Rate this page:

Integrate Virtual Hold Technologies (VHT) with Twilio Flex

Virtual Hold Technologies (VHT) was the first company to offer callbacks and have 25 years of experience offering this service. This implementation guide will step you through the an integration between Flex and VHT’s Mindful platform. With this integration, you can respect your customers' time and give them control by letting them choose when they'll receive a call. Whether it's two minutes or two hours, your customers will appreciate spending less time waiting on hold.

Background on Callbacks


When your queues are full and your agents are swamped, new inbound calls are going to have longer hold times. This can lead to abandoned calls, poor customer experience, increased service costs (as you deal with repeat attempts by a customer), and fewer opportunities to help and engage with your customers.

One of the best ways to reduce hold time is to give callers an option to skip it altogether. With Virtual Hold, you can offer customers the option to call them back at a more convenient time. This works great for contact centers that have peaks and valleys in their contact volume; you take the contact volume that arrives during your peak volume time and you handle it when your staff isn’t as busy.

Implementation Guide Disclaimer

This guide has two primary sections:

  1. Setting up the connection with Virtual Hold
  2. Demonstrating two example call flows that incorporate callbacks.

The example call flows are a guide, and you should adapt them to fit your desired customer experience. Similarly the sample code provided is a guide and doesn't incorporate standard techniques like error handling. These should be refined based on your needs before integrating into a live environment.

Setting up connectivity with Virtual Hold

Here is a high-level overview of the connectivity with VHT:

vht connectivity

You will need to set up a SIP Domain and SIP Trunk which will be used to transfer calls to VHT, as well as a dedicated phone number, which will be used for the Callbacks.

Follow these steps to establish connectivity with VHT:

1. Provision phone numbers in Twilio

In your Twilio console, go to Active Phone Numbers

This solution relies on two phone numbers, one for incoming calls from your customers and one for incoming calls from VHT. These numbers will route to separate Studio Flows which will be configured later on in this document. Don’t forget to update the routing of these phone numbers after you have completed the Studio Flows.

vht phone numbers.jpg

2. Set up an IP Access Control List

1. In your Twilio console, go to IP / CIDR Access Control List

2. Create a new ACL with the following addresses: - VHT SIP Proxy - VHT SIP Proxy - VHT RTP Proxy - VHT RTP Proxy

Screen Shot 2021-03-09 at 1.13.15 PM.png

3. Create a SIP Domain

1. In your Twilio console, go to Programmable SIP Domains

2. Create a new SIP Domain

Hint: Use the IP Access Control List defined in step 1

Screen Shot 2021-03-09 at 1.17.23 PM.png

In this example, incoming calls to this SIP Domain route to a Studio flow, which is configured later in this document. You are welcome to use other options as well, such as a Function that returns TwiML.

4. Provision a new Call Target

1. In Mindful, go to Call Targets, Add New CT; or click here

2. Change the Telephony Type to SIP

3. For the Call Center Phone Number, enter the phone number which you use for the return calls from VHT.

4. For Callback CID, enter the Twilio phone numbers callers are calling. This will be used as the caller ID when returning the callback

Screen Shot 2021-03-09 at 1.58.46 PM.jpg

5. Go to the Metadata tab

6. Add a metadata item called x-user-to-user with the following settings:

pasted image 0.png

5. In Mindful, provision a new phone number

1. Go to Phone Numbers

2. Provision a new phone number and assign it to the Call Target configured in step 4

Screen Shot 2021-03-09 at 2.00.58 PM.jpg

Scenario 1: Offer a callback before the caller enters a task queue

Before sending an incoming call to a TaskRouter task queue, we will call a Function that will use the TaskRouter Statistics API to determine the Estimated Wait Time. If this is above a threshold, we will offer the caller the option of a callback instead of sending the call to a queue.

The TaskRouter Statistics API does not provide a true Estimated Wait Time, but there are multiple ways to calculate it. In this example, we will keep this simple, and use the AvgTaskAcceptanceTime value that is available in TaskRouter for the previous five minutes.

pasted image 0 (1) (1).png

In diagram above, the Estimated Wait Time is obtained by using Studio’s Run Function widget (in the diagram called check_average_wait_time) and invoking a Twilio Function with following code:

 * Function to read avgTaskAcceptanceTime statistics from TaskRouter's cumulative statistics.
 * It returns JSON object with following fields:
 * - avgTaskAcceptanceTime - number of seconds
 * Expected variables from context:
 * - Queue_Estimated_Wait_Time - initial value of 0, used by script to cache value of average task acceptance time
 * - Queue_Estimated_Wait_Time_Last_Updated - initial value of 0, used by script to cache timestamp of last update of
 * - Queue_Update_Interval - average time update interval in milliseconds, initial value of 60000
 * - Workspace_SID
 * - Task_Queue_SID
 * - Service_SID
 * - Environment_SID
 * - VAR_QEWTLU_SID - SID of variable Queue_Estimated_Wait_Time_Last_Updated
 * - VAR_QEWT_SID - SID of variable Queue_Estimated_Wait_Time
 * Following twilio-cli calls are useful for setting up environment variables for this script:
 * twilio api:taskrouter:v1:workspaces:list
 * twilio api:taskrouter:v1:workspaces:task-queues:list \
 * twilio api:serverless:v1:services:list
 * twilio api:serverless:v1:services:environments:list \
 * twilio api:serverless:v1:services:environments:variables:list \

exports.handler = function (context, event, callback) {

    const response = new Twilio.Response();
    response.appendHeader('Access-Control-Allow-Origin', '*');
    response.appendHeader('Access-Control-Allow-Methods', 'OPTIONS POST');
    response.appendHeader('Content-Type', 'application/json');
    response.appendHeader('Access-Control-Allow-Headers', 'Content-Type');

    get_wait_time(context, event, callback).then(value => {
                             'avgTaskAcceptanceTime': value
        return callback(null, response);


async function get_wait_time(context, event, callback) {
    const client = context.getTwilioClient();

    let current_timestamp = new Date().getTime();

    if ((current_timestamp - context.Queue_Estimated_Wait_Time_Last_Updated) > context.Queue_Update_Interval) {

        let average_task_acceptance_time = await get_queue_cumulative_statistics(client, context.Workspace_SID, context.Task_Queue_SID, 'avgTaskAcceptanceTime');

        context.Queue_Estimated_Wait_Time = parseInt(average_task_acceptance_time);

                        key: 'Queue_Estimated_Wait_Time',
                        value: average_task_acceptance_time
                        key: 'Queue_Estimated_Wait_Time_Last_Updated',
                        value: current_timestamp


    return context.Queue_Estimated_Wait_Time;

async function get_queue_cumulative_statistics(twilio_client, workspace_sid, task_queue_sid, stat_name) {
    return twilio_client.taskrouter.workspaces(workspace_sid)
        .then(stats => {
            return (stats[stat_name]);

If you receive the following error during execution, go to service dependencies and update the twilio library's version (e.g., to *) and redeploy.

UnhandledPromiseRejectionWarning: Unhandled promise rejection: TypeError: 
Cannot read property 'services' of undefined at get_wait_time
(/var/task/handlers/ZN016166710a27ef5a1f9efa721c2809e2.js:40:33) at
processTicksAndRejections (internal/process/task_queues.js:97:5)

If the Estimated Wait is greater than 120 seconds, the call is transferred to Virtual Hold via the SIP trunk. Otherwise, the call is sent to Flex via a TaskRouter task queue.

In the toolstep named transfer_to_vht, we define the SIP endpoint to which we transfer the call. This is the phone number that you set up above in step 5, in the following format:{{}}

We also send VHT the caller’s phone number through the user-to-user SIP Header

Scenario 2: Offer a callback while the customer is on hold in a task queue

In this example, the incoming call reaches the TaskRouter task queue in the Studio Flow.

Screen Shot 2021-02-18 at 3.01.46 PM.jpg

The average wait time check is performed in the widget Send to Flex (send_to_flex1 in the diagram above) by using the Hold Music TwiML URL parameter. Instead of pointing to actual Hold Music TwiML, you can point to a custom Twilio Function called hold_treatment(), which in turn calls another custom function transfer_to_vht().


  • When quoting the estimated wait time prior to offering a callback, it is a best practice to set an upper limit for the amount of time that can be quoted. For example, if the wait time exceeds 10 minutes, you might choose to say "...more than than 10 minutes from now" rather than quoting an exact time.
  • We recommend setting a minimum offer threshold based on the current estimated wait time to ensure that callback offers are not made when wait times are very low. You may also wish to check agent availability prior to offering a callback.

Code for hold_treatment:

 * This function expects the same environment variables as the wait_time.js
 * Expected variables from context:
 * - Queue_Estimated_Wait_Time - initial value of 0, used by script to cache value of average task acceptance time
 * - Queue_Estimated_Wait_Time_Last_Updated - initial value of 0, used by script to cache timestamp of last update of
 * Queue_Estimated_Wait_Time
 * - Queue_Update_Interval - average time update interval in milliseconds, initial value of 60000
 * - Workspace_SID
 * - Task_Queue_SID
 * - Service_SID
 * - Environment_SID
 * - VAR_QEWTLU_SID - SID of variable Queue_Estimated_Wait_Time_Last_Updated
 * - VAR_QEWT_SID - SID of variable Queue_Estimated_Wait_Time
exports.handler = function(context, event, callback) {
    let twiml = new Twilio.twiml.VoiceResponse();

    get_wait_time(context).then(avg_wait_time => {
        let action_url = "https://" + context.DOMAIN_NAME + '/' + 'transfer_to_vht';
        if (avg_wait_time > 120) {
            twiml.gather({action: action_url})
                .say(`Your call will be routed to an agent in approximately ${avg_wait_time / 60} minutes. Press 1 if you want to be called back`);
        return callback(null, twiml);


Code for transfer_to_vht:

exports.handler = function(context, event, callback) {
    let response = new Twilio.twiml.VoiceResponse();

    let gathered_digits = 0;
    if (event['Digits']) {
        gathered_digits = parseInt(event['Digits']);
    if (gathered_digits && (gathered_digits === 1)) {
        // leave Flex queue and continue with next node connected to Task completed

    return callback(null, response);


When the Estimated Wait Time exceeds the threshold, in this case, 120 seconds, the function returns <Leave> TwiML which returns control back to Studio and continues with the next widget.

On the Task Created condition, Studio will use the Connect Call To widget and redirect to VHT.

Routing the Return Call

The last step of this Implementation Guide is to route the return calls from VHT straight to Flex, but with a higher priority. This should be used as a followup for either of the two scenarios above.

We can do this with the following Studio Flow:

Screen Shot 2021-03-09 at 1.50.35 PM.png

This Flow will grab the Sip Header x-user-to-user and set that as an attribute to the TaskRouter task. It also changes the Priority from a default value of 0 to 5, to give these incoming calls a higher priority. Depending on your organization’s TaskRouter setup, you may need to change this value to one that is appropriately ranked.

Once you’ve connected all of your Studio Flows to the appropriate phone numbers, your callback solution should be ready to go. Well done - your customers will thank you for saving them from those long hold times!

Rate this page:

Need some help?

We all do sometimes; code is hard. Get help now from our support team, or lean on the wisdom of the crowd browsing the Twilio tag on Stack Overflow.


        Thank you for your feedback!

        We are always striving to improve our documentation quality, and your feedback is valuable to us. How could this documentation serve you better?

        Sending your feedback...
        🎉 Thank you for your feedback!
        Something went wrong. Please try again.

        Thanks for your feedback!

        Refer us and get $10 in 3 simple steps!

        Step 1

        Get link

        Get a free personal referral link here

        Step 2

        Give $10

        Your user signs up and upgrade using link

        Step 3

        Get $10

        1,250 free SMSes
        OR 1,000 free voice mins
        OR 12,000 chats
        OR more