What TTFB measures
TTFB is the stopwatch between the moment a browser asks for a URL and the moment the first byte of the answer comes back. It is not one thing. Inside it sit the DNS lookup, the connection and encryption handshake, any redirects along the way, the time the server spends building the page, and the trip back over the network.
That mix matters, because two sites with the same number can have completely different problems. One may be running slow database queries. Another may have a fast server sitting a long way from its visitors. A third may be losing the time to a chain of redirects before the real page is even requested.
Why TTFB matters
Nothing can be drawn before the first byte arrives, so TTFB sets a floor under every other loading metric. First Contentful Paint and Largest Contentful Paint can only be as good as the wait that precedes them, and no amount of image or script work will push them below it.
It matters twice over when visitors sit far from the server. A business in Kathmandu hosted in the United States, or a Nepali company selling into Australia, pays a distance cost on every single request. Search engine crawlers feel it too: a slow-responding server is crawled more cautiously, which slows down how quickly new pages are found.
Where TTFB goes wrong
The usual cause on small business sites is oversubscribed shared hosting, where your response waits behind other people’s traffic. Next comes a page being rebuilt from scratch on every request because no full-page caching is in place, which is the normal state of an unconfigured content management system carrying a stack of plugins.
Measurement goes wrong just as often. Testing from your own office, in the same city as the server, produces a flattering figure that no overseas visitor will ever see. A logged-in test is misleading in the other direction, because it bypasses the cache and reports the slow path rather than the one real visitors get.
How to act on it
Work outwards from the server. Put full-page caching in front of it so most requests are answered from a stored copy, then examine what the uncached path is doing: slow queries, bloated plugins, an outdated language runtime. Once the origin itself is quick, distance is the remaining cost, and a content delivery network or a host in the right region answers it.
Then remove the avoidable hops — redirect chains, stray internal links that bounce through an old address, third-party lookups that happen before the page can respond. If the ceiling stays high after all of that, the platform is the constraint, and a hosting migration is the honest fix rather than another optimisation plugin.
