# Setting up forecasting and scheduling in Amazon Connect Systems: Amazon Connect For Amazon Connect administrators, forecasters and schedulers who own workforce management setup for an existing instance. Canonical: https://warmtransfer.net/guides/amazon-connect-forecasting-scheduling-setup Last verified: 2026-10-02 Amazon Connect generates schedules from a published short-term forecast, shift profiles, staffing groups and human resources and business rules[^59]. AWS describes Amazon Connect agent event streams and contact events as usable for integration into workforce management solutions[^16][^20]. ## Before you start AWS documentation states that Amazon Connect Customer is the current name for the product previously called Amazon Connect[^75]. AWS orders native scheduling setup as: add users, confirm security profile permissions, create staff rules, create shift activities, create shift profiles, create staffing groups, then generate, review and publish a schedule[^61]. By default the Admin security profile has permissions for all forecasting, capacity planning and scheduling activities[^12]. - The instance Region, because AWS lists forecasting and agent scheduling as available only in a specific set of Regions[^58]. - Your interval choice, because after the forecast and schedule interval is set it cannot be changed[^46]. - The queues that share the same pool of agents, because AWS recommends grouping those queues into the same forecast group[^29]. - Prior-system history, because with less than one year of historical data in Amazon Connect, AWS strongly recommends importing data from the previous system to form a continuous one-year history[^41]. - Your employment-law obligations, because AWS states that Amazon Connect scheduling is not designed to ensure compliance with any particular laws and that the customer is responsible for employment-law compliance[^60]. - For an external WFM, an Amazon Kinesis data stream, because agent event streams are enabled by selecting an existing Kinesis data stream[^13]. See also [Setting up queues and routing profiles in Amazon Connect](https://warmtransfer.net/knowledge/amazon-connect-routing-profile-setup). ## What changes by situation - Which engine will forecast and schedule your agents? Amazon Connect forecasting, capacity planning and scheduling; An external WFM product fed from Amazon Connect data. - Which short-term forecast and schedule interval will you lock in? 15-minute intervals; 30-minute intervals. - How much usable contact history does Amazon Connect already hold for the queues you will forecast? More than one continuous year of representative history in Amazon Connect; Less than one year, or history containing anomalies you need to correct. ## Step 1: Confirm the instance Region supports native forecasting and scheduling **Amazon Connect forecasting, capacity planning and scheduling** ### Do Confirm that the instance's Region is one where AWS lists Amazon Connect forecasting and agent scheduling as available: US West (Oregon), Asia Pacific (Tokyo), Asia Pacific (Seoul), Asia Pacific (Singapore), Asia Pacific (Sydney), Canada (Central), Europe (Frankfurt), Europe (London), AWS GovCloud (US-West) or the US East Region in Virginia[^58]. WarmTransfer's reading of the sources is that an instance in Africa (Cape Town) should not be expected to offer native forecasting and scheduling, because that Region appears in the instance Region table but not in the forecasting and agent scheduling list[^54]. ### Verify The forecast and schedule interval tab sits under Analytics and optimization, Forecasting[^48]. Suggested check: sign in as an administrator and confirm that the forecasting area appears in the admin website navigation. **An external WFM product fed from Amazon Connect data** ### Do Skip this step: the Region list applies to Amazon Connect forecasting and agent scheduling[^58]. ### Verify Suggested check: confirm your instance will feed an external WFM product rather than use native forecasting and scheduling. ## Step 2: Grant security profile permissions to each WFM persona **Amazon Connect forecasting, capacity planning and scheduling** ### Do AWS advises using a default security profile only if it matches what users need, and otherwise creating a security profile that grants only the needed permissions[^76]. Create a profile under Users, Security profiles, Add new security profile; choosing an action automatically selects the actions it depends on, such as View when Edit is chosen[^77]. Grant Analytics, Forecast and schedule interval - Edit to whoever sets the interval[^48] and the forecast time zone[^74]. Grant Forecasting - Edit to forecasters, since creating forecast groups, importing historical data, and creating, inspecting, downloading and editing forecasts each require it[^23]. Grant Forecasting - Publish to whoever publishes forecasts[^55]. Grant Scheduling, Schedule manager - Edit to schedulers, since creating staff rules, shift activities, shift profiles, staffing groups and schedules requires it[^62]. Grant agents Agent application schedule calendar - View or Edit to see their schedule, and Edit to request time off[^17]. Grant Schedule manager View plus Analytics and Optimization, Historical metrics - Access to anyone reviewing historical schedule adherence[^6]. ### Verify A user without Analytics, Forecast and schedule interval - Edit cannot set the forecast and schedule interval[^48]. Suggested check: sign in as one user from each persona and confirm each reaches the pages their role needs and no others. ### Rollback Suggested rollback: remove the added permissions from the security profile, or reassign the affected users to their previous profile. **An external WFM product fed from Amazon Connect data** ### Do Skip this step: the permissions in this step govern Amazon Connect forecasting, capacity planning and scheduling activities[^12]. ### Verify Suggested check: confirm your instance will feed an external WFM product rather than use native forecasting and scheduling. ## Step 3: Set the forecast and schedule interval **15-minute intervals + Amazon Connect forecasting, capacity planning and scheduling** ### Do The interval must be set before forecasts can be generated or forecast groups created[^45]. In Analytics and optimization, Forecasting, open the forecast and schedule interval tab[^48] and select the 15-minute interval, which Amazon Connect then uses for both short-term forecasts and schedules[^47]. ### Verify Forecast group creation depends on the interval being set[^45]. Suggested check: confirm the tab shows your chosen interval as the saved value. ### Rollback After the forecast and schedule interval is set it cannot be changed[^46]. **30-minute intervals + Amazon Connect forecasting, capacity planning and scheduling** ### Do The interval must be set before forecasts can be generated or forecast groups created[^45]. In Analytics and optimization, Forecasting, open the forecast and schedule interval tab[^48] and select the 30-minute interval, which Amazon Connect then uses for both short-term forecasts and schedules[^47]. ### Verify Forecast group creation depends on the interval being set[^45]. Suggested check: confirm the tab shows your chosen interval as the saved value. ### Rollback After the forecast and schedule interval is set it cannot be changed[^46]. **An external WFM product fed from Amazon Connect data** ### Do Skip this step: the interval is a setting for Amazon Connect short-term forecasts and schedules[^47]. ### Verify Suggested check: confirm your instance will feed an external WFM product rather than use native forecasting and scheduling. ## Step 4: Set the forecast time zone **Amazon Connect forecasting, capacity planning and scheduling** ### Do Change the forecast time zone from its UTC default to the zone your forecasters plan in; editing it requires Analytics, Forecast and schedule interval - Edit[^74]. ### Verify Changing the forecast time zone regenerates both short-term and long-term forecasts, with updated forecasts available within 24 hours and adjusted for daylight saving where the zone observes it[^73]. Suggested check: confirm the saved zone is the one you intended. ### Rollback Suggested rollback: set the forecast time zone back to its previous value. Any such change regenerates both short-term and long-term forecasts again[^73]. **An external WFM product fed from Amazon Connect data** ### Do Skip this step: the forecast time zone applies to Amazon Connect short-term and long-term forecasts[^73]. ### Verify Suggested check: confirm your instance will feed an external WFM product rather than use native forecasting and scheduling. ## Step 5: Create every forecast group before any forecast **Amazon Connect forecasting, capacity planning and scheduling** ### Do Forecast groups are associated with staffing groups for scheduling, so AWS recommends grouping queues that share the same pool of agents into the same forecast group[^29]. Each queue can belong to only one forecast group, and at least one forecast group must exist before any forecast can be generated[^28]. Keep each group within 200 queues; an instance allows at most 500 forecast groups[^50]. AWS strongly recommends creating all forecast groups before creating any forecast, because historical data for queues in all forecast groups trains the forecasting model[^26]. ### Verify Each queue can belong to only one forecast group[^28]. Suggested check: list every queue you intend to forecast and confirm each sits in exactly one group. ### Rollback A forecast group that was used to create forecasts cannot be removed until those forecasts are deleted from the Forecasts tab[^27]. Adding or removing queues in an existing forecast group can trigger recomputation of its forecasts and affects downstream capacity plans and schedules built on that group[^25]. That recomputation typically completes within 4 hours[^30]. **An external WFM product fed from Amazon Connect data** ### Do Skip this step: forecast groups are associated with staffing groups for Amazon Connect scheduling[^29]. ### Verify Suggested check: confirm your instance will feed an external WFM product rather than use native forecasting and scheduling. ## Step 6: Import historical data where Amazon Connect history is short or wrong **Less than one year, or history containing anomalies you need to correct + 15-minute intervals + Amazon Connect forecasting, capacity planning and scheduling** ### Do AWS strongly recommends importing data from the previous system to form a continuous one-year history[^41]. Build CSV files with the columns QueueName, QueueId, ChannelType (VOICE, CHAT, TASK or EMAIL in capitals), TimeStamp (ISO8601), IntervalDuration, IncomingContactVolume, AverageHandleTime in seconds and ContactsHandled[^37]. Upload one set with IntervalDuration 15mins for short-term forecasts and a separate set with IntervalDuration daily for long-term forecasts, because interval-only data does not allow long-term forecasts[^43]. Keep each upload within 1GB, and within 300 files each for interval and daily historical actuals[^51]. Amazon Connect then forecasts from both its own data and the imported data, and the imported data takes precedence[^42]. Where files overlap, the last uploaded records are used[^40]. ### Verify A successful upload shows Status Complete with today's upload date, and a failed upload offers a download details link with an error log identifying the faulty row[^39]. ### Rollback Imported historical data can be deleted; deletion triggers an immediate change in associated forecasts and the baseline falls back to previously uploaded or native data for the affected dates[^38]. **Less than one year, or history containing anomalies you need to correct + 30-minute intervals + Amazon Connect forecasting, capacity planning and scheduling** ### Do AWS strongly recommends importing data from the previous system to form a continuous one-year history[^41]. Build CSV files with the columns QueueName, QueueId, ChannelType (VOICE, CHAT, TASK or EMAIL in capitals), TimeStamp (ISO8601), IntervalDuration, IncomingContactVolume, AverageHandleTime in seconds and ContactsHandled[^37]. Upload one set with IntervalDuration 30mins for short-term forecasts and a separate set with IntervalDuration daily for long-term forecasts, because interval-only data does not allow long-term forecasts[^43]. Keep each upload within 1GB, and within 300 files each for interval and daily historical actuals[^51]. Amazon Connect then forecasts from both its own data and the imported data, and the imported data takes precedence[^42]. Where files overlap, the last uploaded records are used[^40]. ### Verify A successful upload shows Status Complete with today's upload date, and a failed upload offers a download details link with an error log identifying the faulty row[^39]. ### Rollback Imported historical data can be deleted; deletion triggers an immediate change in associated forecasts and the baseline falls back to previously uploaded or native data for the affected dates[^38]. **More than one continuous year of representative history in Amazon Connect + Amazon Connect forecasting, capacity planning and scheduling** ### Do Skip the import: with more than one year of historical data in Amazon Connect, AWS states that import can be skipped[^41]. ### Verify Suggested check: confirm that the history for every queue you will forecast spans more than one continuous year without gaps you would want to correct. **An external WFM product fed from Amazon Connect data** ### Do Skip this step: imported historical data feeds Amazon Connect's own forecasts[^42]. ### Verify Suggested check: confirm your instance will feed an external WFM product rather than use native forecasting and scheduling. ## Step 7: Create, inspect and publish forecasts **Amazon Connect forecasting, capacity planning and scheduling** ### Do Choose Analytics and optimization, Forecasting, the Forecast tab and Create Forecast, then select the forecast groups and the Short-term and Long-term types[^21]. Short-term forecasts are generated for 18 weeks, updated automatically every day and used for scheduling[^71]. Long-term forecasts are generated for 64 weeks, updated automatically every week and used for capacity planning[^53]. Before publishing, inspect the forecast by comparing Last computed forecast with Last published forecast and Actuals against the forecast; imported history is labelled Actuals Overrides[^44]. Publish the forecast to make it visible to capacity planners and schedulers[^55]. ### Verify Status moves from Scheduled to Complete when computation finishes[^21], typically within 4 hours and often in about 1 hour[^30]. After publishing, the status reads Complete or Publish failed[^55]. Intraday forecasts update every 15 minutes and are generated only for queues with at least 5000 unique contacts per week per queue-channel over the last 4 weeks[^49]. ### Rollback Amazon Connect retains only the last published forecast, so AWS advises downloading it with Actions, Download last published forecast before publishing a new one[^56]. Only the most current computed forecast is kept, so download a day's forecast before Amazon Connect overwrites it[^24]. Forecasts can be deleted, but downstream capacity plans and schedules created from them are impacted[^22]. **An external WFM product fed from Amazon Connect data** ### Do Skip this step: Amazon Connect short-term forecasts are used for scheduling[^71] and long-term forecasts for capacity planning[^53]. ### Verify Suggested check: confirm your instance will feed an external WFM product rather than use native forecasting and scheduling. ## Step 8: Create shift activities and choose the adherence method **Amazon Connect forecasting, capacity planning and scheduling** ### Do Create shift activities of type Productive, Non-productive or Time off, giving break activities the Non-productive type with the Break or meal sub-type[^2]. Work activity can be set only on Productive activities, an existing activity cannot be changed to a work activity, and a saved work activity cannot be changed back[^3]. Create at least one work activity, because a shift profile's default activity must be a work activity[^70]. Set Adherence to Yes to enable the Default or Custom adherence method[^1]. Under Default, an agent is adherent when scheduled Productive and in Available status, or scheduled Non-productive and Offline or in any custom status[^5]. Under Custom, each shift activity is mapped to specific agent statuses, and any unmapped status during that activity counts as non-adherent[^4]. Optionally set adherence thresholds per activity and per staffing group, up to a maximum of 10 minutes early or late[^10]. ### Verify An activity with Adherence set to No produces no Adherence metric[^1]. Suggested check: review each activity and confirm its type, work flag and adherence setting match your plan. ### Rollback A saved work activity cannot be changed back[^3]. Threshold changes apply only to future adherence calculations[^10]. Suggested rollback: edit any other activity setting back to its previous value. **An external WFM product fed from Amazon Connect data** ### Do Skip this step: shift activities control whether the Amazon Connect Adherence metric is calculated[^1]. ### Verify Suggested check: confirm your instance will feed an external WFM product rather than use native forecasting and scheduling. ## Step 9: Create shift profiles **Amazon Connect forecasting, capacity planning and scheduling** ### Do Give each shift profile its own time zone[^69]. In the profile's Schedule Window, set per day a Working or Flex status, earliest start and latest end time, optional maximum shift length and a default activity that is a work activity[^70]. ### Verify Amazon Connect automatically adjusts shifts in a profile for daylight saving changes in that profile's time zone[^69]. Suggested check: open each saved profile and confirm every day's window and default activity. ### Rollback After a shift profile is saved it can be edited or removed from the Shift Profiles list view[^68]. **An external WFM product fed from Amazon Connect data** ### Do Skip this step: shift profiles are one of the inputs Amazon Connect uses to generate schedules[^59]. ### Verify Suggested check: confirm your instance will feed an external WFM product rather than use native forecasting and scheduling. ## Step 10: Create staffing groups and staff rules **Amazon Connect forecasting, capacity planning and scheduling** ### Do Create staffing groups, associating each with one forecast group, and place every agent you want scheduled in a staffing group[^65]. Keep each staffing group at or under 350 agents, a hard limit that cannot be increased[^52], and each forecast group at or under 300 staffing groups[^50]. Add rules from Minimum staff required, Shift start time, Working time, Minimum rest between shifts, Consecutive working days, Maximum consecutive day of the week worked and Minimum consecutive rest period per week[^66]. Add anyone who needs to view published agent schedules in the Published calendar view as a supervisor[^67]. Do not make a user an agent in one staffing group and a supervisor in another, because that is not allowed[^63]. Use individual staff rules for time zone, start and end dates, shift profile override and contract rules; they override staffing group rules when a schedule is generated[^72]. AWS's documented setup order creates staff rules before shift activities and shift profiles[^61]; this guide places them here, after the shift profiles they can override and before the schedule is generated. ### Verify Amazon Connect generates a schedule only for agents who are in a staffing group[^65]. Suggested check: compare your agent roster against the staffing group membership and confirm nobody is missing. ### Rollback Removing an agent from a staffing group does not delete their shifts and adherence continues to be tracked, so AWS recommends removing their shifts first if they are not moving to another group[^64]. **An external WFM product fed from Amazon Connect data** ### Do Skip this step: staffing groups determine which agents Amazon Connect generates a schedule for[^65]. ### Verify Suggested check: confirm your instance will feed an external WFM product rather than use native forecasting and scheduling. ## Step 11: Generate, review and publish a schedule **Amazon Connect forecasting, capacity planning and scheduling** ### Do In Schedule Manager, generate a schedule for a single forecast group, for up to 18 weeks ahead[^35]. A single generation run covers at most 5,000 agents[^52]. Choose optimization goals per channel: Service level or Average speed of answer for voice and chat, and Service level or Average time to complete for task and email[^34]. Before publishing, edit shift and activity times, add, replace or remove activities, copy shifts and recompute metrics as needed[^33]. Check working-time rules against your own obligations, because AWS states that the customer is responsible for employment-law compliance[^60]. Publish with Publish and Proceed so the agents and supervisors in its staffing groups can view the schedule[^57]. ### Verify Generation shows In progress and takes 30 minutes to 3 hours depending on agents, rules and duration, ending in Complete or Failed[^32]. Resolve every HIGH warning, which means an agent was not scheduled; MEDIUM means an agent was scheduled but not all requirements were met, and LOW covers minor issues such as overstaffing[^36]. ### Rollback A generated schedule is a draft that agents and supervisors cannot see until it is published[^31]. Schedulers can edit the draft or regenerate the schedule after configuration changes[^33]. If a published schedule is changed for a date within the last 30 days, adherence is recalculated against the new schedule[^8]. **An external WFM product fed from Amazon Connect data** ### Do Skip this step: Amazon Connect schedules are generated in Schedule Manager for a single forecast group[^35]. ### Verify Suggested check: confirm your instance will feed an external WFM product rather than use native forecasting and scheduling. ## Step 12: Verify agent schedule view and live adherence **Amazon Connect forecasting, capacity planning and scheduling** ### Do Have a test agent open the calendar icon at https://{instance name}/connect/agent-app-v2/, or at /connect/agent-app-v2/scheduling when using Salesforce CTI or a custom agent application[^18]. When the first published shift starts, open the Agent adherence widget of the Queue and agent performance dashboard[^7]. ### Verify After publishing, the agents and supervisors in the schedule's staffing groups can view it[^57]. Amazon Connect begins generating schedule adherence automatically as soon as a published schedule starts that contains shift activities with Adherence set to yes[^9]. An agent with no schedule for a period shows adherence status Unscheduled[^11]. Adherent time, Adherence, Scheduled time and Non-adherent time appear on the Historical metrics and Real-time metrics pages[^7]. ### Rollback If an agent's schedule is changed for a date within the last 30 days, adherence is recalculated against the new schedule[^8]. **An external WFM product fed from Amazon Connect data** ### Do Skip this step: native schedule adherence starts only when a published Amazon Connect schedule starts[^9]. ### Verify Suggested check: confirm your instance will feed an external WFM product rather than use native forecasting and scheduling. ## Step 13: Stream agent and contact activity to a third-party WFM **Amazon Connect forecasting, capacity planning and scheduling** ### Do No streaming is needed for native setup: Amazon Connect begins generating schedule adherence automatically once a published schedule containing activities with Adherence set to yes starts[^9]. ### Verify Suggested check: confirm that adherence figures are populating for your published schedule. **An external WFM product fed from Amazon Connect data** ### Do Agent event streams are not enabled by default; after creating a Kinesis data stream, open the instance in the Amazon Connect console, go to its Data streaming page and select the stream under Agent Events[^13]. If server-side encryption is enabled on that stream with a key Amazon Connect cannot use, Amazon Connect cannot publish to it, so use the same KMS key used for recordings or scheduled reports[^14]. Agent event streams carry agent login, logout, contact connection and status-change events, which AWS describes as usable for integration into workforce management solutions[^16]. Contact events are a near real-time stream of voice, chat, task and email contact events that AWS also describes as usable for that integration[^20]. ### Verify Agent events are written with the agent ARN as partition key, so all events for one agent go to the same shard[^15]. Suggested check: log a test agent in, change its status, and confirm new records for that agent arrive in the stream. ### Rollback Suggested rollback: clear the stream selection under Agent Events on the Data streaming page. ## Applicability Applies to: Amazon Web Services Amazon Connect and Amazon Web Services Amazon Connect Customer. Deployments: multi-tenant. Sources checked 2026-10-02. AWS lists Amazon Connect forecasting and agent scheduling as available in US West (Oregon), Asia Pacific (Tokyo), Asia Pacific (Seoul), Asia Pacific (Singapore), Asia Pacific (Sydney), Canada (Central), Europe (Frankfurt), Europe (London), AWS GovCloud (US-West) and the US East Region in Virginia[^58]. WarmTransfer's reading of the sources is that an instance in Africa (Cape Town) should not be expected to offer native forecasting and scheduling[^54]. The limits of 350 agents per staffing group and 5,000 agents per schedule generation run are hard limits that cannot be increased[^52]. Amazon Connect Customer is the current name for the product previously called Amazon Connect[^78]. ## What remains uncertain Whether any particular external WFM product accepts the Kinesis agent event stream format for its real-time adherence is not covered by the sources below. How to subscribe to contact events, and how an external product consumes them, is not covered by the sources below. Shift bidding, shift trades, time-off group allowances, forecast overrides and adherence notification rules are not covered by the sources below. Capacity planning scenarios, shrinkage and FTE uploads are not covered by the sources below. Pricing for native forecasting and scheduling is not covered by the sources below. ## Sources [^1]: When a shift activity's Adherence is set to No, the Adherence metric is not calculated for it; Yes enables the Default or Custom adherence method. Source: [Create daily activities in Connect Customer for an agent's shift in your contact center](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling-create-shift-activities.html), procedure step 4, Adherence field. Checked 2026-10-02. [^2]: Shift activities have the type Productive, Non-productive or Time off, and a Non-productive activity can carry the Break or meal sub-type. Source: [Create daily activities in Connect Customer for an agent's shift in your contact center](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling-create-shift-activities.html), procedure step 4, Type and Sub-type fields. Checked 2026-10-02. [^3]: Work activity can be set only on Productive activities; an existing activity cannot be changed to a work activity and a saved work activity cannot be changed back. Source: [Create daily activities in Connect Customer for an agent's shift in your contact center](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling-create-shift-activities.html), procedure step 4, Work activity field and Note. Checked 2026-10-02. [^4]: Under the Custom adherence method each shift activity is mapped to specific agent statuses, and any unmapped status during that activity counts as non-adherent. Source: [Schedule Adherence for agent productivity in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/schedule-adherence.html), section 'How adherence is determined', Custom bullet. Checked 2026-10-02. [^5]: Under the Default adherence method an agent is adherent when scheduled Productive and in Available status, or scheduled Non-productive and Offline or in any custom status. Source: [Schedule Adherence for agent productivity in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/schedule-adherence.html), section 'How adherence is determined', Default bullet. Checked 2026-10-02. [^6]: Viewing historical schedule adherence metrics requires the Schedule manager View permission plus Analytics and Optimization, Historical metrics - Access. Source: [Security profile permissions for forecasting & agent scheduling in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/required-optimization-permissions.html), section 'Scheduling', Schedule manager Note. Checked 2026-10-02. [^7]: Schedule adherence metrics (Adherent time, Adherence, Scheduled time, Non-adherent time) appear on the Historical metrics and Real-time metrics pages, and real-time adherence appears in the Agent adherence widget of the Queue and agent performance dashboard. Source: [Schedule Adherence for agent productivity in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/schedule-adherence.html), section 'Where you can view schedule adherence'. Checked 2026-10-02. [^8]: If an agent's schedule is changed for a date within the last 30 days, adherence is recalculated against the new schedule. Source: [Schedule Adherence for agent productivity in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/schedule-adherence.html), section 'What happens when ....', 'A historical schedule is changed'. Checked 2026-10-02. [^9]: Amazon Connect begins generating schedule adherence automatically as soon as a published schedule starts that contains shift activities with Adherence set to yes. Source: [Schedule Adherence for agent productivity in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/schedule-adherence.html), section 'How adherence is determined', paragraph 1. Checked 2026-10-02. [^10]: Adherence thresholds can be set per activity and per staffing group up to a maximum of 10 minutes early or late, and threshold changes apply only to future adherence calculations. Source: [Schedule Adherence for agent productivity in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/schedule-adherence.html), section 'Configure adherence thresholds', Important box. Checked 2026-10-02. [^11]: An agent with no schedule for a period is shown with adherence status Unscheduled. Source: [Schedule Adherence for agent productivity in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/schedule-adherence.html), section 'How adherence is determined', paragraph after Custom bullet. Checked 2026-10-02. [^12]: By default the Admin security profile has permissions for all forecasting, capacity planning and scheduling activities. Source: [Security profile permissions for forecasting & agent scheduling in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/required-optimization-permissions.html), final paragraph. Checked 2026-10-02. [^13]: Agent event streams are not enabled by default; after creating a Kinesis data stream, enable them in the Amazon Connect console under the instance's Data streaming page by selecting the stream under Agent Events. Source: [Enable agent event streams to report agent activity in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/agent-event-streams-enable.html), introduction and procedure 'To enable agent event streams'. Checked 2026-10-02. [^14]: If server-side encryption is enabled on the agent event Kinesis stream with a key Amazon Connect cannot use, Amazon Connect cannot publish to it; the workaround is to use the same KMS key used for recordings or scheduled reports. Source: [Enable agent event streams to report agent activity in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/agent-event-streams-enable.html), Note in introduction. Checked 2026-10-02. [^15]: Agent events are written to Kinesis with the agent ARN as partition key, so all events for one agent go to the same shard. Source: [Enable agent event streams to report agent activity in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/agent-event-streams-enable.html), introduction paragraph 2. Checked 2026-10-02. [^16]: Amazon Connect agent event streams are Kinesis data streams of agent login, logout, contact connection and status-change events, which AWS describes as usable for integration into workforce management solutions. Source: [Connect Customer agent event streams](https://docs.aws.amazon.com/connect/latest/adminguide/agent-event-streams.html), introduction paragraphs 1-2. Checked 2026-10-02. [^17]: Agents need the Agent Applications permission Agent application schedule calendar - View or Edit to see their schedule, and Edit to request time off. Source: [Security profile permissions for forecasting & agent scheduling in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/required-optimization-permissions.html), section 'Agent Applications'. Checked 2026-10-02. [^18]: Agents open their schedule from the calendar icon at https://{instance name}/connect/agent-app-v2/, or at /connect/agent-app-v2/scheduling when using Salesforce CTI or a custom agent application. Source: [How agents view their schedule in the Connect Customer agent workspace](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling-view-schedule-agents.html), introduction bullet list. Checked 2026-10-02. [^19]: Capacity planning offers a Hiring plan (FTE estimates up to 64 weeks, using long-term and short-term forecasts) and a Scheduling plan that needs only a published short-term forecast. Source: [Capacity planning in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/capacity-planning.html), introduction paragraphs 2-5. Checked 2026-10-02. [^20]: Amazon Connect contact events are a near real-time stream of voice, chat, task and email contact events that AWS describes as usable for integration into workforce management solutions. Source: [Connect Customer contact events](https://docs.aws.amazon.com/connect/latest/adminguide/contact-events.html), introduction paragraphs 1-2. Checked 2026-10-02. [^21]: To create a forecast, choose Analytics and optimization, Forecasting, the Forecast tab and Create Forecast, select forecast groups and the Long-term and/or Short-term type; status moves from Scheduled to Complete when computation finishes. Source: [Create forecasts in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/create-forecasts.html), procedure 'To create a forecast', steps 3-8. Checked 2026-10-02. [^22]: Forecasts can be deleted, but downstream capacity plans and schedules created from them are impacted. Source: [Create forecasts in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/create-forecasts.html), introduction bullet list, last bullet. Checked 2026-10-02. [^23]: Creating forecast groups, importing historical data, creating, inspecting, downloading and editing forecasts each require the Forecasting - Edit permission. Source: [Security profile permissions for forecasting & agent scheduling in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/required-optimization-permissions.html), section 'Analytics and Optimization permissions', Forecasting bullet. Checked 2026-10-02. [^24]: Only the most current computed forecast is kept, so a day's forecast must be downloaded before Amazon Connect overwrites it. Source: [Forecasting in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/forecasting.html), Important box before 'Getting started with forecasting'. Checked 2026-10-02. [^25]: Adding or removing queues in an existing forecast group can trigger recomputation of its forecasts and affects downstream capacity plans and schedules built on that group. Source: [Create forecast groups in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/create-forecast-groups.html), section 'How to create forecast groups', step 7. Checked 2026-10-02. [^26]: AWS strongly recommends creating all forecast groups before creating any forecast, because historical data for queues in all forecast groups trains the forecasting model. Source: [Create forecast groups in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/create-forecast-groups.html), section 'Important things to know', bullet 4. Checked 2026-10-02. [^27]: A forecast group that was used to create forecasts cannot be removed until those forecasts are deleted from the Forecasts tab. Source: [Create forecast groups in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/create-forecast-groups.html), section 'How to create forecast groups', step 8. Checked 2026-10-02. [^28]: In Amazon Connect each queue can belong to only one forecast group, and at least one forecast group must exist before any forecast can be generated. Source: [Create forecast groups in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/create-forecast-groups.html), section 'Important things to know', bullets 2-3. Checked 2026-10-02. [^29]: Forecast groups are associated with staffing groups for scheduling, so AWS recommends grouping queues that share the same pool of agents into the same forecast group. Source: [Create forecast groups in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/create-forecast-groups.html), section 'Important things to know', bullet 1. Checked 2026-10-02. [^30]: A newly created short-term or long-term forecast is typically computed within 4 hours and often in about 1 hour, and forecast group changes trigger recomputation in the same timeframe. Source: [Forecasting in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/forecasting.html), Note following the introduction bullet list. Checked 2026-10-02. [^31]: Generate schedule produces a draft that agents and supervisors cannot see until it is published. Source: [Generate, review, and publish a schedule by using Schedule Manager in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling-publish-schedule.html), procedure step 8, Note. Checked 2026-10-02. [^32]: Schedule generation shows In progress and takes 30 minutes to 3 hours depending on agents, rules and duration, ending in Complete or Failed. Source: [Generate, review, and publish a schedule by using Schedule Manager in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling-publish-schedule.html), procedure step 9. Checked 2026-10-02. [^33]: Before publishing, schedulers can edit shift and activity times, add, replace or remove activities, copy shifts and recompute metrics, and can regenerate schedules after configuration changes. Source: [Generate, review, and publish a schedule by using Schedule Manager in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling-publish-schedule.html), sections 'Edit a schedule' and 'Regenerate a schedule'. Checked 2026-10-02. [^34]: Schedule optimization goals are chosen per channel: Service level or Average speed of answer for voice and chat, and Service level or Average time to complete for task and email. Source: [Generate, review, and publish a schedule by using Schedule Manager in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling-publish-schedule.html), procedure step 7. Checked 2026-10-02. [^35]: A schedule is generated in Schedule Manager for a single forecast group, for up to 18 weeks ahead. Source: [Generate, review, and publish a schedule by using Schedule Manager in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling-publish-schedule.html), procedure steps 5-6. Checked 2026-10-02. [^36]: Schedule generation warnings are HIGH (an agent was not scheduled), MEDIUM (an agent was scheduled but not all requirements were met) or LOW (minor issues such as overstaffing). Source: [Generate, review, and publish a schedule by using Schedule Manager in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling-publish-schedule.html), procedure step 10. Checked 2026-10-02. [^37]: The historical import CSV uses the columns QueueName, QueueId, ChannelType (VOICE, CHAT, TASK or EMAIL in capitals), TimeStamp (ISO8601), IntervalDuration, IncomingContactVolume, AverageHandleTime in seconds and ContactsHandled. Source: [Import historical data for forecasting in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/import-data-for-forecasting.html), section 'Important things to know', bullet 1 requirements list. Checked 2026-10-02. [^38]: Imported historical data can be deleted; deletion triggers an immediate change in associated forecasts and the baseline falls back to previously uploaded or native data for the affected dates. Source: [Import historical data for forecasting in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/import-data-for-forecasting.html), section 'Delete imported historical data'. Checked 2026-10-02. [^39]: A successful historical upload shows Status Complete with today's upload date; a failed upload offers a download details link with an error log identifying the faulty row. Source: [Import historical data for forecasting in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/import-data-for-forecasting.html), section 'How to import historical data', steps 6-7. Checked 2026-10-02. [^40]: Historical data can be split across multiple files, and where files overlap the last uploaded records are used. Source: [Import historical data for forecasting in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/import-data-for-forecasting.html), section 'Important things to know', multiple-files bullet. Checked 2026-10-02. [^41]: With less than one year of historical data in Amazon Connect, AWS strongly recommends importing data from the previous system to form a continuous one-year history; with more than one year, import can be skipped. Source: [Import historical data for forecasting in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/import-data-for-forecasting.html), section 'When to import data'. Checked 2026-10-02. [^42]: When historical data is imported, Amazon Connect forecasts from both its own data and the imported data, and the imported data takes precedence. Source: [Import historical data for forecasting in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/import-data-for-forecasting.html), introduction paragraph. Checked 2026-10-02. [^43]: Historical data must be uploaded separately for short-term forecasts (IntervalDuration 15mins or 30mins matching the configured interval) and long-term forecasts (IntervalDuration daily); interval-only data does not allow long-term forecasts. Source: [Import historical data for forecasting in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/import-data-for-forecasting.html), section 'Important things to know', IntervalDuration requirement and separate-upload bullet. Checked 2026-10-02. [^44]: Before publishing, a forecast can be inspected by comparing Last computed forecast with Last published forecast and comparing Actuals against the forecast, with imported history labelled Actuals Overrides. Source: [Inspect a forecast in Amazon Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/inspect-forecast.html), introduction bullet list. Checked 2026-10-02. [^45]: The forecast and schedule interval must be set before forecasts can be generated or forecast groups created. Source: [Set the forecast and schedule interval in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/set-forecast-scheduling-interval.html), section 'Important things to know', bullet 4. Checked 2026-10-02. [^46]: After the forecast and schedule interval is set in Amazon Connect it cannot be changed. Source: [Set the forecast and schedule interval in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/set-forecast-scheduling-interval.html), section 'Important things to know', last bullet. Checked 2026-10-02. [^47]: Amazon Connect supports a 15-minute or 30-minute interval for short-term forecasts and schedules, and the chosen interval is used for both. Source: [Set the forecast and schedule interval in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/set-forecast-scheduling-interval.html), section 'Important things to know', bullet 3. Checked 2026-10-02. [^48]: Setting the forecast and schedule interval requires the security profile permission Analytics, Forecast and schedule interval - Edit, and the tab is under Analytics and optimization, Forecasting. Source: [Set the forecast and schedule interval in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/set-forecast-scheduling-interval.html), section 'Important things to know' bullet 1 and procedure steps 2-3. Checked 2026-10-02. [^49]: Intraday forecasts update every 15 minutes and are generated only for queues with at least 5000 unique contacts per week per queue-channel over the last 4 weeks. Source: [Forecasting in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/forecasting.html), introduction bullet 'Intraday forecasts are updated every 15 minutes'. Checked 2026-10-02. [^50]: Amazon Connect allows at most 200 queues per forecast group, 500 forecast groups per instance and 300 staffing groups per forecast group. Source: [Connect Customer feature specifications](https://docs.aws.amazon.com/connect/latest/adminguide/feature-limits.html), table 'Forecasting, capacity planning, and scheduling feature specifications', rows Queues per forecast group, Forecast groups per instance, Staffing groups per forecast group. Checked 2026-10-02. [^51]: A single historical actuals upload may be up to 1GB, with up to 300 files each for interval and daily historical actuals. Source: [Connect Customer feature specifications](https://docs.aws.amazon.com/connect/latest/adminguide/feature-limits.html), table 'Forecasting, capacity planning, and scheduling feature specifications', historical actuals rows. Checked 2026-10-02. [^52]: Amazon Connect allows at most 350 agents per staffing group and 5,000 agents per schedule generation run, and these specifications are hard limits that cannot be increased. Source: [Connect Customer feature specifications](https://docs.aws.amazon.com/connect/latest/adminguide/feature-limits.html), table 'Forecasting, capacity planning, and scheduling feature specifications', rows 1-2; Important box at top. Checked 2026-10-02. [^53]: Amazon Connect long-term forecasts are generated for 64 weeks and updated automatically every week, and are used for capacity planning. Source: [Create forecasts in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/create-forecasts.html), introduction bullet list. Checked 2026-10-02. [^54]: Africa (Cape Town) appears in the Amazon Connect instance Region table but not in the forecasting and agent scheduling Region list, so an instance in af-south-1 should not be expected to offer native forecasting and scheduling (inferred). Source: [Availability of Connect Customer features by Region](https://docs.aws.amazon.com/connect/latest/adminguide/regions.html), sections 'Connect Customer availability by Region' and 'Forecasting & agent scheduling' compared. Checked 2026-10-02. [^55]: Publishing a forecast requires the Forecasting - Publish permission and makes the forecast visible to capacity planners and schedulers; the published status is Complete or Publish failed. Source: [Publish a forecast in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/publish-forecast.html), introduction and procedure steps 1-5. Checked 2026-10-02. [^56]: Amazon Connect retains only the last published forecast, so AWS advises downloading it with Actions, Download last published forecast before publishing a new one. Source: [Publish a forecast in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/publish-forecast.html), Important box and section 'Download the last published forecast'. Checked 2026-10-02. [^57]: After a schedule is published with Publish and Proceed, the agents and supervisors specified in its staffing groups can view it. Source: [Generate, review, and publish a schedule by using Schedule Manager in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling-publish-schedule.html), procedure step 12. Checked 2026-10-02. [^58]: Amazon Connect forecasting and agent scheduling is listed as available in US East (N. Virginia), US West (Oregon), Asia Pacific (Tokyo), Asia Pacific (Seoul), Asia Pacific (Singapore), Asia Pacific (Sydney), Canada (Central), Europe (Frankfurt), Europe (London) and AWS GovCloud (US-West). Source: [Availability of Connect Customer features by Region](https://docs.aws.amazon.com/connect/latest/adminguide/regions.html), section 'Forecasting & agent scheduling'. Checked 2026-10-02. [^59]: Amazon Connect generates schedules from a published short-term forecast, shift profiles, staffing groups and human resources and business rules. Source: [Scheduling in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling.html), introduction bullet list. Checked 2026-10-02. [^60]: AWS states that Amazon Connect scheduling is not designed to ensure compliance with any particular laws, and the customer is responsible for employment-law compliance. Source: [Scheduling in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling.html), Note under the introduction. Checked 2026-10-02. [^61]: AWS orders scheduling setup as: add users, confirm security profile permissions, create staff rules, create shift activities, create shift profiles, create staffing groups, then generate, review and publish a schedule. Source: [Scheduling in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling.html), section 'Getting started'. Checked 2026-10-02. [^62]: Scheduling, Schedule manager grants permission to view, edit and publish generated schedules, and creating staff rules, shift activities, shift profiles, staffing groups and schedules requires Schedule manager - Edit. Source: [Security profile permissions for forecasting & agent scheduling in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/required-optimization-permissions.html), section 'Scheduling', Schedule manager bullet. Checked 2026-10-02. [^63]: A user cannot be an agent in one staffing group and a supervisor in another. Source: [Create groups and rules for staffing and scheduling in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling-create-staffing-groups.html), section 'Important things to know', bullet 4. Checked 2026-10-02. [^64]: Removing an agent from a staffing group does not delete their shifts and adherence continues to be tracked, so AWS recommends removing their shifts first if they are not moving to another group. Source: [Create groups and rules for staffing and scheduling in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling-create-staffing-groups.html), section 'Important things to know', bullet 5. Checked 2026-10-02. [^65]: Every agent must be in a staffing group for Amazon Connect to generate a schedule for them, and each staffing group is associated with one forecast group. Source: [Create groups and rules for staffing and scheduling in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling-create-staffing-groups.html), section 'Important things to know' bullet 1 and 'Create group and add staff' step 4. Checked 2026-10-02. [^66]: Staffing group rules include Minimum staff required, Shift start time, Working time, Minimum rest between shifts, Consecutive working days, Maximum consecutive day of the week worked and Minimum consecutive rest period per week. Source: [Create groups and rules for staffing and scheduling in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling-create-staffing-groups.html), section 'Add rules', procedure 'To add rules'. Checked 2026-10-02. [^67]: A user who needs to view published agent schedules in the Published calendar view must be added to the staffing group as a supervisor. Source: [Create groups and rules for staffing and scheduling in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling-create-staffing-groups.html), section 'Important things to know', bullet 3. Checked 2026-10-02. [^68]: After a shift profile is saved it can be edited or removed from the Shift Profiles list view. Source: [Create a template for an agent's weekly shift in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling-create-shift-profiles.html), procedure step 10. Checked 2026-10-02. [^69]: A shift profile has its own time zone, and Amazon Connect automatically adjusts shifts in that profile for daylight saving changes. Source: [Create a template for an agent's weekly shift in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling-create-shift-profiles.html), procedure step 4. Checked 2026-10-02. [^70]: A shift profile's Schedule Window sets per day a Working or Flex status, earliest start and latest end time, optional maximum shift length and a default activity that must be a work activity. Source: [Create a template for an agent's weekly shift in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling-create-shift-profiles.html), procedure step 5, Schedule Window section. Checked 2026-10-02. [^71]: Amazon Connect short-term forecasts are generated for 18 weeks and updated automatically every day, and are used for scheduling. Source: [Create forecasts in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/create-forecasts.html), introduction bullet list. Checked 2026-10-02. [^72]: Individual staff rules (time zone, start and end dates, shift profile override, contract rules) override staffing group rules when a schedule is generated. Source: [Create staff rules for scheduling in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/scheduling-create-staff-rules.html), introduction Note and 'Create staff rules for individuals' steps 4-6. Checked 2026-10-02. [^73]: Changing the forecast time zone regenerates both short-term and long-term forecasts, with updated forecasts available within 24 hours and adjusted for daylight saving where the zone observes it. Source: [Set the forecast time zone](https://docs.aws.amazon.com/connect/latest/adminguide/set-forecast-timezone.html), section 'Important things to know', bullet 3. Checked 2026-10-02. [^74]: The default Amazon Connect forecast time zone is UTC, and editing it requires the Analytics, Forecast and schedule interval - Edit permission. Source: [Set the forecast time zone](https://docs.aws.amazon.com/connect/latest/adminguide/set-forecast-timezone.html), section 'Important things to know', bullets 1-2. Checked 2026-10-02. [^75]: AWS documentation states that Amazon Connect Customer is the current name for the product previously called Amazon Connect. Source: [What is Connect Customer?](https://docs.aws.amazon.com/connect/latest/adminguide/what-is-amazon-connect.html), Note box at top of page 'What is Connect Customer?'. Checked 2026-09-30. [^76]: AWS advises using a default security profile only if it matches what users need and otherwise creating a security profile that grants only the needed permissions, and notes new permissions are added regularly. Source: [Default security profiles in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/default-security-profiles.html), First paragraph and Note below table. Checked 2026-09-30. [^77]: A security profile is created under Users, Security profiles, Add new security profile by a user with Security profiles - Create permission, and choosing an action automatically selects the actions it depends on, such as View when Edit is chosen. Source: [Create a security profile in Connect Customer](https://docs.aws.amazon.com/connect/latest/adminguide/create-security-profile.html), How to create security profiles, items 2-7; Required permissions to create security profiles. Checked 2026-09-30. [^78]: Amazon Connect Customer is the current name for the product previously called Amazon Connect; the name Amazon Connect now refers to a set of agentic AI solutions for different business functions. Source: [What is Connect Customer?](https://docs.aws.amazon.com/connect/latest/adminguide/what-is-amazon-connect.html), What is Connect Customer? > Note at top of page. Checked 2026-09-30.