Outbound Call Queueing

[sallen@healthcall.com]

We recently had our Plum Voice Platform upgraded to version 3.1.1. We are trying to better understand how the changes to the campaign manager have affected call queueing.

It looks like resources are divided between campaigns with all campaigns having an equal chance to provide calls to the queue. For example, say there are 1000 calls scheduled for 10am in campaign one. There are another 1000 calls scheduled for 11am in campaign two. If the calls from campaign one are still going out at 11am, then calls from campaign two will beging going out at 11am in addition to the calls from campaign one. Is this correct?

What is the best way to schedule an immediate call? It looks like we would add it to a campaign with no scheduled calls with a scheduled timestamp of NOW or a time in the near future. It seems that if you add a call scheduled for 11am to campaign one when calls scheduled for 10 am are still going out for that campaign, the new call would not go until after the calls scheduled at 10am were finished. In the past we could schedule a call with a time in the past in the default campaign and it would go out almost immediately.

We're trying to replicate the monitoring system we had where we scheduled outbound calls in the default campaign for each hour of the day. If the call failed, an alert was sent. After the upgrade, not all of the hourly calls went out. We scheduled them for 15 minutes after each hour with an expiration timestamp 44 minutes later. If we schedule these calls in a separate campaign can we be assured that they will go out in a timely manner?

[support]

Hi Scott,

About question 1: Yes, this is correct. All outbound campaigns are weighted equally, with each campaign having a First In First Out (FIFO) queue. To clarify your example, at 10 am, the 1000 calls for campaign one will begin going out. Then, at 11 am, the 1000 calls for campaign two will begin to go out and take turns with campaign one once a port becomes available for making an outbound call.

About question 2: The best way to schedule an immediate call is to queue it to the "default" campaign. You wouldn't have to set a scheduled timestamp of NOW or a time in the near future as calls queued to the default campaign are by default, set to go out immediately (if no timestamp is provided) once a port becomes available for an outbound call.

About question 3: Could you please us with more details on how you have queued these calls to replicate your monitoring system? How are you setting up your expiration timestamp when making these calls? Also, when you mentioned that "not all of the hourly calls went out", do you mean that the calls weren't attempted at all? Is there a status returned for those call attempts? This information would help us better understand this issue.

Regards,
Plum Support

[sallen@healthcall.com]

The calls that didn't go out were not attempted. The status returned is failed. Here's an example row from the outbound_call table:
1332260,"failed",2,12194763459,"http://10.0.1.76/_VoiceXML/callmonitor/ ... 1291444201

The scheduled timestamp is 1291515300 and the expiration timestamp is 1291517940. They are scheduled to expire 44 minutes after the scheduled time. I believe they were queued to the default campaign.

[sallen@healthcall.com]

As I think about this, would the best way to make hourly, outbound monitoring calls be to schedule them as immediate calls each hour? We could run a script each hour from a cron job that schedules an immediate call using the method you described. We would schedule the immediate call in the default campaign with no timestamp.

[support]

Hi Scott,

Yes, this would be the best way to make those hourly, outbound monitoring calls as these calls would be queued up immediately to go out as soon as a port was available for making an outbound call.

Regards,
Plum Support

[sallen@healthcall.com]

I have tested sending a couple immediate outbound calls by scheduling them to the default campaign with no scheduled timestamp. The first one went out within 15 minutes. It has been 40 minutes since I scheduled the second one and it hasn't been made yet. My goal is to schedule hourly outbound calls to test that the server is functioning. However, I need to know how long to wait to check the call status for completed. Please look into this to see why the "immediate" call was not made as expected. The row from the outbound_call table is below.

1369838,"uncalled",2,12194763459,"http://10.0.1.76/_VoiceXML/callmonitor/ ... 1291997296

[support]

Hi Scott,

Our engineers have taken a look at this issue and have resolved a problem with one of your campaigns that was preventing queued calls from going out.

Please let us know if you experience any further issues.

Regards,
Plum Support

[sallen@healthcall.com]

Thank you for resolving the issue.

I have one more question. The immediate outbound calls were going out as expected until we got to the 3pm call. There are 2000+ calls scheduled in the default campaign starting at 2:30 PM. Now that those are going out, the latest immediate calls we have scheduled are sitting in the queue. Is that expected behavior? If so, what are our options?

I can think of a couple things to try and will do that for now.

[support]

Hi Scott,

Yes, this is the expected behavior. Our understanding is that you had initially queued calls to go out on the default campaign at 2:30pm.

However, since the outbound system is first in, first out (FIFO), those calls that you queued at 3pm are sent to the back of the queue for the default campaign and thus, would not be sent out immediately.

As for options, we would suggest queuing non-immediate calls in a separate campaign (not using the default campaign) and just queue your immediate calls to the default campaign.

Regards,
Plum Support

[sallen@healthcall.com]

Thanks. That makes sense. I appreciate your continued responses to my questions. This helps us define the expected functionality of the updated campaign manager in the Plum platform.