Knowing where you stand.
WHAT TO MEASURE
Time until meaningful content appears Time until the page responds to input Layout movement after appearing Total transferred size Number of requests
WHAT TOOLS PROVIDE
Laboratory measurement, under controlled conditions Field measurement, from real visitors
WHY BOTH
Laboratory tests are repeatable; field data reflects reality.
WHAT FIELD DATA REVEALS
That real visitors have slower devices and connections than you assumed.
WHAT TO TEST ON
A modest device, on a throttled connection.
WHY
Your own machine conceals every problem.
WHAT TO ESTABLISH
A baseline, before optimising.
WHY
Without it you cannot tell whether anything improved.
WHAT TO PRIORITISE
The largest contributor, measured rather than assumed.
WHAT PEOPLE USUALLY GUESS WRONGLY
Where the time goes.
WHAT TO MONITOR CONTINUOUSLY
Performance over time, so regressions are noticed.
WHAT CAUSES REGRESSIONS
A new dependency An added third-party script An unoptimised image
WHAT TO CHECK
Before every release.
