# 1. Risk classification **HIGH risk.** This touches almost every exposure in the definition at once: - **Money:** Automated billing controls cash collection for all 400 customers. A misconfigured proration, tax rule, or charge amount bills real money incorrectly across the entire base at once, not one invoice at a time. - **Trust:** B2B customers tolerate a manual invoice error quietly; a wrong automated charge to their card or account hits their finance team and erodes confidence fast. At 400 customers, a bad batch is a churn event. - **Compliance:** Subscription billing brings tax handling (VAT/sales tax by jurisdiction), payment-data handling (PCI scope via the processor), and dunning/refund obligations. Manual invoicing largely shielded you from this. - **Reversibility:** Low. Once you've migrated contracts, payment tokens, and billing cycles into the new system and cut over, rolling back to manual mid-cycle is slow and messy—you'd be reconciling two systems against live customer money. The objective (cut errors, save 10 hrs/week) is operationally minor; the *exposure* is what sets the rating, not the size of the win. # 2. Scrutinize before proceeding - **Contract/price mapping** — every customer's plan, discount, custom terms, and renewal date ported correctly. *Validate: Finance lead + Sales/account owner for non-standard deals.* - **Tax configuration** — rates and rules per jurisdiction. *Validate: external accountant or tax advisor.* - **Payment processor + PCI scope** — tokenization, who holds card data, processor contract. *Validate: Finance lead + security-responsible engineer.* - **Proration, mid-cycle changes, refunds, failed-payment (dunning) logic.** *Validate: Finance lead.* - **Revenue reconciliation** — new system's output must tie to your books. *Validate: Finance lead / bookkeeper.* - **Customer-facing comms** about the billing change. *Validate: Customer Success lead.* # 3. Ship fast and verify - Invoice/email **templates and branding** — adjust live, low stakes. - **Internal dashboards and reporting views** in the new tool — iterate as you go. - **Admin notifications and internal alerts** for failed payments. - **The 10-hours-saved workflow itself** (who stops doing what) — refine after cutover; no sign-off needed. # 4. Small-scale test - **Group:** 15–25 customers — mix of standard plans, one custom-discount deal, and at least one in a different tax jurisdiction. Bias toward your more forgiving / engaged accounts. - **Trial period:** Two full billing cycles (≈2 months) so renewals and a real charge run, not just setup. - **Run in parallel:** Keep manual invoicing live for the other ~375 customers during the test—two systems, no full cutover. - **Rollback path:** Because the rest are still manual, reverting the pilot group means switching them back to manual invoicing next cycle and refunding/correcting any mischarge. Define the refund-and-apology script *before* you start. # 5. Three go/no-go questions 1. Did every pilot customer get charged the **exact correct amount** (price, discount, tax, proration) across two cycles, with zero corrections needed? 2. Does the new system's revenue output **reconcile cleanly** to our books, and can Finance close the month from it? 3. Do we have a **tested rollback and refund process**, so if a bad batch goes out to all 400 we can stop it, reverse it, and notify customers within 24 hours?