple metrics: defect detection rate, escaped defects (found in production), test coverage, automation coverage, mean time to detect defects, and ultimately—customer-reported issues. Quality improves when these trends move in the right direction.