New Data Model Metrics Examples

Talkdesk® New Data Model methods for curating call data allows for more granularity into your organization’s metrics. While powerful, these new methods also diverge from Talkdesk’s historical reporting. 

The purpose of this document is to compare how the new metrics and methods differ from the old reporting model in some specific examples. 

Note: This article is an in-depth accompaniment to the article New Contacts Data Model - New and Updated Metrics


Inbound Contacts

(Replaces inbound calls) Inbound contacts include all contacts that reach a queue, regardless of the contact’s final status (e.g., abandoned, short abandoned, answered, etc.).


These examples illustrate scenarios that would be counted as an inbound contact. Scenario two illustrates an abandoned contact, but it could also be considered a short abandon depending on the short abandon threshold setting. 


Date Filtering 

The old reporting model filtered inbound contacts by their end date and time, however, the New Data Model will now use their beginning date and time. 


For example, if an agent transfers a call at 1/17/20 23:59:30 and that contact finishes at 1/18/20 00:07:12, the old model would consider the contact to occur on Jan 18th. In the new model, this contact occurs on Jan 17th. 

The expected behavior is that this change will only impact calls that cross time intervals. Additionally, inbound contacts are not used for standard KPI (Key Performance Indicator) metrics, since they would include short abandoned calls. 

You will also notice an increase in inbound contact count relative to the amount of transfers that take place in your organization. The new model will report one additional contact per transfer. 

Note: We discourage aggregating this metric by phone number or other omnichannel points of contact (e.g., email address, SMS short code, among others).



The old reporting model made no distinction between interactions and contacts, however, you now have the ability to see queue-specific contacts per call by leveraging contact-level reporting. 

Using the same call as the prior example, a call transferred from Queue A to Queue B would have counted one contact for Queue B. 

Under the new model, this call would count as two contacts: one for Queue A and another for Queue B. 


Abandoned Contacts 

(Replaces abandoned calls) Abandoned contacts are inbound contacts where the caller disconnected after entering a queue, before getting connected to an agent.


Date Filtering 

As with inbound contacts, the New Data Model will now key off abandoned contacts’ beginning date and time, as opposed to the end date and time used in the old reporting model.


In this example, someone calls your organization and is routed to Queue A at 1/17/20 23:59:02 and then disconnects at 1/18/20 00:00:42. In the old model, this call would have been counted as one abandoned contact for Queue A on Jan 18th. Now, New Data Model will consider this to be one abandoned contact on Jan 17th. 

This change will be most noticeable in the answered metric for any queue the call went through before the final queue where the abandon occurred. Specifically, you should notice an increase in the answered count for all queues prior to the abandon. 

Tip: Evaluating this metric by interaction is useful for understanding the average customer journey; i.e., what is the likelihood that a customer will abandon when they contact the contact center, or while using phone channel vs. chat channel, among others. 

For understanding staffing and service level impacts of abandoned contacts, the new data model is the correct data to use.



Also, in-line with inbound contacts, the New Data Model offers queue-specific visibility for abandoned contacts. 

In this example, someone calls your organization at 1/17/20 23:55:02 and reaches Queue A. 

An agent answers their call and transfers to Queue B at 1/17/20 23:59:30. Subsequently, the customer disconnects at 1/18/20 00:00:56, before they’ve reached an agent. 

The old model would have counted one abandoned contact for Queue B with no consideration for the successful answer on Queue A. With the new version of Live, Talkdesk counts one handled contact on Queue A and one abandoned contact on Queue B.


Disconnected Calls While Ringing 

In the old reporting model, calls that disconnected while ringing an agent (i.e., after leaving the queue) were considered missed calls. The New Data Model considers calls that disconnect while ringing as abandoned contacts.


In this example, a customer calls your organization at 1/17/20 23:55:02. Using the IVR, they’re routed to Queue A where they enter the waiting queue. Talkdesk finds an available agent, begins ringing them, but the customer hangs up before the agent answers. 

The old reporting model would have considered a missed call, however, the New Data Model will consider this an abandoned contact. 


Service Level 

(New calculation method) Service level percent is the number of contacts answered within your organization's waiting time threshold. This value is configurable through Admin > Preferences as well as through numbers’ custom settings. 

The New Data Model will calculate service level as follows: 

[All contacts answered within threshold / 

(All inbound contacts - Short abandoned contacts)] x 100

Warning. This is a queue-specific metric. You should never attempt to measure agent-level dimensions using service level percentage.


Date Filtering 

Now that other metrics rely on contacts’ beginning date and time, service level calculation must likewise change. The New Data Model measures service level from the beginning date and time. 

The expected behavior is that this should only impact contacts and interactions that cross a time interval (e.g., 15 minutes, one hour, one week, etc). Additionally, changes to service level will become more noticeable the smaller interval you use. 


Wait Time 

The old reporting model added wait times together and associated them with at the interaction level. 

To increase reporting accuracy, the New Data Model preserves contact-level wait times to reflect where customers waited in their call experience. 

As an example, consider a call where a customer reaches your organization and is routed to Queue A. There, they wait for 40 seconds before ringing an agent. The agent picks up and transfers them to Queue B where they wait for another 15 seconds before reaching the second agent. 

In the old model, wait time visibility would stop at the call level for a total of 55 seconds (the sum of both wait times). 

In the New Data Model, the interaction’s wait time would be divided for each contact. Queue A would reflect a wait time of 40 seconds, and Queue B would reflect a wait time of 15 seconds. 


Abandons Included 

The old reporting model excludes abandons from your service level calculation, but this is now taken into consideration under the New Data Model. The expected behavior is that your organization will see a decrease in your service level to one degree or another now that abandoned contacts negatively impact it. 

Note: This does not include pre-queue abandons. In other words, Talkdesk will only consider contacts where customers would have any possibility of reaching an agent. 

For example, imagine ten customers call into your organization. Of those ten, two customers abandon their call. Of the eight remaining calls, seven are answered within the wait time threshold.

Using the old model, your service level percentage would be 87.5%, e.g.: 

[(7 calls answered within threshold + 0 calls missed within threshold)/ (8 total answered calls + 0 total missed calls)] x 100 

= 87.5%

With the New Data Model, your service level percentage is an expression of all contacts answered within threshold divided by the total number of contacts. With two abandoned contacts, and one contact answered outside of threshold, we’re left with 7 contacts within the threshold for a total of 70%. 

[7 contacts answered within threshold / 

(10 total contacts - 0 short abandoned contacts)] x 100 = 70% 


Average Handle Time (AHT) 

(New metric, no equivalent in old reporting) Average handle time (AHT) is a new metric that measures how long an agent is occupied while handling a contact for a given queue. Individual values for handle time are a sum of a contact’s talk, hold, and ACW time. The expected use case for this metric is for staffing and forecasting.


Note: “Hold time” above refers to the time when the customer is explicitly put on hold by an agent. It does not count the time spent in a waiting queue before an agent answers. 

As with other metrics, Talkdesk measures AHT based on the beginning date and time of each contact and queue in an interaction.


In this example, a call reaches your organization and is routed to Queue A and an agent answers their call. This agent transfers the call to Queue B where a different agent answers. The second agent ends the call without a transfer, thus ending the interaction. 

The handle time for Contact 1 begins at 1/17/20 23:56:49 when Agent 1 answers the phone (signified by the Talk Time segment). Handle time ends at 1/18/20 00:01:21 after Agent 1 concludes After Call Work. 

The handle time for Contact 2 begins at 1/18/20 00:01:33 Agent 2 answers. Handle time for this contact ends at 1/18/20 00:07:12 once Agent 2 is finished with After Call Work.


Additional Resources: 


All Articles ""
Please sign in to submit a request.