What is TTFB?
Time to first byte is the wait between a browser asking for a page and the first byte of the answer arriving. web.dev’s definition lists the phases it contains: redirect time, service worker startup time if there is one, DNS lookup, connection and TLS negotiation, the request, up until the point at which the first byte of the response has arrived. So it is not “server speed”. It is server speed plus the distance to the server plus the handshake, measured from wherever the probe sits.
That is why it is the metric hosting reviewers lean on: it isolates the part of page speed that hosting controls, before images, scripts and the theme add their own delay.
What is a good TTFB?
web.dev says most sites should strive to have a TTFB of 0.8 seconds or less, and treats it as a supporting metric for Largest Contentful Paint, whose “good” threshold is 2.5 seconds. For a hosting comparison the useful bands are narrower. Across the published measurements collected in our reviews, shared plans land between about 140 milliseconds (HostScore’s Flashcloud test from Singapore, one month, September 2026) and 745 milliseconds (HostingStep’s SiteGround window, eight months of US probes to May 2026). Under 300 milliseconds from a nearby probe is fast; 300 to 600 is ordinary; above 600 is slow for a cached page and worth asking about.
Why do reviewers get different numbers for the same host?
Because they measure different things and call them all TTFB. Three things move the number most:
- Where the probe is. A probe next to the server reports a few milliseconds of network time; one on another continent adds 100 to 200 milliseconds before the server does anything. HostingStep probes from 19 North American locations; WPBeginner averages six regions worldwide; Kripesh Adwani measures from India against servers he placed in Mumbai, Singapore or New York. Hostinger scores 227 milliseconds in WPBeginner’s six-region average and 447 in HostingStep’s US-only window, and both are correct.
- What page is requested. A cached home page returns from memory; an uncached one runs PHP and the database first. Reviewers rarely say which, and some tools bypass the cache.
- How often and how long. A single run on a quiet afternoon is a different measurement from an average of 349,920 checks over 243 days, which is what HostingStep’s eight-month windows contain.
The honest comparison is between measurements with the same method, which is why our reviews list every published figure with its tool, interval, plan and location, and why we do not average them.
What does hosting control, and what does it not?
Hosting controls the server’s thinking time (PHP, the database, the cache in front of them), the distance to the visitor (which data center you chose), and the connection setup (HTTP/2 or HTTP/3, TLS configuration). It does not control a slow theme, a plugin that queries the database forty times per page, or an image carousel, none of which appears in TTFB at all. A site with a 150 millisecond TTFB and a 4 second LCP has a theme problem, not a hosting problem.
How do you measure your own TTFB?
From a browser, the Network panel in developer tools shows the waiting time for the document request; that is TTFB from your chair. For a number you can compare over time, use a monitoring service that probes from a fixed location at a fixed interval and reports the average, which is what the reviewers above do and what HostIdea’s own windows will do when they open. One-off tests from a speed-test site are fine for spotting a problem and poor for ranking hosts.