3D AHT: Unlocking the Secrets of AHT


How many times in your contact centre career have you been asked for explanation of average handle time (AHT) variance and struggled to provide a reason with confidence? It’s easier when variance delivers an AHT reduction. Great agent coaching or successful initiative delivery can easily be attributed. But when AHT increase arises, when it hasn’t been sought, explanations can become vague and tenuous at best.

So why after 30 years of Contact Centre evolution do we still not know the reason why AHT varies? The answer is simple; contact centre systems that provide interaction handle time data necessary for variance analysis do so from a customer routing perspective, and not from a perspective that attributes the content and purpose of the interaction. And if both sets of data are available, it is rarely presented in consolidation to support meaningful performance analysis.

We can usually see within which subset of customers AHT has varied (ACD Skill), but we usually cannot see which call types within that subset are at the root of the variance.

Emerging contact centre technologies, such as the QPC Management Information Gateway (MIG), are set to change how we analyse AHT variance in a way that provides timely, specific, and actionable data to drive contact centre performance improvement with 3D AHT reporting.

But before we consider 3D AHT, let’s define and examine the limitations of existing 2D AHT data models.


Most existing telephony systems provide Operations with one of two views of AHT:

  • Call data x Skill data
  • Call data x Agent data

In some rare cases the telephony system will provide all 3 dimensions for analysis … but very rarely. This is what we will refer to as 2D AHT, the analysis of the AHT by 2 key dimensions Call data x Skill / Agent data.

2D AHT fails to reveal the detailed answers we seek because it relies upon a single assumption:

  •  All calls are about the same thing

It’s only when this assumption is true, that Agent performance can be the key reason for AHT variance as all other variables should have a negligible impact on variance in this scenario. This concept is best understood if we compare a call centre to the production line of a car manufacturer. The factory will mass produce the same car throughout the day and as such requires the components and materials to be the same throughout production cycle, otherwise the outcome is not consistent. In an ideal world Contact Centres would have this level of consistency. We’d have lots of very happy customers if this was the case. The reality is quite the opposite and for one reason … there are lots of types of calls/conversations customers want to have with us, so the assumption doesn’t hold true and a different approach is required.

3D AHT – Comparing Apples with Apples

3D AHT - apples and oranges

Imagine for a second we are Scientists and we’ve been given both an Apple and an Orange. Our hypothesis is that the Apple is different to the Orange. How would we prove this to be true? You could argue it’s fairly obvious they are different. They have different textures, smells, coloured skins, seeds, edible/non-edible skins and not least forgetting tastes. The burden of proof all points to the hypothesis being true. Not surprisingly enough, an Apple is not the same as an Orange. So following this through, if you were then presented with a crate load of mixed Apples / Oranges and the crate was marked as Apples, you would know straight away it was mislabelled and not take the label description for granted. Why then when presented with an AHT do we accept that we have a lorry load of Apples and nothing else?

3D AHT – Breaking down the AHT into Call Types

By combining the Call statistics from the telephony platform with the Call Type, we create a new dimension in AHT which allows the Analyst to allocate the handle time to the appropriate call type and thus separate the additive qualities of AHT into the appropriate “buckets” and for the first time allow for a true comparison of variance within AHT because we are comparing the same call shapes and drivers.

The diagrams below provide a visual representation of how, the new dimension of Call Type / Category allows for a more appropriate allocation of handle time.

Diagram 1 – The current view provided by 2D AHT

2D AHT graph

This is a typical handle time distribution which shouldn’t look unfamiliar; the AHT of this distribution is approx. 229s, pretty much the visual peak of the distribution. It’s important to note that even with Skill level data from the telephony platform, the reality is that all call centres still receive a mix of call types within those Skill groups. Customers will often find the shortest route into the call centre, so don’t assume that just because you have a “Sales” skill, it means they’re all Sales calls.

Diagram 2 – 3D AHT: How breaking out the Call Type / Category uncovers the underlying factors to 2D AHT

3D AHT graph

The diagram is intended to demonstrate the “hidden” impact of the specific call types on the shape of the aggregated 2D AHT. This demonstrates, in a simple way, how the different Call Types all have different handle time distributions. Without this information, each Call Type contributes in an additive way to the Total AHT. Imagine you had an agent who “randomly” handled only Sales calls and as such their AHT was around the 171s point, when you compare their performance with an agent who “randomly” handled only Billing calls with an AHT around 304s, on the surface you would look unfavourably upon the “Billing” call agent as their AHT is nearly double the “Sales” call agent, but you obviously wouldn’t be accounting for the mix of calls they have both handled individually.

Table 1 – 2D AHT example by Agents

2D AHT table

The above table shows a typical view of 2D AHT by agent. In this scenario it would not be untypical of the Team Leader to focus on the agents whose AHT is higher than their peers and target them to reduce their AHT in line with the rest of the team. In some cases this could be the start of a “Performance Improvement Plan” which may eventually lead to disciplinary action. Consider Agent 8, they are the currently one of the “worst” performers in terms of AHT, if we take a traditional approach.

Table 2 – 3D: The “hidden” context of AHT

3D AHT table

By allocating the handle time to the Call Types, we start to see a different picture in relation to the individual performance of agents for the different Call Types. This 3D view highlights more effectively the reality that all agents struggle with some Call Types but as this dimension of AHT is hidden, you are not as targeted or focused in the management of agent performance.

Being able to measure it, allows us to control it.

Look again at Agent 17, before our data highlighted they had almost the highest AHT in the team. Now, with 3D AHT, we can see that of the 7 Call Types Agent 17 handled, they struggled with 3 of those 7 Call Types (Cancellations, Enquiries, Refund), were within the acceptable range for 3 Call Types (Billing, Sales – Marketing Promotion, Technical Faults) and one of the most efficient performers on Complaint calls. It becomes clear that with some additional training on their 3 highest Call Types, Agent 17 would have a significant reduction in Overall AHT, bringing them in line with their peers (they are no longer an outlier). Taking this concept one step further; a number of agents “fly under the radar” due to their 2D AHT being within an acceptable range. But with the additional dimension of Call Type, you would “unhide” the fact that 5 other Agents in the team (Agents 1, 2, 8, 10 & 18) would also benefit from a “Refunds” training course. By targeting coaching and development effort into the specific Call Types by Agent, you will benefit from a more effective use of your coaching resources into the specific areas that will impact performance the most.


In order to control and reduce AHT variation in a Contact Centre you must first be able to isolate and then allocate the handle time to each Call Type. 3D AHT brings Call Type into the AHT calculation as a new dimension, which easily identifies the specific areas for targeted performance coaching, focusing your coaching efforts/time towards the Call Types which will have the greatest impact on performance improvement.

4D AHT Teaser

What if you could also analyse AHT using the final Outcome / Voice of Customer as a 4th dimension! Was the customer satisfied with the service, if so how did they score in performance? Did the customer buy something? Did the customer cancel their account, if so why? How does interaction success versus with failure impact AHT?

We’ve all at some time wrestled with the debate around, should we spend more time on the first call trying to successfully resolve, or should we just focus on shorter AHT to reduce costs. With the 4th dimension of AHT, you could finally understand how AHT impacts the total end to end experience of a customer when measured against tangible outcomes.

Gary Smith, Director of Product Management QPC

Gary Smith is a guest blogger from our European office who is able to share insights into emerging technologies currently being developed by QPC to enable far greater visibility and application of AHT and other contact centre metrics for effective performance management.