How do non Key User Actions effect an Application's overall Apdex?

[36Krazyfists]

I understand Apdex and how it's calculated in general.  What I don't get is how Dynatrace uses non Key User Actions to calculate the overall Apdex for an application.

I know you can set the KPM metrics for XHR/Load/Custom action of the overall application.  What I don't get about its use is the following:

- Does Dynatrace calculate an Apdex rating for each non Key User Action against the general Application-level Apdex threshold?  So, if an Application's XHR Action Threshold is 5/15 seconds (Tolerating/Frustrated) and a XHR non Key User Action action has an average of 20 seconds, it will be marked as Frustrated?

- I assume it uses a weighted formula versus the Key User Actions?  

[juan_mesa · Accepted solution]

Hey! The overall Apdex score for an application in Dynatrace takes all user actions into account, that means both Key User Actions and non-Key User Actions. So, if a non-Key action runs much slower than the threshold, it will still be marked as Frustrated and pull the overall Apdex score down.

Also, there isn’t a separate formula for Key vs. non-Key actions, Dynatrace uses the same Apdex formula across the board. The main difference is that Key User Actions let you set custom thresholds and track them more closely, but when it comes to the global Apdex, everything is counted together.

https://docs.dynatrace.com/docs/observe/digital-experience/rum-concepts/scores-and-ratings/apdex-rat...  

[36Krazyfists]

So that means that, if you want your overall application's Apdex to be even close to accurate, you have to mark all your most common actions as Key User Actions because, otherwise, they are just gonna use the application's overall threshold which could result in a lot of actions being incorrectly counted as either Satisfied/Frustrated.

Interesting.  I know with RUM on Grail, this will go away because there is no such thing as Key User Actions (but there's also no such thing as Apdex either, which is kind of a shame because it's a convenient way to manage thresholds and comparing performance across applications), but until then we'll have to be really diligent about marking our top actions as KUA's.