Letterbook
All docs

Letterbook Docs

Manage conversations

Use statuses, assignments, priorities, and internal notes to keep support moving.

Last updated June 25, 2026

What this guide covers#

Use the inbox to review customer context, collaborate with teammates, approve AI drafts, and move every conversation toward resolution.

A conversation is more than the customer message. It can include channel history, internal notes, customer data from integrations, AI-generated reply drafts, suggested actions, status, priority, assignment, and the final resolution.

Daily workflow#

  1. Open the newest unassigned conversation.
  2. Read the customer message and previous thread history.
  3. Review matched customer context from connected integrations.
  4. Check the AI draft and any proposed action.
  5. Add an internal note if another teammate needs context.
  6. Assign the conversation or take ownership.
  7. Edit and send the reply, or escalate if the answer needs more information.
  8. Resolve the conversation when the customer issue is handled.

Statuses and ownership#

Use status and assignment consistently so the inbox stays readable as volume grows.

  • Open: the customer needs a response or the team needs to take action.
  • Pending: the team is waiting on the customer, a teammate, or an external system.
  • Resolved: the issue is complete and no further action is expected.
  • Reopened: a customer replied after the conversation was resolved.

Every open conversation should have either an owner or a clear triage queue. Unowned tickets are the easiest way to miss urgent customers.

Internal notes#

Use internal notes for collaboration that should not be sent to the customer:

  • Summarize what has already been checked.
  • Ask a teammate to verify a policy decision.
  • Record why a refund, cancellation, or exception was approved.
  • Add engineering context before escalating a bug.

Do not put secrets, raw credentials, or unnecessary sensitive customer data in notes. If a teammate needs secure data, use the connected system of record.

Reviewing AI drafts#

When reviewing a draft, check:

  • The answer addresses the actual customer question.
  • The draft uses the right account, subscription, order, or database context.
  • The tone matches your support style.
  • Any promised action is actually staged, completed, or assigned.
  • The draft does not invent product behavior or policy exceptions.

Edit the draft before sending if any of those are wrong. Update the knowledge base if the AI repeatedly misses the same policy or product detail.

If useful drafts are hidden too often or low-quality drafts appear too often, adjust Draft Eagerness. If agents repeatedly type the same approved language, add a snippet.

Escalation habits#

Create a clear convention for the tickets the first reviewer cannot resolve. For example:

  • Billing exceptions go to the teammate who can approve credits or refunds.
  • Product bugs get a bug label and an internal note with reproduction details.
  • Account access issues require identity verification before changes are made.
  • Legal, privacy, and security issues follow a separate escalation path.

Team habits#

Document the conventions your support team should follow so the inbox stays consistent as volume grows. Include naming conventions for labels, when to assign versus leave unassigned, what counts as resolved, and which actions always need human approval.