Estimated wait time in Skill based routing
Updated
How you can leverage estimated wait time and handling time for cases to improve case resolution time and overall customer satisfaction.
You can view and share an expected wait time with your customers for a case or a message in the Assignment Work Queue. This information is sent out to customers via auto-response. This will enable the customers, who have requested to engage with a live agent, to have realistic expectations for time to receive a response.
Estimated Work Queue Wait Time - It defines the approximate wait time of the case/message that is waiting in the work queue (Not assigned to the agent).
(Work Queue - It is like a bucket where all cases are put for assignment and when users associated to that work queue get free, cases are assigned to them.)
Average Handling Time - It is a metric that calculates the average time duration that the agent is taking to handle one case or message.
Note:
Estimated Wait Time for a work queue for the 1st case is always set as null.
This article covers how Estimated Wait Time is calculated in Unified Routing.
How Estimated Wait Time is Calculated
Estimated Wait Time can be calculated by two methods:
Work Queue Wait Time Method
User Handling Time Method
Work Queue Wait Time Method
Work Queue Wait time is the average of the time duration of message or cases that are waiting in the Work Queue. Single Exponential Smoothing is used to predict the expected average wait time when any case comes for assignment. When a case is put for assignment in the work queue, it will wait there until it is assigned to an agent. Expected Average Wait Time will be given to the customer for that case. Single Exponential Smoothing computes the running average of the time that recent cases have spent waiting in the assigned work queue. More weightage is given to recent cases compared to old ones.
Mathematically speaking, to calculate estimated wait time for case number n, when n-1 cases have already been assigned, we use the formula :
Estimated Wait for nth case = α * Actual Wait Time of (n-1)th case + (1- α) * Estimated Wait Time of (n-1)th case |
Therefore, it accounts for actual wait time and error wait time in predicting the last wait time. Here, Alpha (α) is a smoothing parameter that defines the weighting and should be greater than 0 and less than 1, which you can put after optimizing from your end.
For example, to calculate estimated wait time for case number 10, when 9 cases have already been assigned, using the formula we get-
Estimated Wait time for 10th case = α * Actual Wait time of 9th case + (1- α) * Estimated Wait time of 9th case
This can be also be explained via another example too.
There are 7 cases already assigned in the assignment work queue and a new case number 8 comes in. You can calculate the estimated average wait time for case number 8 using the above formula:
Estimated Wait for 8th case = 0.9 * 10 + (1-0.9) * 10.6 = 10.06 ≈ 10.1 mins (Here α is assumed 0.9)
Case Number | Actual Wait Time | Estimated Wait Time |
1 | 9 |
|
2 | 10 | 9.0 |
3 | 8 | 9.8 |
4 | 9 | 8.4 |
5 | 9 | 8.9 |
6 | 11 | 9.0 |
7 | 10 | 10.6 |
8 |
| 10.1 |
So, you can get the estimation which is very close to the actual value if the actual wait time is not facing any outliers or sharp trends.
Bootstrapping for Subsequent Cases
In case you want to calculate the estimated wait time of 9th case and beyond, when only the first 7 cases are actually assigned, you can use the bootstrapping method as explained below:
Estimated wait time 9th case = α *actual wait time 7th case + (1-α)*estimated wait time 8th case
Estimated wait time 9th case = 0.9 *10+ (1-0.9)*10.06 = 10.006 (10 mins)
And so on for subsequent cases.
User Handling Time Method
User Handling Time is the method through which you can calculate the time that the agents have taken to solve a case/message. It is basically the difference of time between the assignment of case/message and unassignment of case/message.
The formula to calculate estimated wait time via User Handling Time method is:
Estimated Wait Time = {AHT(Average Handling Time) * PIQ (Position In Queue)} / AA (Number of Active Agents) |
For example, refer to the below table.
Average Handling Time (AHT) | Position in Queue (PIQ) | Active Agents (AA) | Estimated Wait Time | |
1 | 12 minutes | 3 | 7 | 5.1 |
2 | 8 minutes | 2 | 3 | 5.3 |
3 | 6 minutes | 5 | 6 | 5 |
4 | 4 minutes | 8 | 8 | 4 |
5 | 2 minutes | 9 | 9 | 2 |
In the above table, if you refer to the 4th case then you can calculate Estimated Wait time by applying the above formula:
Estimated Wait Time for 4th case = 4*8/8 => 4 minutes
Similarly, you can calculate it for 5th case = 2*9/9 => 2 minutes
To Copy Estimated Wait Time to Cases/Messages Custom Fields via Rule Engine
By creating a rule in the Rule Engine, you can set actions under Actions to ‘Copy properties from source to destination’, to copy Estimated Wait Time (User Handling Time) and Estimated Wait Time (Work Queue Wait Time) into Case/Message custom properties fields.