Execution Engine

The execution engine automates your collection strategy by scheduling and dispatching contact actions without manual intervention. It processes eligible accounts based on their assigned strategy, respecting frequency limits and time windows.

How It Works

1. Eligible accounts are identified — accounts with an active strategy, not paused, and in a contactable status (pending_contact, in_auto_management, promise_overdue).
2. The engine determines the next action from the strategy sequence and the account’s current step index. For voice/WhatsApp steps, it may retry the same step (if configured) after no answer, a dropped call, or a step timeout before advancing.
3. Actions are placed in the execution queue with a scheduled time.
4. The queue processor picks up pending items, validates dispatch rules (opt-out, frequency, allowed hours), and dispatches through the appropriate channel.

The Execution Page

The Execution page provides real-time visibility into the engine:

KPIs

• Pending — Actions waiting to be processed.
• Processing — Actions currently being dispatched.
• Completed Today — Successfully dispatched actions today.
• Failed Today — Actions that failed (provider error, validation failure).
• Skipped Today — Actions skipped (opt-out, frequency limit, outside hours).

Queue Table

The table shows recent queue items with:

• Status — Current state of the queue item.
• Scheduled For — When the action was scheduled to run.
• Attempts — Which step in the strategy sequence this represents.
• Error — Error message if the action failed.

Controls

Pause / Resume

Toggle the engine on or off for your organization. When paused, no new actions are dispatched but existing queue items are preserved.

Manual Enqueue

Click Enqueue Campaign to manually trigger the engine to evaluate eligible accounts and add them to the queue. Use the campaign dropdown next to the button to enqueue only a specific campaign's accounts, or select "All eligible accounts" to process everything.

This is useful when:

• You've just imported new accounts.
• You've modified a strategy and want changes to take effect immediately.
• You want to force a re-evaluation outside the normal schedule.
• You want to enqueue a specific campaign without affecting others.

Queue Statuses

Status	Meaning
Pending	Waiting to be processed at the scheduled time
Processing	Currently being dispatched
Completed	Successfully dispatched and interaction recorded
Failed	Dispatch failed — will retry if attempts remain
Skipped	Skipped due to validation (opt-out, frequency, hours, etc)

Callback Handling

When a debtor requests a callback during a voice call, the current strategy step is paused (status: callback pending) until the scheduled callback time. At that point, the engine automatically creates a new queue item for the return call.

If a callback is not fulfilled within 24 hours of the requested time, the engine automatically advances to the next strategy step.

Delivery failures

When an automated message cannot be delivered (for example, email bounce, or SMS/WhatsApp failed / undelivered), the engine treats the step as completed with a delivery failure, clears conflicting pending queue rows for that account, and advances to the next step in the strategy. The same applies to voice when the provider reports a call initiation failure before a conversation starts.

Manual sends from the dashboard are not tied to the strategy engine in this way.

Adaptive timing after voice calls

After every AI voice call, the engine analyzes the debtor's sentiment and willingness to pay (scored 1–5 by the AI) to automatically adjust the timing and tone of the next step:

• Positive signals (high willingness, positive/neutral sentiment) — the next step is scheduled sooner to maintain momentum.
• Negative signals (low willingness, hostile sentiment, emotional crisis) — the next step is delayed to allow a cooling period and the tone softens to empathetic.
• Neutral signals — the standard delay from the strategy is applied unchanged.

These adjustments are recorded in the strategy execution log for full auditability. You can view the AI analysis signals (sentiment, willingness, timing adjustment) in the Strategy Execution card on each account's detail page.

Promise follow-up after overdue

If a promise is broken and the account moves to promise overdue, the next time accounts are enqueued (scheduler or Enqueue on the Execution page), accounts that had already finished the sequence are reset to the beginning of the strategy and return to in auto management, so the playbook can run again from step one.

TIP

The engine runs automatically on a configurable interval. You don't need to manually enqueue unless you want immediate processing of newly added accounts.