The blocker
The collections team, the renewals team or the appointment desk needs to reach a few thousand customers this week. Agents work through a list, dialing each number by hand. Most calls ring out, hit voicemail or a busy tone. An agent might spend more of the hour waiting than talking.
The compliance risk is just as real. Lists are exported from different systems and nobody checks the customer’s local time. The same person gets called three times in a day by three different teams. When a supervisor turns up the pace to hit targets, customers pick up to silence and hang up.
Why it happens
- Dialing is manual. Agents spend their time on calls that never connect.
- Pacing is a guess. Too slow wastes agents; too fast creates silent, abandoned calls.
- No shared limits. Each team runs its own list, so nobody sees how often a customer has been contacted.
- Time zones are ignored. A single calling window is applied to customers across the country.
- Results are hard to see. Campaign outcomes live in spreadsheets, not alongside inbound metrics.
How Amazon Connect solves it
Outbound calling rules differ by country and state. You are responsible for how you configure and use these features. This is not legal advice. Confirm your settings with your own compliance team, and see AWS’s shared responsibility model.
Amazon Connect outbound campaigns is a feature of Amazon Connect for high-volume outbound communication by voice, SMS, email and other messaging channels. You can create campaigns in the guided campaign builder for single-channel campaigns, or in the visual journey builder for multi-step, multi-channel journeys. A campaign can also start from a customer event, such as a change in your Customer Profiles data.
1. Pick the dialing mode that fits the job
AWS’s best practices describe the modes:
- Predictive dials ahead, predicting when agents will be free. Use it when agent productivity and cost per call matter most.
- Progressive dials the next number only after an agent finishes the previous call, which reduces the chance of a customer waiting.
- Agentless (automated voice) delivers reminders and notifications, or self-service through an IVR, with no agent.
The guided campaign builder also offers a preview mode for agent-assisted voice.
2. Spend agent time on people, not voicemail
AWS notes that 40 to 60 percent of calls to consumers go to voicemail. Answering machine detection (AMD), through the Check call progress flow block, uses a machine learning model to tell people from machines, so people reach an agent and machines get a message or a retry. AWS is clear about its limits: a short “Hello” greeting can be mistaken for a person, and a long live greeting for a machine. AWS also states that AMD might not comply with telemarketing laws, and that you are responsible for using it in a compliant way.
3. Keep abandoned calls under control
For predictive campaigns, abandonment controls let you set a target abandonment rate, a measurement window from 1 to 24 hours, and when the abandonment timer starts. When the rate reaches your target, the dialer switches to progressive dialing to bring it back down. AWS advises consulting your compliance team on the right values, because regulations vary.
4. Call at the right time, and not too often
Each campaign has active communication times by day of week, plus optional exception days when nothing is sent. You choose a standard time zone or the recipient’s local time zone, inferred from the profile’s address or phone area code. Recipients whose time zone cannot be determined are dropped, not called at a guess.
Communication limits cap how many times a recipient is contacted in a time frame, for example 4 per 2 days and 6 per 14 days. Total communication limits apply the same idea across every campaign in the instance, so separate teams cannot overwhelm one customer. A voicemail still counts as a communication, and AWS says the count errs on the side of over-counting.
5. Retry with rules, and measure the result
Retry rules act on dispositions such as busy, unanswered or voicemail, with an optional wait before the next attempt, and can move to the customer’s next number. The outbound campaigns performance dashboard shows delivery attempts, human answered rate, abandonment and more, and campaign metrics are available for reporting.
Two setup points to plan for: the default quota for concurrent campaign calls is 0, so you request an increase first, and event-driven mass notifications need pre-authorization through AWS Support.
What good looks like
Agents talk to people instead of voicemail. With abandonment controls on, the dialer switches to progressive dialing on its own when abandonment reaches your target. Every customer is called only inside the hours you set, in their own time zone, and no more often than your limits allow, whichever team owns the list. Supervisors see campaign results on the same platform as inbound, and compliance can point to the settings behind each rule.
Where to start
- List every outbound program today, who runs it and how often each customer is contacted.
- Agree calling windows, contact limits and abandonment targets with your compliance team.
- Request the outbound campaigns quota and set up a dedicated campaign queue and flow.
- Start one campaign in progressive mode, review the dashboard, then try predictive with abandonment controls.
- Use forecasting and scheduling so enough agents are logged in when campaigns run.










