Re-engage Quiet Accounts

When to use it: against your paying accounts, to find the ones that have gone quiet and send each a short, specific note grounded in their own history — never a blast.

              Natural language input in Claude
              "Find our quiet paying accounts and draft a re-engagement note for each."
                         │
                         ▼
┌─────────────────────────────────────────────────┐
│  1 · Define the target segment                  │   folk_record
│  A revenue floor, not churned, and quiet        │
│  longer than your own threshold.                │
└────────────────────────┬────────────────────────┘
                         ▼
┌─────────────────────────────────────────────────┐
│  2 · Pick the contact and gather context        │   folk_record
│  The most senior person on file, and            │
│  one real detail from their own notes.          │
└────────────────────────┬────────────────────────┘
                         ▼
┌─────────────────────────────────────────────────┐
│  3 · Draft the email                            │
│  Short, warm, specific — a two-word subject     │
│  and a concrete ask, never an open one.         │
└────────────────────────┬────────────────────────┘
                         ▼
╔═════════════════════════════════════════════════╗
║  4 · Review, then send                          ║   email_send
║  A human look before anything goes out —        ║
║  this is outbound to paying customers.          ║
╚════════════════════════╤════════════════════════╝
                         ▼
┌─────────────────────────────────────────────────┐
│  5 · Log the run                                │   data_write
│  Stamp today's date on every account            │
│  emailed, so nobody's touched twice.            │
└─────────────────────────────────────────────────┘

1. Define the target segment

• folk_record — filter paying accounts to your own revenue floor, not already churned, and quiet longer than your own threshold (e.g. no logged interaction in 3+ months). Combine all three filters in one query rather than fetching everything and filtering client-side — this list gets re-run regularly, keep it cheap.

2. Pick the contact and gather context

• folk_record — prefer the account's stated main point of contact; otherwise the most senior person on file with an email, someone the account has actually interacted with before over a name that's merely listed. Skip contacts from clearly unrelated categories, and skip anyone with no email rather than guessing one.
• Before drafting anything, pull what the CRM already knows: past notes, description and status fields. Search by both company and contact name, plus a few topical terms relevant to your product. Pull out exactly one concrete detail worth referencing — something they actually said or did. If the notes are empty or too generic, fall back to a plain template rather than inventing a detail — a fabricated one is worse than a generic email.

3. Draft the email

Short, warm, specific. Reference how long it's actually been since the last interaction and weave in the one real detail so it reads as "we noticed," not "we're checking a box." A two-word subject line reads as a real note, not a campaign. Offer a short call and ask for a couple of concrete times — a specific ask converts better than an open-ended "let's catch up."

4. Review, then send

• email_send — deliver the full list (contact, email, drafted note) for a human look before anything goes out; this is outbound to paying customers, it earns a review pass. Send with whoever on the team should have visibility CC'd, so a reply doesn't land in one inbox nobody else can see.

5. Log the run

• data_write — one row per account: the segment fields that qualified it, the contact used, the detail referenced, and the outcome. Stamp today's actual date (fetched at run time, not hardcoded) on every account you emailed, so the next run knows who's already been touched. Leave the date blank for anyone held back from sending.

Output

Report: how many accounts matched the segment, how many were emailed, how many were held back and why, and how many drafts used a real context detail vs. the plain fallback.