as per these test-results https://www.webpagetest.org/graph_page_data.php?tests=161227_B7_YAE&medianMetric=loadTime&fv=1&rv=1&control=NOSTAT:
- the TTFB for “repeat view” is considerably and consistently (5 test-runs) higher then for “first view”, which is weird.
- but for “time to title” in the same test, “repeat view” is faster then “first view”.
- to make matters even more fishy, for “repeat view” TTFB > TTTitle.
is this an anomaly in the test result? a bug in wpt? or is there a perfectly logical explanation for this?
It’s a ‘quirk’ of the test TTFB in the repeat view isn’t measuring the same thing as TTFB in the first view.
In the First View TTFB is measured on the response for the 1st request i.e. https://guides.wp-bullet.com/
In the Repeat View it’s measured for https://guides.wp-bullet.com/wp-admin/admin-ajax.php?action=get_recently&widget_id=5 which is request #21 in the first view but is the first resource that can’t be retrieved from cache in the Repeat View
As TTFB measures the first byte from the network this is the ‘right’ thing to do but it means it’s not a very useful metric for comparing first and repeat views in this case.
that indeed makes sense Andy, thanks!