What to measure
REST: time to first byte. WebSocket: inter-message gaps and clock skew. Always test from app servers, not only browsers.
Common mistakes
One-off browser tests misrepresent server-to-server paths. Sample many times and report P50/P95.
TIQEX latency tool
The on-site tool measures browser timing to TIQEX public endpoints only—not a multi-vendor leaderboard. Retest from your server region.
Baselines
Fix symbol, window, and sample size; measure REST TTFB and WS gaps; record egress and plan tier.
Tool
Use /tools/latency-benchmark, then compare with your own harness.