Understanding Estimated Wait Time

Estimated Wait Time (EWT) is a projection of how long a person can expect to wait before connecting with a live agent. The estimated wait time gives people transparency and helps them make educated decisions.

From the consumer standpoint, if someone knows that they will have to wait an hour in a queue, they can make alternative plans such as:

Requirements for Estimated Wait Time feature

Note: Estimated Wait Time feature is supported for voice, video and messaging queues.

Table of Contents

How Estimated Wait Time is calculated

To determine the Estimated Wait Time, Zoom Contact Center must have:

 

If the engagement requires a live agent, Zoom Contact Center will first select a pool of agents suitable to taking that engagement. This can be accomplished using one of two ways:

 

The strategy for managing EWT ensures precise outcomes by consistently updating after each engagement, addressing new scenarios, and accounting for fluctuations in daily, weekly, or seasonal patterns.

How to test the Estimated Wait Time functionality

To accurately test the Estimated Wait Time (EWT) feature, follow the guidelines below. These steps help ensure consistent and reliable results during testing.

Ensure all agents are busy

EWT is only triggered when agents assigned to the target queue are in a busy state.
If agents are offline, ready, or in any non-busy state, it will prevent EWT from being activated.

Maintain a stable queue configuration

EWT predictions depend on agent skills and queue setup. During testing:

Any changes during testing can lead to inaccurate or fluctuating EWT estimates.

Provide sufficient historical data

The EWT model relies on historical call data. Make sure that:

Note: If most past calls had wait times under two minutes or consumers were not served, the system may play default EWT prompt, regardless of current conditions.

Allow time for model updates

The EWT model updates approximately every 15 minutes. The model will retain data for up to a week. To ensure the model reflects recent activity:

This delay ensures the latest data is factored into the prediction.

Testing in untrained queues or fallback

If a queue has not yet accumulated enough historical data (i.e., it’s untrained) or during fallback when the EWT service is not accessible:

Note:  In production deployment, when agents are actively using the system, training for EWT model is not needed. However, during evaluation or testing, this training data is needed. If queues are not trained, fallback message will be played.