Why it matters on an HVAC site
Most contractor sites are WordPress on shared hosting, and shared hosting is where TTFB goes wrong quietly. It is acceptable on a Tuesday morning and poor in a heatwave, which is exactly when the traffic is worth the most.
It is also the one number an agency cannot fix with front-end work. If the server takes 1.4 seconds to respond, no amount of image optimisation gets LCP under 2.5.
How to check it yourself
- Field data from the Chrome User Experience Report reports TTFB separately.
- Measure repeatedly across a day rather than once, since the failure is load-dependent.
- Compare a cached page against one that always misses the cache, such as a search result, to separate hosting from caching.
Worked through in numbers
A site measures TTFB at 380ms most mornings and 1,900ms between 2pm and 6pm in July.
The host was shared, and the sustained load of the season was enough to queue requests. Page-level caching was in place but the booking page was excluded from it, and the booking page was where the paid traffic landed.
Moving to a host with guaranteed resources took the afternoon figure to 410ms. That single change took 1.5 seconds off LCP on the worst-performing page without anyone touching the page.
Where this sits in the rankings
Scored under Technical SEO. How long the server takes to start answering.