Independent study lab
This is an independent, free, learning-only student project. It respects every company and vendor. Trademarks belong to their owners. It is not affiliated with or endorsed by Amazon Web Services, Cisco, Genesys, Salesforce, Twilio, or any other company. If you have a concern, the page will be removed or corrected promptly through the contact form. Nothing here is sold. There are no ads. The author is happy to volunteer for any organization.
Forecasts, capacity, and schedules
Confirm before relying. Quotas, prices, and names change.
The forecast interval is a one-time choice
Forecasting predicts contact volume and handle time from historical metrics. Short-term forecasts update daily and are the input for scheduling up to 18 weeks ahead. Long-term forecasts update every week, on the weekday you created them, and are the input for capacity planning up to 64 weeks ahead.
Before any of that, set the forecast and schedule interval to 15 or 30 minutes. The interval page says you need Forecast and schedule interval - Edit, you must set it before you create forecast groups, and you cannot change it later. Intraday forecasts update every 15 minutes, and only for a queue-channel with at least 5,000 unique contacts a week over the last 4 weeks.
- The first short-term or long-term forecast is typically ready within 4 hours, and often in about 1 hour.
- Only the newest forecast stays on the page. Download a copy if you need to keep that day.
- The forecasting page lists the order: interval, forecast groups, historical import, create, inspect, publish.
- A change to the forecast group, such as adding or removing a queue, starts that computation again.
Source: Set the forecast and schedule interval
Hiring plans and scheduling plans
A capacity plan turns forecasts and a scenario into staffing targets. A Hiring plan estimates long-term full-time equivalents, up to 64 weeks, and needs both a long-term and a short-term forecast. The capacity page recommends at least 4 weeks of overlap and requires at least 1 day.
A Scheduling plan needs only a published short-term forecast. It returns headcount in 15- or 30-minute intervals, matching that forecast, and the plan dates must sit inside the forecast dates. Plan override is not available for a Scheduling plan.
- Create a scenario before you create the plan.
- Shrinkage and available full-time equivalents can be imported. Hiring plans use day-level files. Scheduling plans use interval-level files.
- Task and email plans can start from an automatic backlog when you use an average time to complete.
- The getting-started list ends with review, then publish. Override and re-run are for hiring plans.
Source: Capacity planning
Publish the schedule before you talk about adherence
The forecasting and scheduling overview describes forecasting, scheduling, schedule adherence, and capacity planning. Scheduling builds agent schedules for day-to-day work. Adherence metrics are available after the agent schedules are published.
The forecasting page says you cannot build a schedule or a capacity plan until a corresponding forecast exists. Publish the short-term forecast when schedulers should use it. That page shows when a forecast was last updated and when it was published.
- Forecasts update on their own. Publishing is the step that hands one to schedulers.
- Short-term forecasts feed scheduling, up to 18 weeks ahead.
- Long-term forecasts feed capacity planning, up to 64 weeks ahead.
- The overview names administrator, forecaster, scheduler, capacity planner, and agent as the people in the workflow.
Source: Forecasting and agent scheduling
Sources and go deeper
- Set the forecast and schedule interval in Connect Customer - Amazon Connect Customer AWS docs. last checked 2026-10
- Forecasting in Connect Customer - Amazon Connect Customer AWS docs. last checked 2026-10
- Forecasting & agent scheduling in Connect Customer - Amazon Connect Customer AWS docs. last checked 2026-10
- Capacity planning in Connect Customer - Amazon Connect Customer AWS docs. last checked 2026-10
sources.json (wfm-interval) sources.json (wfm-capacity) sources.json (wfm-schedule)