What is Latency?

Latency is the delay between a request and a response, often measured as round-trip time (RTT) in milliseconds. It includes signal propagation, serialization on the wire, queuing in routers, server processing, and return travel. Low latency feels instant; high latency feels sluggish even when bandwidth is ample.

What contributes to latency?

Propagation delay follows physics: longer fiber routes mean more milliseconds. Serialization delay depends on packet size and link speed. Queuing delay grows when routers are congested. On the web, DNS lookup, TCP and TLS handshakes, and server time-to-first-byte add to perceived delay before content renders. Each hop can add variability; jitter is the spread of delay samples over time.

Bandwidth is different: it caps throughput, not single-packet RTT. A fat pipe with a long path can still feel slow for interactive work.

Latency versus throughput and jitter

Throughput is how much data arrives per second once the pipe is full. Jitter is how much latency fluctuates. Real-time apps (voice, gaming) care about both low latency and stable jitter; batch downloads care more about throughput.

Latency and invalid traffic on ads

Humans on residential links often show RTT and jitter distributions that differ from data centers, proxies, or distant VPN exits. Fraud systems may use timing alongside other signals to flag suspicious behavior or bot-like automation. Latency alone never proves fraud, but it helps context (for example, impossible geography versus measured RTT).

Slow landing pages also depress conversions, so marketers optimize Core Web Vitals and hosting. That is separate from abuse yet interacts with campaigns: if click fraud or ad fraud wastes budget, fixing speed without detection leaves the underlying invalid traffic in place. See also suspicious clicks for how platforms may classify traffic.