Best practices for Amazon Connect outbound campaigns using Pinpoint

The topics in this section explain best practices for outbound calling campaigns using Pinpoint. These practices can increase agent productivity, help you comply with regulations, and help protect the integrity of your phone numbers.

Choose the right campaign

Amazon Connect provides several types of dialing campaigns. The following sections describe each type so that you can implement the campaign that best meets your needs.

Predictive campaigns

When agent productivity, cost per calls, or contact center efficiency are critical metrics, use predictive dialers. Predictive dialers anticipate that many calls won't be answered. They counterbalance that by dialing as many phone numbers in a list as possible during an agent's shift by making predictions about agent availability.

The predictive algorithm calls ahead based on certain performance metrics. This means that calls can be connected before an agent becomes available, and a customer is connected to the next available agent. The predictive algorithm constantly analyzes, evaluates, and makes agent availability predictions in real-time so that agent productivity and efficiency can improve.

Progressive campaigns

When you need to reduce answer speeds, use progressive dialers. A progressive dialer dials the next phone number in a list after an agent completes the previous call. The dialer only makes as many outbound calls as there are agents available.

You can use integrated answering machine detection to help identify a live customer pickup or a voicemail, and customize your contact strategy accordingly. For example, if a person answers a call, you can present options for them to select. If a call goes to voicemail, you can leave a message.

You can also manage pacing by specifying dial capacity for each campaign. For example, you can send more voice notifications faster by setting a higher dial capacity for a given agentless campaign compared to other dialer campaigns.

Agentless campaigns

You use agentless campaigns to send high-volume personalized voice notifications, appointment reminders, or to enable self-service using the Interactive Voice Response (IVR) with no agents needed.

Agent staffing best practices

When call recipients answer a call and hear silence in return, they often hang up. For predictive campaigns, use the following best practices to help reduce that silence:

Connection latency best practices

Successful outbound calling campaigns avoid silent calls, the period of silence after a person answers a call and before an agent comes on the line. Legal requirements to limit the number of silent or abandoned calls and keep the called party informed may also apply. You can configure Amazon Connect in different ways to reduce call connection delays.

Pinpoint segment attributes

When creating an Amazon Pinpoint segment file, add the data (attributes) required for routing logic, custom greetings, or agent screen pop. Do not use Lambda functions in the flow to extract additional information, such as EffectiveDate, Attributes.CampaignIdentifier, or User.UserId before connecting to an agent.

For more information, see Supported attributes in the Amazon Pinpoint User Guide.

Outbound agent-staffed calling

When using the Check call progress flow block:

Outbound agentless calling

Outbound campaigns often use custom greetings and self service functions. Do not use Lambda functions to get contact attributes. Instead, provide customer data (attributes) via the campaign segment. Use these attributes from the campaign segment to play custom greetings.

Whisper and queue flow best practices

User administration best practices

We recommend setting the following options for your users to reduce connection times. To access these settings, in the Amazon Connect admin website navigate to Users, User management, Edit.

These options apply to soft phones only.

The following image shows the Settings section of the Edit user page.

Workstation and network best practices

The following best practices can help optimize agent efficiency by ensuring adequate hardware and network resources.

Testing best practices

As a best practice, run tests at scale. To achieve the lowest call connection latency, use outbound campaigns to make hundreds of thousands of continuous calls to mimic your production environment. Call connection latency can be relatively high when making a handful of campaign calls.

Best practices for answering machine detection

To use Answering Machine Detection (AMD) in a campaign, use the Check call progress flow block. It provides call progress analysis. This is an ML-model that detects an answered call condition so that you can provide different experiences for calls answered by people and calls answered by machine, with or without a beep. The flow block also provides a branch for routing calls when the ML-model can't distinguish between people and voicemail, or when errors occur in call processing.

AMD uses the following criteria to detect live calls:

Forty to 60-percent of calls to consumers go to voicemail. AMD helps eliminate the number of voicemail calls over live calls. However, the detection accuracy has limitations.

The pros, cons, and best uses of Answering Machine Detection

The use of Answering Machine Detection (AMD) may not comply with telemarketing laws. You are responsible for implementing AMD in a manner that is compliant with applicable laws, and you should always consult your legal advisor regarding your specific use case.

Use case 1: AMD is on and leaving automatic voicemails

Use case 2: AMD is on but not leaving automatic voicemails

Use case 3: AMD is off and agents can leave manual voicemails

Use case 4: AMD is off and agents can leave a prerecorded voicemail

Journey best practices

As a best practice, create a well-defined scenario for each Amazon Pinpoint journey. Limit the scope of a scenario to a specific aspect of a larger customer experience that enables you to monitor, refine, and manage a customer's specific experience. You can then create a sequence of related journeys.

For example, a journey can welcome new customers and provide recommended first steps during their first seven days as a customer. Based on each customer's actions during the first journey, you can route them to additional journeys tailored to their initial level of engagement. One journey might provide next steps for customers who were highly engaged in the first journey. Another subsequent journey might promote different products or services to customers who were less engaged in the first journey. By creating a sequence of scoped journeys, you can continually refine and manage the customer experience throughout the customer lifecycle.

After you define a scenario, choose journey settings that support your goals for the scenario. The settings define the timing, volume, and frequency with which any part of a journey can engage participants.

Schedule best practices

Amazon Connect outbound campaigns enable you to limit calls to certain times of day and avoid calls during quiet times in the evening or during weekends. You can also set calling exceptions in an Amazon Pinpoint journey. The exceptions overwrite the sending times configured for days of the week.

We recommend using both features. For more information about scheduling in Amazon Connect, see . For more information about scheduling in Amazon Pinpoint, see Step 4: Choose when to send the campaign, in the Amazon Pinpoint User Guide.

In addition to exceptions, you can:

Best practices for activity settings

In the Entry activity of your journey, only use the Add participants from a segment option.

Do Not Call best practices

Many countries have created DNC (Do Not Call) lists. These allow telephone subscribers to not receive marketing calls. Companies must check customer phone numbers against such DNC lists, and remove those numbers before placing a call. You use Amazon Pinpoint to manage DNC lists in outbound campaigns.

Journeys allow you to check the status of an endpoint against third-party data sources before sending the messages. You can also add an AWS Lambda function that conducts external DNC checks and does or does not dial based on the response.

The following image shows the suggested DNC flow.

Best practices for managing redials

The following sections provide best practices for managing redials and differentiating your call center from spammers.

Automate workflows and use multiple channels

As a best practice, don't persistently call leads and hope the contact answers. The more you call, the less likely the contact may be to answer. Instead, use automation to move the contact to another list and call back 30 days later, then perhaps 60 days later.

Also, look at the number of times a call goes to voicemail. At some point, you may want to stop calling that lead.

An optimal strategy uses automated workflows with multiple communication channels to build an outreach cadence. For example, you start with a phone call, then send an SMS message, then an email. This can significantly increase the chances of contacting the lead. For more information about setting up multiple channels, see:

The next sections provide other tips for managing redials.

Manage your call volume

The following best practices can help differentiate your call center from spammers and help protect the integrity of your phone numbers.