Time to first byte
The delay between sending a request and the first byte of the response arriving, the performance number that isolates the proxy path from download size.
Time to first byte measures the gap between your request leaving and the very first byte of the answer coming back. It is a more useful number than total response time for diagnosing a proxy, because it isolates the round trip and the server's thinking time from how long the body then takes to download. A slow TTFB points at the path or the origin; a fast TTFB followed by a slow transfer points at bandwidth or page weight instead.
Through a proxy the figure decomposes into parts you can reason about. It includes the leg from you to the proxy, the leg from the proxy to the destination, the destination's own processing, and the connection setup on each hop, the TCP handshake and the TLS negotiation, each of which costs a round trip the first time. This is why a fresh connection through a distant exit has a much worse TTFB than a reused one, and why measuring a single cold request overstates the steady-state cost.
It is also where the geometry lesson from latency shows up as a number. If your exit is far from the destination, every request pays that distance before the first byte, so TTFB is the metric that makes an unnecessary routing triangle visible: put the exit near the target and the first byte arrives sooner, because the leg that repeats is shorter. A high TTFB with a nearby exit, by contrast, points at the origin or at connection setup rather than at distance.
The practical use is comparison, not an absolute target. Measure TTFB across exits and across providers against the same destination, prefer the ones that reuse connections, and separate it from download time so you are optimising the right half. A pool with a low, consistent TTFB is doing the part a proxy controls well; the rest is your target and your bandwidth.
Frequently asked questions
What is a good time to first byte for a proxy?
There is no absolute figure, only one relative to the route. Compare against the physical floor for the distance and against a direct connection to the same target: a nearby exit should add little, while a transcontinental one pays the round trip before the first byte. Judge it by comparison across exits and providers, not against a fixed number.
Why is my proxy's TTFB so high?
Common causes, in order: a distant exit paying the round trip to the target, a cold connection paying TCP and TLS setup that a reused one would not, or the origin itself being slow. Measure a warm connection rather than a single cold request, and check the geometry, since an exit far from the destination inflates every first byte.
Is TTFB the same as latency?
Related but not identical. Latency is the round-trip delay of the connection; TTFB is the time until the first response byte, which includes that latency plus connection setup and the server's processing time. TTFB is the more complete measure of how quickly a request starts returning, while raw latency isolates just the network round trip.
Back to the full glossary.