Meaning
The time delay between a user or client initiating a request and the receiving system returning the response.
Three examples from work
Performance optimization
DevOur p99 latency increased from 150ms to over 800ms during peak hours.
LeadCheck for unindexed database queries and slow third-party webhooks.
Global deployment
ArchitectDeploying edge servers in Europe reduced round-trip latency by seventy percent.
POOur European enterprise clients immediately noticed faster page load speeds.
AI response streaming
Frontend DevGenerating full AI answers takes five seconds; how do we mask the latency?
LeadStream the tokens as Server-Sent Events so the time-to-first-token is under 300ms.
Measure latency using percentiles (p50, p95, p99) rather than misleading simple averages. Low latency is critical for interactive user experiences.