The appYdex definition, UX index for web performance

Our monitoring solution appYuser used to base on the Apdex norm to estimate users’ satisfaction toward web performance. This was close but we could do even better.

This page is a summary of the steps that lead to the creation of the new appYdex norm.

Apdex is calculated based on a target display time.

Users are sorted depending on the comparison between the loading time they experienced (Te) and the target time (Tt).

Above 8s, the user’s attention is quickly decreasing (sources Zona Research Group & MIT)​

It is a statistical indicator.

The Apdex score is calculated from the formula on the opposite. It depends on how many users there are in each category. The users’ global satisfaction are then defined as follow:


  •  Excellent from 0.94 to 1
  •  Good from 0.85 to 0.93
  •  Fair from 0.7 to 0.84
  •  Poor from 0.5 to 0.69
  •  Unacceptable from 0 to 0.49

But users' satisfaction is not just one time.

The beginning of the display is as least as important as the end.

Limiting the assesment to one or the other can lead to an erroneous estimate of satisfaction.

The best approximation would be the combined evaluation of both.

Hence the idea of combining two Apdex:

Start display Apdex

The satisfaction cannot be good without a good start display time.
A good start display often compensate for a bad total load time.
This time ignores the problems of the web page.


Total load time Apdex

The satisfaction cannot be excellent without a good total load time.
An excellent total load time can sometimes catch up a bad start display time.
This time takes into account network issues and web pages problems.

Introducing appYdex!

The appYdex is calculated on the basis of a complex mathematical model that combines the two Apdex; the user experience of beginning and end of display.
The result gives an index that ranges from 0 to 1 and is associated with satisfaction levels.
This model gives a very realistic estimate of the actual user satisfaction, many experiences on the Web Performance have confirmed this.

The appYdex model, the UX index of Web Performance and its different levels of user satisfaction.