> ## Documentation Index
> Fetch the complete documentation index at: https://helpdocs.getthread.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Stale Ticket Follow-Up Cadence

> Drive a configurable follow-up cadence (for example 24/48/72 hours, three attempts) on tickets waiting for a client reply — draft each friendly nudge, count documented attempts, and skip tickets in legitimate waits.

<Info>
  **Category:** QA & Closure · [View source ↗](https://github.com/bryan-getthread/skills/blob/main/skills/qa-and-closure/stale-ticket-followup-cadence/SKILL.md)
</Info>

**Connectors:** none — works with Thread out of the box

**Role:** [Service & Ops Manager](/start-here/roles/service-ops-manager)

**Outcome:** Fewer Escalations & Less Noise

**When to use:** "Follow up on my waiting tickets" / "who hasn't replied to us?" — running the desk's standard cadence (24/48/72h, or the house variant) across a queue, or drafting the next nudge for one specific quiet ticket.

**Run it:** across all waiting-on-client tickets (manually or on a schedule).

## Prompt

```
Keep waiting-on-client tickets moving on a predictable rhythm. Count the attempts already
documented in the thread, draft the next follow-up in a friendly client-facing voice, and know
when a ticket is waiting for a good reason and should be left alone.

1. Confirm the cadence. Default: first follow-up 24h after the last outbound client-facing
   message, second at 48, third at 72 — three attempts total, then hand off to the No-Response
   Closure Sequence skill. Honor a desk-specific cadence if the user states one ("3 days / 3
   attempts / 3 emails" is a common house rule).

2. Find candidates: open tickets in waiting-on-client statuses where the last message is outbound
   and older than the current cadence step.

3. For each candidate, count documented attempts: client-visible follow-up messages since the
   client last replied. Only messages the client could actually see count — internal notes are
   not attempts, and never assume a phone attempt happened unless a note records it.

4. Skip legitimate waits: a future scheduled appointment, a client-stated timeline ("back from
   vacation Monday"), a vendor/third-party dependency, or an approval pending. Don't nudge someone
   who told you when they'll respond — the skip-list is sacred; when wait legitimacy is unclear,
   skip and flag rather than send.

5. Draft the follow-up for each remaining ticket in the client-facing voice: friendly, short,
   references the original issue in one line, states what's needed from them, and offers an easy
   out ("if this is resolved, let us know and we'll close it"). Escalate tone gently across
   attempts — attempt three notes the ticket will be closed soon without a reply, and never
   sounds annoyed. No internal jargon, no ticket-system boilerplate, no invented details. Match
   the client's language where the thread isn't in English.

6. Present drafts for review, then post each approved follow-up as a client-facing note and set
   the appropriate waiting status. Never bulk-send without the user reviewing the drafts.

7. Output a summary table: ticket, client, attempt number just sent (or skipped and why), and the
   date the next cadence step falls due. Never exceed the attempt count — if the thread already
   shows three documented attempts, route to the closure sequence instead of a fourth nudge.

This is a cadence sweep, not a Flow — Thread Flows fire on ticket events only, with no
schedule/cron/age/elapsed-time trigger. Run it manually on demand, or from an external scheduler
that invokes Super Magic. Running unattended: your entire reply is the client-facing follow-up
message, posted verbatim — no narration, no attempt counters, no internal commentary. Send only
when (a) the ticket is in a waiting-on-client status, (b) the last message is outbound, (c) the
cadence step is due, and (d) documented attempts are below the maximum. If any condition fails or
the wait looks legitimate, produce no output and stop. Never draft attempt three or beyond
unattended if the desk requires human sign-off before final notices.
```


## Related topics

- [Using Variables in Magic AI and Magic Agents](/assistive-ai/variables.md)
- [Aging, SLA & Follow-Up](/skill-library/qa-and-closure/aging-and-sla-cleanup.md)
- [Follow-up Chaser](/skill-library/communication/followup-chaser.md)