Lab vs real-user performance

Speed numbers come in two kinds, and they often disagree.

Lab tests A lab test loads your page once, on a simulated device and network, under controlled conditions. It is repeatable and great for spotting problems, but it is not what your visitors experience.

Real-user data Real-user data is collected from actual visitors' browsers over weeks. It reflects real devices, real networks and real behaviour, but it needs enough traffic to exist, and it changes slowly after you make fixes.

Why they disagree Your visitors may have faster or slower phones than the lab assumes. Caching, location and logged-in states also differ. A page can look slow in the lab and fine for real users, or the other way round.

Which to trust Use real-user data to decide whether you have a problem. Use lab tests to find the cause and check that a fix works. When real-user data isn't available, treat lab results as a guide, not a verdict.