I'm working on optimizing our monitoring for PureCloud as we currently have some issues where the edges get stuck in some weird state and calls are not routed to our agents. (This is an intermittent behavior and is currently investigated with the support team so the point of this post is not to resolve the whole thing).
We noticed that the temporary solution for this issue is to reboot the edges but we only apply it once the service is noticeably bad and end-users start reporting via email/chat that they can't connect with an agent over our hotline. Our goal is to have proactive monitoring of the number of abandoned calls and to take actions if we reach a certain threshold.
I've tried using the analytics query builder but I can't really achieve the output we look for. Perhaps I'm not using the builder properly but before I dig in further, I would like to confirm with you if the things I want are possible at all.
To summarize what we need:
Live tracking of the number of abandoned calls for each queue
We would like to receive a notification if we reach a pre-set threshold. (This is not a top priority as I have an idea of how to implement this with a third-party system or Zendesk)
I didn’t find a similar feature in the standard admin web interface and therefore I started looking into the API. Please let me know if I missed something there as it would be quite easier to use it instead of building something from the ground up.
You could use the notification service and subscribe to the v2.routing.queues.{id}.conversations.calls topic. Any conversation where the acd and customer participant disconnect and the conversation ends without reaching an agent is considered abandoned.
If you are unfamiliar with the notification service then the notification service overview is a good place to start. This tutorial subscribes to a similar topic using JavaScript and might be helpful.
I think that the oWaiting metric might be useful in our case since the calls which end up as abandoned are left waiting in the queue. So for example, if we notice an abnormal amount of waiting calls, we can proactively reboot the edges. I'm not sure how to get a summarized value for all queues in the org - is there a guide for that too?
I would also agree your waiting use case could work as well, if calls are having issues routing due to edge issues then they could backlog in the waiting queue.
For the customization route, we do have an event bridge topic for metrics https://developer.genesys.cloud/notificationsalerts/notifications/available-topics#v2-analytics-conversation--id--metrics The caveat to this is the metrics will come in raw meaning you'll get singular values for each conversation that abandons. This route won't do the aggregation for you, so if you wanted to trigger an alert for an abandon count going above a certain threshold, you'd need to keep track of the aggregation on your application end and fire the alert based on your own application's logic.
