customer-win-back-sequencer

name	customer-win-back-sequencer
description	For churned accounts, research what has changed since they left (new funding, team growth, competitor dissatisfaction, product updates that address their original pain), assess re-engagement potential, and generate a personalized, timing-aware win-back sequence — no generic "we miss you" emails. Composite that hands off to cold-email-outreach or linkedin-outreach.
metadata	{"version":"1.0.1","category":"outreach","type":"composite"}

Customer Win-Back Sequencer

A composite: scripts filter the churn list and research changes / contact status; the agent scores re-engagement, picks the angle, and drafts a timing-aware win-back sequence. Only pursue accounts where something material actually changed.

When to use

• "Which churned customers should we win back?" / "build a win-back campaign for [customer]."
• "Research our churned accounts for re-engagement" / "run the win-back scan."

How it works (scripts + agent)

1. Filter — filter_churn.py keeps accounts within the recency window (default 3-18 months since churn) and at/above --min-mrr. Too recent = too soon; too old = stale.
2. Research company changes — detect_signal.py --signal winback per account (funding, growth/hiring, launches). (agent) summarize what changed.
3. Research contact + competitor changes — find_contacts.py against the domain to check if the contact is still there / promoted / replaced and re-find the email (likely stale post-churn). For competitor-dissatisfaction, detect_signal.py --signal general --extra "{competitor} alternative OR complaints". (agent) read the evidence.
4. (agent) Score + pick the angle — change-triggered / fixed-pain / fresh-DM. Don't re-pitch the value that caused churn; anchor on the fixed reason or a new change. If the contact left, find the new decision-maker.
5. (agent) Draft the win-back sequence with email-drafting (change as why-now) + timing.
6. Hand off to cold-email-outreach / linkedin-outreach; slack approval.

How to run

# 1 — filter the churn list
python3 ${SKILL_DIR}/scripts/filter_churn.py \
  --input ${WORKSPACE}/churn.csv --min-months 3 --max-months 18 --min-mrr 500 \
  --output ${WORKSPACE}/pursue.json

# 2 — research what changed per account
python3 ${SKILL_DIR}/scripts/detect_signal.py \
  --companies ${WORKSPACE}/pursue.json --signal winback --extract \
  --output ${WORKSPACE}/changes.json

# 3 — re-verify / re-find the contact (email likely stale)
python3 ${SKILL_DIR}/scripts/find_contacts.py \
  --domains ${WORKSPACE}/pursue.json --titles "their role,VP,Director" \
  --enrich --verify --output ${WORKSPACE}/contacts.json

Outputs

• pursue.json (recency/MRR-filtered), changes.json, contacts.json, the agent's ranked re-engagement list (per-account findings + angle + score) and win-back sequences with timing — handed to the launch skill + review table.

Credentials / env

• env.required: none. Change research is keyless, drafting is the agent, and the launch step hands off to cold-email-outreach (CSV export with no key).
• env.optional (all degrade): APOLLO_API_KEY — if set → org/contact enrichment; else → serp research + pattern-guess. MILLIONVERIFIER_API_KEY — if set → verify (re-verify the stale post-churn email); else → local syntax/dedup (bounce risk). Send/launch — if a send key (LEMLIST_API_KEY / INSTANTLY_API_KEY / SENDGRID_API_KEY / RESEND_API_KEY, or PHANTOMBUSTER_API_KEY for LinkedIn) is set → launch the sequence; else → export a CSV to send manually (the keyless default).

Notes & edge cases

• Timing + relevance are everything — only pursue accounts where something material changed.
• Don't re-pitch the value that caused churn — anchor on the fixed reason or a new change. If the contact left, find the new decision-maker (fresh eyes, no baggage).
• Re-verify the contact email (likely stale post-churn) before send.