What time to first byte measures
The clock starts when the browser asks for a page and stops when the first byte of the reply arrives. Everything in between is included: finding the server through the domain name system, opening the connection, negotiating the secure handshake, the server’s own thinking time, and the journey back. In practice two things dominate — how far the visitor sits from the server, and how much work the server has to do before it can say anything at all.
On a database-driven site such as WordPress, that work means running theme and plugin code and querying the database to assemble the page. When the page is cached, the server hands over a file it prepared earlier and answers almost immediately. When it is not, it rebuilds the page from scratch for every visitor. The gap between those two states is usually the single largest number in the whole measurement, which is why caching is the first thing to look at.
Why time to first byte matters
Nothing on the page can begin until the first byte lands. The browser cannot parse markup it has not received, cannot discover the stylesheet, cannot start downloading the hero image. Every later milestone inherits the delay, including Largest Contentful Paint, which is one of the three Core Web Vitals.
It also shapes how much of your site gets crawled. Search Console reports an average server response time in its crawl statistics, and a server that responds slowly is visited less enthusiastically. For a large catalogue or a site publishing regularly, that is a real constraint rather than a theoretical one.
Common mistakes with TTFB
The most common is measuring from the wrong place. A test run from a data centre near the origin server flatters it; the same page requested on a mobile connection in Kathmandu, or by a customer in Sydney, tells a different story. Test from where your customers actually are.
The second is misdiagnosis. Teams compress images and defer scripts for weeks while the real delay happens before a single pixel is requested. The third is trusting a content delivery network to fix everything: if it only caches images and stylesheets while the HTML is still built on a distant, overloaded origin, the first byte still waits for that work. Redirect chains add a whole extra round trip each, and on shared hosting a neighbouring site’s traffic spike can slow your responses without anything on your side changing.
How to act on it
Measure the same page cached and uncached, from a location that matches your audience, and compare. If the cached response is fast and the uncached one is slow, the fix is caching and code efficiency. If both are slow, the problem is the server or the distance, and no amount of front-end work will help.
From there the moves are ordinary: cache full page output rather than assets alone, put a delivery network in front that will serve HTML, remove unnecessary redirect hops, keep the database and plugin list tidy, and keep server software current. When the plan is simply undersized or in the wrong region for the audience, the honest answer is a better plan or a move, which is what hosting and migration work exists to handle.
