In addition, the steepness of the curve at different points can tell you how much you will gain in return by optimizing speed by a specific amount. in your trade-off decisions, because your time as a designer or engineer is also valuable. So the goal is not to find a single magic number, but to find useful ranges of values and reasonable guidelines in research. For example:
On unfamiliar sites, a 2-second delay was enough to cause most of the drop-off — familiar sites bottomed out after longer delays. Performance also suffered on unfamiliar sites, most often with a 4-second delay.
Another study involved navigating nested menus on a web page. A luxembourg mobile database series of delays, spaced 3 seconds apart, were tested as each panel loaded. Satisfaction dropped when the delay increased from 0 to 3 seconds, or from 9 to 12 seconds. Willingness to return also dropped with a 12-second delay. A 6-second delay was enough for some participants to say a site was slow.
One study found that mobile web users don’t focus on a screen for more than 4-8 seconds. This means that if they shift their attention before your page has loaded, the time they spend shifting their attention will further delay when they finally see the page. So a 5-second load time could turn into a 10-second effective delay.
It was argued that the system should react at a speed comparable to the latency experienced by humans when interacting. This led to guidance that responses should take approximately 1-4 seconds.