Callbacks & retries

Attempt caps and holds

An attempt cap limits how many times a contact is tried. When it is reached, or an outcome is ambiguous, the item is held.

Contents
Also in Callbacks & retries
Scheduling a callback in local timeDeferrals and manager approvalRetry cadence by location

Each location sets an attempt cap for its retry cadence. When a contact reaches the cap, DialSplice stops retrying and holds the work item for a manager. A hold is also opened when a provider outcome is ambiguous. In both cases a person reviews the item; nothing is retried blindly.

Attempt caps

The cap is part of the location configuration, for example “cap 5”. The queue row and the Follow-up & retry control card show progress, such as “Attempt 3 of 5” with five segment bars. Every touch counts, whatever the channel, and the preflight refuses the touch once the cap is reached.

Kinds of hold

  1. Attempt-cap hold. The distribution rule reads “No answer at cap · Attempt 5 of 5 · manager review”. The item leaves the agent queue and waits on the manager desk under Attempt caps reached.
  2. Reconciliation hold. A provider returned an unknown or ambiguous outcome, for example an SMS with “Reconciliation · provider outcome unknown”. The effect was journaled once; it is not resent. The item waits under Reconciliation holds.

What you’ll see

  • Agents: the item disappears from Power Hour and Callbacks. If you search for the contact, its status line shows the hold reason.
  • Managers: the Callback & retry health card shows counts for “Attempt caps reached” and “Reconciliation holds”, with a Release holds button.
  • The call log shows the sync state as “Pending” or “Reconciling” until the hold is resolved.
  • A “Reconciliation hold opened” alert can be sent by push or Teams / Slack, depending on policy.

Steps for managers

  1. Open the manager desk and click Release holds.
  2. Review each held item. For an attempt-cap hold, decide whether to close the item, reset the cadence, or move it to Coverage.
  3. For a reconciliation hold, confirm the real outcome in your provider, then record it. The ledger writes the resolution as a new event.

If something looks wrong

  • A contact was tried after the cap. Check the ledger for preflight.evaluated; the cap is enforced per channel, and a different channel may have its own count.
  • Holds keep growing. Ask your admin to check connector status. A degraded connector produces ambiguous outcomes.
  • Retry cadence by location
  • Reading queue health
  • Connector status and credentials
01Book a demo

Can't find it?

Ask in the demo, or tell us what the article should have said.

Book a DemoSee it work